Multi-view video based multiple objects segmentation using graph cut and spatiotemporal projections...

34
Multi-view video based multiple objects segmentation using graph cut and spatiotemporal projections Journal of Visual Communication and Image Representation Volume 21, Issues 5–6, July–August 2010, Pages 453–461 Qian Zhang, King Ngi Ngan Department of Electronic Engineering, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong Speaker : Yi-Ting Chen

Transcript of Multi-view video based multiple objects segmentation using graph cut and spatiotemporal projections...

Multi-view video based multiple objects segmentation using graph cut and spatiotemporal projections

Journal of Visual Communication and Image Representation

Volume 21 Issues 5ndash6 JulyndashAugust 2010 Pages 453ndash461

Qian Zhang King Ngi Ngan

Department of Electronic Engineering The Chinese University of Hong Kong Shatin New Territories Hong Kong

Speaker Yi-Ting Chen

2 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

3 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

4 Introduction

Most of the interest has been focused on the research of single view segmentation

Depth information in the 3D scene can be reconstructed from multi-view images but multiple view segmentation has not attracted much attention

Most of the classical and start-of-the-art graph cut based segmentation algorithms require userrsquos interventions to specify the initial foreground and background regions as hard constraints

5 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

6 Overview of the proposed framework We built a five-view camera system for views v 01234isin

To reduce the projection error and avoid extensive computational load we select view 2 as the key view to start the segmentation process

7 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

8 Automatic initial interested objects (IIOs) extraction based on saliency model (SM)

Inspired by the work in [33] more sophisticated cues such as motion and depth are combined into our topographical SM

(a) input image (b) saliency map using depth and motion and (c) extracted IIOs

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

9 Multiple objects segmentation using graph cut

For individual object we construct a sub-graph for the pixels belonging to its ldquoObject Rectanglerdquo an enlarged rectangle to encompass the whole object

and restricts the segmentation region

we convert multiple objects segmentation into several sub-segmentation problems

10 Objects segmentation by using graph cut

Graph cut

The general formulation of energy function

Data term smoothness term

11 Basic energy function

Data term evaluate the likelihood of a certain pixel p

assigned to the label fp

color (RGB) and depth information are combined

is the color distributions modeled by the Gaussian Mixture Model (GMM) is the depth modeled by the histogram modeg() denotes a Gaussian probability distributionh() is the histogram modew() is the mixture weighting coefficient is GMM component variable =d r g b is a four-dimensional feature vector for pixel p

12Basic energy function

Smoothness term Epq(fpfq) measures the penalty of two neighboring

pixels p and q with different labels

dist(pq) is the coordinate distance between p and q

diff(cpcq) is the average RGB color difference between p and q

βr=(2〈 (rp-rq)2〉 )-1 where 〈 middot〉 is the expectation operator for the red channel

13

basic energy function using (a) color (b) depth (c) combined color and depth

The result with basic energy function

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

2 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

3 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

4 Introduction

Most of the interest has been focused on the research of single view segmentation

Depth information in the 3D scene can be reconstructed from multi-view images but multiple view segmentation has not attracted much attention

Most of the classical and start-of-the-art graph cut based segmentation algorithms require userrsquos interventions to specify the initial foreground and background regions as hard constraints

5 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

6 Overview of the proposed framework We built a five-view camera system for views v 01234isin

To reduce the projection error and avoid extensive computational load we select view 2 as the key view to start the segmentation process

7 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

8 Automatic initial interested objects (IIOs) extraction based on saliency model (SM)

Inspired by the work in [33] more sophisticated cues such as motion and depth are combined into our topographical SM

(a) input image (b) saliency map using depth and motion and (c) extracted IIOs

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

9 Multiple objects segmentation using graph cut

For individual object we construct a sub-graph for the pixels belonging to its ldquoObject Rectanglerdquo an enlarged rectangle to encompass the whole object

and restricts the segmentation region

we convert multiple objects segmentation into several sub-segmentation problems

10 Objects segmentation by using graph cut

Graph cut

The general formulation of energy function

Data term smoothness term

11 Basic energy function

Data term evaluate the likelihood of a certain pixel p

assigned to the label fp

color (RGB) and depth information are combined

is the color distributions modeled by the Gaussian Mixture Model (GMM) is the depth modeled by the histogram modeg() denotes a Gaussian probability distributionh() is the histogram modew() is the mixture weighting coefficient is GMM component variable =d r g b is a four-dimensional feature vector for pixel p

12Basic energy function

Smoothness term Epq(fpfq) measures the penalty of two neighboring

pixels p and q with different labels

dist(pq) is the coordinate distance between p and q

diff(cpcq) is the average RGB color difference between p and q

βr=(2〈 (rp-rq)2〉 )-1 where 〈 middot〉 is the expectation operator for the red channel

13

basic energy function using (a) color (b) depth (c) combined color and depth

The result with basic energy function

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

3 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

4 Introduction

Most of the interest has been focused on the research of single view segmentation

Depth information in the 3D scene can be reconstructed from multi-view images but multiple view segmentation has not attracted much attention

Most of the classical and start-of-the-art graph cut based segmentation algorithms require userrsquos interventions to specify the initial foreground and background regions as hard constraints

5 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

6 Overview of the proposed framework We built a five-view camera system for views v 01234isin

To reduce the projection error and avoid extensive computational load we select view 2 as the key view to start the segmentation process

7 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

8 Automatic initial interested objects (IIOs) extraction based on saliency model (SM)

Inspired by the work in [33] more sophisticated cues such as motion and depth are combined into our topographical SM

(a) input image (b) saliency map using depth and motion and (c) extracted IIOs

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

9 Multiple objects segmentation using graph cut

For individual object we construct a sub-graph for the pixels belonging to its ldquoObject Rectanglerdquo an enlarged rectangle to encompass the whole object

and restricts the segmentation region

we convert multiple objects segmentation into several sub-segmentation problems

10 Objects segmentation by using graph cut

Graph cut

The general formulation of energy function

Data term smoothness term

11 Basic energy function

Data term evaluate the likelihood of a certain pixel p

assigned to the label fp

color (RGB) and depth information are combined

is the color distributions modeled by the Gaussian Mixture Model (GMM) is the depth modeled by the histogram modeg() denotes a Gaussian probability distributionh() is the histogram modew() is the mixture weighting coefficient is GMM component variable =d r g b is a four-dimensional feature vector for pixel p

12Basic energy function

Smoothness term Epq(fpfq) measures the penalty of two neighboring

pixels p and q with different labels

dist(pq) is the coordinate distance between p and q

diff(cpcq) is the average RGB color difference between p and q

βr=(2〈 (rp-rq)2〉 )-1 where 〈 middot〉 is the expectation operator for the red channel

13

basic energy function using (a) color (b) depth (c) combined color and depth

The result with basic energy function

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

4 Introduction

Most of the interest has been focused on the research of single view segmentation

Depth information in the 3D scene can be reconstructed from multi-view images but multiple view segmentation has not attracted much attention

Most of the classical and start-of-the-art graph cut based segmentation algorithms require userrsquos interventions to specify the initial foreground and background regions as hard constraints

5 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

6 Overview of the proposed framework We built a five-view camera system for views v 01234isin

To reduce the projection error and avoid extensive computational load we select view 2 as the key view to start the segmentation process

7 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

8 Automatic initial interested objects (IIOs) extraction based on saliency model (SM)

Inspired by the work in [33] more sophisticated cues such as motion and depth are combined into our topographical SM

(a) input image (b) saliency map using depth and motion and (c) extracted IIOs

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

9 Multiple objects segmentation using graph cut

For individual object we construct a sub-graph for the pixels belonging to its ldquoObject Rectanglerdquo an enlarged rectangle to encompass the whole object

and restricts the segmentation region

we convert multiple objects segmentation into several sub-segmentation problems

10 Objects segmentation by using graph cut

Graph cut

The general formulation of energy function

Data term smoothness term

11 Basic energy function

Data term evaluate the likelihood of a certain pixel p

assigned to the label fp

color (RGB) and depth information are combined

is the color distributions modeled by the Gaussian Mixture Model (GMM) is the depth modeled by the histogram modeg() denotes a Gaussian probability distributionh() is the histogram modew() is the mixture weighting coefficient is GMM component variable =d r g b is a four-dimensional feature vector for pixel p

12Basic energy function

Smoothness term Epq(fpfq) measures the penalty of two neighboring

pixels p and q with different labels

dist(pq) is the coordinate distance between p and q

diff(cpcq) is the average RGB color difference between p and q

βr=(2〈 (rp-rq)2〉 )-1 where 〈 middot〉 is the expectation operator for the red channel

13

basic energy function using (a) color (b) depth (c) combined color and depth

The result with basic energy function

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

5 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

6 Overview of the proposed framework We built a five-view camera system for views v 01234isin

To reduce the projection error and avoid extensive computational load we select view 2 as the key view to start the segmentation process

7 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

8 Automatic initial interested objects (IIOs) extraction based on saliency model (SM)

Inspired by the work in [33] more sophisticated cues such as motion and depth are combined into our topographical SM

(a) input image (b) saliency map using depth and motion and (c) extracted IIOs

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

9 Multiple objects segmentation using graph cut

For individual object we construct a sub-graph for the pixels belonging to its ldquoObject Rectanglerdquo an enlarged rectangle to encompass the whole object

and restricts the segmentation region

we convert multiple objects segmentation into several sub-segmentation problems

10 Objects segmentation by using graph cut

Graph cut

The general formulation of energy function

Data term smoothness term

11 Basic energy function

Data term evaluate the likelihood of a certain pixel p

assigned to the label fp

color (RGB) and depth information are combined

is the color distributions modeled by the Gaussian Mixture Model (GMM) is the depth modeled by the histogram modeg() denotes a Gaussian probability distributionh() is the histogram modew() is the mixture weighting coefficient is GMM component variable =d r g b is a four-dimensional feature vector for pixel p

12Basic energy function

Smoothness term Epq(fpfq) measures the penalty of two neighboring

pixels p and q with different labels

dist(pq) is the coordinate distance between p and q

diff(cpcq) is the average RGB color difference between p and q

βr=(2〈 (rp-rq)2〉 )-1 where 〈 middot〉 is the expectation operator for the red channel

13

basic energy function using (a) color (b) depth (c) combined color and depth

The result with basic energy function

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

6 Overview of the proposed framework We built a five-view camera system for views v 01234isin

To reduce the projection error and avoid extensive computational load we select view 2 as the key view to start the segmentation process

7 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

8 Automatic initial interested objects (IIOs) extraction based on saliency model (SM)

Inspired by the work in [33] more sophisticated cues such as motion and depth are combined into our topographical SM

(a) input image (b) saliency map using depth and motion and (c) extracted IIOs

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

9 Multiple objects segmentation using graph cut

For individual object we construct a sub-graph for the pixels belonging to its ldquoObject Rectanglerdquo an enlarged rectangle to encompass the whole object

and restricts the segmentation region

we convert multiple objects segmentation into several sub-segmentation problems

10 Objects segmentation by using graph cut

Graph cut

The general formulation of energy function

Data term smoothness term

11 Basic energy function

Data term evaluate the likelihood of a certain pixel p

assigned to the label fp

color (RGB) and depth information are combined

is the color distributions modeled by the Gaussian Mixture Model (GMM) is the depth modeled by the histogram modeg() denotes a Gaussian probability distributionh() is the histogram modew() is the mixture weighting coefficient is GMM component variable =d r g b is a four-dimensional feature vector for pixel p

12Basic energy function

Smoothness term Epq(fpfq) measures the penalty of two neighboring

pixels p and q with different labels

dist(pq) is the coordinate distance between p and q

diff(cpcq) is the average RGB color difference between p and q

βr=(2〈 (rp-rq)2〉 )-1 where 〈 middot〉 is the expectation operator for the red channel

13

basic energy function using (a) color (b) depth (c) combined color and depth

The result with basic energy function

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

7 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

8 Automatic initial interested objects (IIOs) extraction based on saliency model (SM)

Inspired by the work in [33] more sophisticated cues such as motion and depth are combined into our topographical SM

(a) input image (b) saliency map using depth and motion and (c) extracted IIOs

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

9 Multiple objects segmentation using graph cut

For individual object we construct a sub-graph for the pixels belonging to its ldquoObject Rectanglerdquo an enlarged rectangle to encompass the whole object

and restricts the segmentation region

we convert multiple objects segmentation into several sub-segmentation problems

10 Objects segmentation by using graph cut

Graph cut

The general formulation of energy function

Data term smoothness term

11 Basic energy function

Data term evaluate the likelihood of a certain pixel p

assigned to the label fp

color (RGB) and depth information are combined

is the color distributions modeled by the Gaussian Mixture Model (GMM) is the depth modeled by the histogram modeg() denotes a Gaussian probability distributionh() is the histogram modew() is the mixture weighting coefficient is GMM component variable =d r g b is a four-dimensional feature vector for pixel p

12Basic energy function

Smoothness term Epq(fpfq) measures the penalty of two neighboring

pixels p and q with different labels

dist(pq) is the coordinate distance between p and q

diff(cpcq) is the average RGB color difference between p and q

βr=(2〈 (rp-rq)2〉 )-1 where 〈 middot〉 is the expectation operator for the red channel

13

basic energy function using (a) color (b) depth (c) combined color and depth

The result with basic energy function

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

8 Automatic initial interested objects (IIOs) extraction based on saliency model (SM)

Inspired by the work in [33] more sophisticated cues such as motion and depth are combined into our topographical SM

(a) input image (b) saliency map using depth and motion and (c) extracted IIOs

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

9 Multiple objects segmentation using graph cut

For individual object we construct a sub-graph for the pixels belonging to its ldquoObject Rectanglerdquo an enlarged rectangle to encompass the whole object

and restricts the segmentation region

we convert multiple objects segmentation into several sub-segmentation problems

10 Objects segmentation by using graph cut

Graph cut

The general formulation of energy function

Data term smoothness term

11 Basic energy function

Data term evaluate the likelihood of a certain pixel p

assigned to the label fp

color (RGB) and depth information are combined

is the color distributions modeled by the Gaussian Mixture Model (GMM) is the depth modeled by the histogram modeg() denotes a Gaussian probability distributionh() is the histogram modew() is the mixture weighting coefficient is GMM component variable =d r g b is a four-dimensional feature vector for pixel p

12Basic energy function

Smoothness term Epq(fpfq) measures the penalty of two neighboring

pixels p and q with different labels

dist(pq) is the coordinate distance between p and q

diff(cpcq) is the average RGB color difference between p and q

βr=(2〈 (rp-rq)2〉 )-1 where 〈 middot〉 is the expectation operator for the red channel

13

basic energy function using (a) color (b) depth (c) combined color and depth

The result with basic energy function

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

9 Multiple objects segmentation using graph cut

For individual object we construct a sub-graph for the pixels belonging to its ldquoObject Rectanglerdquo an enlarged rectangle to encompass the whole object

and restricts the segmentation region

we convert multiple objects segmentation into several sub-segmentation problems

10 Objects segmentation by using graph cut

Graph cut

The general formulation of energy function

Data term smoothness term

11 Basic energy function

Data term evaluate the likelihood of a certain pixel p

assigned to the label fp

color (RGB) and depth information are combined

is the color distributions modeled by the Gaussian Mixture Model (GMM) is the depth modeled by the histogram modeg() denotes a Gaussian probability distributionh() is the histogram modew() is the mixture weighting coefficient is GMM component variable =d r g b is a four-dimensional feature vector for pixel p

12Basic energy function

Smoothness term Epq(fpfq) measures the penalty of two neighboring

pixels p and q with different labels

dist(pq) is the coordinate distance between p and q

diff(cpcq) is the average RGB color difference between p and q

βr=(2〈 (rp-rq)2〉 )-1 where 〈 middot〉 is the expectation operator for the red channel

13

basic energy function using (a) color (b) depth (c) combined color and depth

The result with basic energy function

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

10 Objects segmentation by using graph cut

Graph cut

The general formulation of energy function

Data term smoothness term

11 Basic energy function

Data term evaluate the likelihood of a certain pixel p

assigned to the label fp

color (RGB) and depth information are combined

is the color distributions modeled by the Gaussian Mixture Model (GMM) is the depth modeled by the histogram modeg() denotes a Gaussian probability distributionh() is the histogram modew() is the mixture weighting coefficient is GMM component variable =d r g b is a four-dimensional feature vector for pixel p

12Basic energy function

Smoothness term Epq(fpfq) measures the penalty of two neighboring

pixels p and q with different labels

dist(pq) is the coordinate distance between p and q

diff(cpcq) is the average RGB color difference between p and q

βr=(2〈 (rp-rq)2〉 )-1 where 〈 middot〉 is the expectation operator for the red channel

13

basic energy function using (a) color (b) depth (c) combined color and depth

The result with basic energy function

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

11 Basic energy function

Data term evaluate the likelihood of a certain pixel p

assigned to the label fp

color (RGB) and depth information are combined

is the color distributions modeled by the Gaussian Mixture Model (GMM) is the depth modeled by the histogram modeg() denotes a Gaussian probability distributionh() is the histogram modew() is the mixture weighting coefficient is GMM component variable =d r g b is a four-dimensional feature vector for pixel p

12Basic energy function

Smoothness term Epq(fpfq) measures the penalty of two neighboring

pixels p and q with different labels

dist(pq) is the coordinate distance between p and q

diff(cpcq) is the average RGB color difference between p and q

βr=(2〈 (rp-rq)2〉 )-1 where 〈 middot〉 is the expectation operator for the red channel

13

basic energy function using (a) color (b) depth (c) combined color and depth

The result with basic energy function

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

12Basic energy function

Smoothness term Epq(fpfq) measures the penalty of two neighboring

pixels p and q with different labels

dist(pq) is the coordinate distance between p and q

diff(cpcq) is the average RGB color difference between p and q

βr=(2〈 (rp-rq)2〉 )-1 where 〈 middot〉 is the expectation operator for the red channel

13

basic energy function using (a) color (b) depth (c) combined color and depth

The result with basic energy function

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

13

basic energy function using (a) color (b) depth (c) combined color and depth

The result with basic energy function

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

14 The segmentation errors in the rectangles

errors occur because their color and depth information are very similar to the foreground data

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

15 Background penalty with occlusion reasoning(12)

Since we capture the same scene at different view points occluded background regions often occur around the object boundary the occluded regions have a higher probability to be the

background than the visible ones

impose a background penalty factor αbp=35

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

16 Background penalty with occlusion reasoning(22)

(a) background probability map without occlusion penalty (b) combined occlusion map (c) background probability map with occlusion penalty

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

17 The erroneous segmentations marked as ellipses

errors are mainly caused by the strong color contrast in the background comparing to the weak contrast across the ldquotruerdquo object boundary

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

18 Foreground contrast enhancement(13)

To make the color contrast representation more efficient

the average color difference is computed in the perceptually uniform L a b color space

To enhance the contrast across foregroundbackground boundary and attenuate the background contrast

adopting the motion residual information

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

19 Foreground contrast enhancement(23)

The motion residual is defined as

the smoothness term combines the L a b color and motion residual contrasts

are the motion residual of p and q

is the reconstructed image from and the motion field

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

20 Foreground contrast enhancement(33)

But combining the color contrast and motion residual contrasts will not only attenuate the background contrast but also weaken the lsquolsquotruerdquo foreground contrast

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

21 Foreground contrast enhancement(33)

we define a local color contrast to enhance the discontinuity distribution in its neighborhood calculate the local mean μ and the local variance δ of

contrast

=[26] J Wang P Bhat RA Colburn M Agrawala MF Cohen Interactive videocutout ACM Transactions on Graphics 24 (2005) 585ndash594

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

22 The result with modified energy function

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

23 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

24

projected by pixel-based disparity compensation exploits the spatial consistency

projected by pixel-based motion compensation from the mask of its previous frame enforces the temporal consistency

Multi-view video segmentation

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

25 Uncertain boundary band validation

To improve the segmentation results we construct an uncertain band along the object boundary based on an activity measure

using our graph cut algorithm to yield more accurate segmentation layers

(a) prediction mask of view 3(b) uncertain band based on the post-processing

(a) (b)

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

26 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

27 Experimental results

five-view camera system

resolution of 640 480 at frame rate of 30 frames per second (fps)

we demonstrated on two types of multi-view videos simulating different scenarios similar and low depth

different depths

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

28 Segmentation of IOs with similar and low depth

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

29 Segmentation of IOs with different depths

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

30 Comparison with othersrsquo methods(12)

compare with Kolmogorovrsquos bilayer segmentation algorithm [21] by using their test images

(a) left view (b) right view (c) result by our proposed algorithm(d) result by Kolmogorovrsquos algorithm

[21] V Kolmogorov A Criminisi A Blake G Cross C Rother Probabilistic fusion of stereo with color and contrast for bi-layer segmentation IEEE Transactions on Pattern Analysis and Machine Intelligence 28 (9) (2006) 1480ndash1492

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

31

compare our proposed algorithm with an existing method employing multi-way cut with α-expansion [33] using our test images

Comparison with othersrsquo methods(22)

[33] WX Yang KN Ngan Unsupervised multiple object segmentation of multiview images Advanced Concepts for Intelligent Vision Systems Conference (2007) pp 178ndash189

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

32 Outline

Introduction

The proposed framework

Method Segmentation for key view

Multi-view video segmentation

Experimental results

Conclusion

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

33 Conclusion

In this paper we propose an automatic segmentation algorithm for multiple objects from multi-view video

The experiment was implemented on two representative multi-view videos

Accurate segmentation results with good visual quality and subjective comparison with othersrsquo methods attest to the efficiency and robustness of our proposed algorithm

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening

34

Thanks for your listening

  • Multi-view video based multiple objects segmentation using grap
  • Outline
  • Outline (2)
  • Introduction
  • Outline (3)
  • Overview of the proposed framework
  • Outline (4)
  • Automatic initial interested objects (IIOs) extraction based on
  • Multiple objects segmentation using graph cut
  • Objects segmentation by using graph cut
  • Basic energy function
  • Basic energy function (2)
  • The result with basic energy function
  • The segmentation errors in the rectangles
  • Background penalty with occlusion reasoning(12)
  • Background penalty with occlusion reasoning(22)
  • The erroneous segmentations marked as ellipses
  • Foreground contrast enhancement(13)
  • Foreground contrast enhancement(23)
  • Foreground contrast enhancement(33)
  • Foreground contrast enhancement(33) (2)
  • The result with modified energy function
  • Outline (5)
  • Multi-view video segmentation
  • Uncertain boundary band validation
  • Outline (6)
  • Experimental results
  • Segmentation of IOs with similar and low depth
  • Segmentation of IOs with different depths
  • Comparison with othersrsquo methods(12)
  • Comparison with othersrsquo methods(22)
  • Outline (7)
  • Conclusion
  • Thanks for your listening