Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1...

40
Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2 , Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City Universi ty of Hong Kong 2. Dept. of CSE, The Chinese University of Hong Kong Accepted by IEEE TPAMI

Transcript of Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1...

Page 1: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Self-Validated Labeling of MRFs for Image Segmentation

Wei Feng

1,2, Jiaya Jia

2 and Zhi-Qiang Liu

1

1. School of Creative Media, City University of Hong Kong2. Dept. of CSE, The Chinese University of Hong Kong

Accepted by IEEE TPAMI

Page 2: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Outline

Motivation Graph formulation of MRF labeling Graduated graph cuts Experimental results Conclusion

Page 3: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Outline

Motivation Graph formulation of MRF labeling Graduated graph cuts Experimental results Conclusion

Page 4: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Self-Validated Labeling

Common problem: segmentation, stereo etc.

Self-validated labeling: two parts Labeling quality: accuracy (i.e., likelihood) and

spatial coherence Labeling cost (i.e., the number of labels)

Bayesian framework: to minimize the Gibbs energy (equivalent form of MAP)

coherencelikelihoodEEE

Page 5: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Motivation

Computational complexity remains a major weakness of the MRF/MAP scheme

Robustness to noise Preservation of soft boundaries Insensitive to initialization

Page 6: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Motivation

Self-validation: How to determine the number of clusters? To segment a large number of images Global optimization based methods are

robust, but most are not self-validated Split-and-merge methods are self-

validated, but vulnerable to noise

Page 7: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Motivation

For a noisy image consisting of 5 segments

Let’s see the performance of the state-of-the art methods

Page 8: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Motivation

Normalized cut (NCut)

[1]

Unself-validated segmentation (i.e., the user needs to indicated the number of segments, bad)

Robust to noise (good) Average time: 11.38s (fast, good) NCut is unable to return satisfying result when fe

eded by the right number of segments 5; it can produce all “right” boundaries, mixed with many “wrong” boundaries, only when feeded by a much larger number of segments 20.

[1] J. Shi and J. Malik, “Normalized cuts and image segmentation”, PAMI 2000.

Page 9: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Motivation

Bottom-up methods E.g., Mean shift [2] E.g., GBS [3]

Self-validated (good) Very fast (< 1s, good) But, sensitive to noise

(bad)

[2] D. Comaniciu and P. Meer. “Mean shift: A robust approach towards feature space analysis”, PAMI 2002.[3] P. F. Felzenszwalb and D. P. Huttenlocher. “Efficient graph based image segmentation”, IJCV 2004.

Page 10: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Motivation

Data-driven MCMC[4]

Self-validated (good) Robust to noise

(good) But, very slow (bad)

[4] Z. Tu and S.-C. Zhu, “Image segmentation by data-driven Markov chain Monte Carlo”, PAMI 2002.

Page 11: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Motivation

As a result, we need a self-validated segmentation method, which is fast and robust to noise.

Our method: graduated graph mincut Tree-structured graph cuts (TSGC) Net-structured graph cuts (NSGC) Hierarchical graph cuts (HGC)

Time #Seg

TSGC 2.96s 5

NSGC 5.7s 5

HGC 2.01s 6

Page 12: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Motivation

[5] C. D’Elia, G. Poggi, and G. Scarpa, “A tree-structured Markov random field model for Bayesian image segmentation,” IEEE Trans. Image Processing, vol. 12, no. 10, pp. 1250–1264, 2003.

[5]

Page 13: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Outline

Motivation Graph formulation of MRF labeling Graduated graph cuts Experimental results Conclusion

Page 14: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Graph Formulation of MRFs

Graph formulation of MRFs (with second order neighborhood system N2): (a) graph G = <V,E> with K segments {L1, L2 . . . LK } and observation Y; (b) final labeling corresponds to a multiway cut of the graph G.

Page 15: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Graph Formulation of MRFs

Property: Gibbs energy of segmentation Seg(I) can be defined as

MRF-based segmentation ↔ multiway (K-way) graph mincut problem (NP-complete, K=2 solvable)

Page 16: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Outline

Motivation Graph formulation of MRF labeling Graduated graph cuts Experimental results Conclusion

Page 17: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Graduated Graph Mincut

Main idea To gradually adjust the optimal labeling acco

rding to the Gibbs energy minimization principle.

A vertical extension of binary graph mincut (in constrast to horizontal extension, α-expansion and α-β swap)

Page 18: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Graduated Graph Mincut

Page 19: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Binary Labeling of MRFs

Page 20: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Binary Labeling of MRFs

Page 21: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Tree-structured Graph Cuts

Page 22: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Tree-structured Graph Cuts

Page 23: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Tree-structured Graph Cuts

: (over-segmentation)

Page 24: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Net-structured Graph Cuts

Page 25: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Net-structured Graph Cuts

Page 26: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Net-structured Graph Cuts

Page 27: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Hierarchical Graph Cuts

Page 28: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Hierarchical Graph Cuts

Page 29: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Graduated Graph Cuts

Summary An effective tool for self-validated labeling

problems in low level vision. An efficient energy minimization scheme by

graph cuts. Converting the K-class clustering into a

sequence of K−1 much simpler binary clustering. Independent to initialization Very close good local minima obtained by α-

expansion and α-β swap

Page 30: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Segmentation Evolution

Iter #1Iter #2Iter #3Iter #4Mean image

Page 31: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Outline

Motivation Graph formulation of MRF labeling Graduated graph cuts Experimental results Conclusion

Page 32: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Comparative Results

Comparative Experiments

Page 33: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Robustness to Noise

Robust to noise

Page 34: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Preservation of Soft Boundary

Page 35: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Consistency to Ground Truth

Page 36: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Coarse-to-Fine Segmentation

Page 37: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Performance Summary

Page 38: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Outline

Motivation Graph formulation of MRF labeling Graduated graph cuts Experimental results Conclusion

Page 39: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Conclusion

An efficient self-validated labeling method that is very close to good local minima and guarantees stepwise global optimum

Provides a vertical extension to binary graph cut that is independent to initialization

Ready to apply to a wide range of clustering problems in low-level vision

Page 40: Self-Validated Labeling of MRFs for Image Segmentation Wei Feng 1,2, Jiaya Jia 2 and Zhi-Qiang Liu 1 1. School of Creative Media, City University of Hong.

Thanks!Thanks!