Image Segmentation - Home | Computer ScienceSegmentation by region growing CSE 166, Fall 2017 16...
Transcript of Image Segmentation - Home | Computer ScienceSegmentation by region growing CSE 166, Fall 2017 16...
Image Segmentation
Image ProcessingCSE 166
Lecture 17
Reading
• Digital Image Processing, 4th edition– Chapter 10: Image segmentation I: edge detection, thresholding, and region detection
CSE 166, Fall 2017 2
Thresholding
CSE 166, Fall 2017 3
Single threshold Dual threshold
Histograms
Noise and thresholding
CSE 166, Fall 2017 4
Noise
Varying background and thresholding
CSE 166, Fall 2017 5
Intensity ramp Product of input and intensity ramp
Input
Basic global thresholding
CSE 166, Fall 2017 6
Intensity ramp ThresholdInput
Optimum global thresholding
CSE 166, Fall 2017 7
Histogram
Basic global thresholding
Optimum global
thresholding using Otsu’s method
Input
Image smoothing to improve global thresholding
CSE 166, Fall 2017 8
Without smoothing
With smoothing
Otsu’s method
Image smoothing does not always improve global thresholding
CSE 166, Fall 2017 9
Without smoothing
With smoothing
Otsu’s method
Edges to improve global thresholding
CSE 166, Fall 2017 10
Mask image (thresholdedgradient
magnitude)
Input
Masked input
Optimum global
thresholding using Otsu’s method
Edges to improve global thresholding
CSE 166, Fall 2017 11
Input
Masked input
Optimum global
thresholding using Otsu’s method
Mask image (thresholded absolute Laplacian)
Variable thresholding
CSE 166, Fall 2016 12
Global thresholding
Local standard deviations
Local thresholding
using standard deviations
Input
Variable thresholding
CSE 166, Fall 2016 13
Global thresholding
Local thresholding using moving averages
Input(spot shading)
Variable thresholding
CSE 166, Fall 2016 14
Global thresholding
Local thresholding using moving averages
Input(sinusoidal shading)
Segmentation by region growing
CSE 166, Fall 2017 15
Initial seedimage
InputX‐ray image
Final seed image
Output image
Segmentation by region growing
CSE 166, Fall 2017 16
Difference image thresholded with the smallest of the dual
thresholds
Segmentation by region growing
Difference image thresholded using dual thresholds
Difference image
Advanced segmentation methods
• k‐means clustering• Superpixels• Graph cuts
CSE 166, Fall 2017 17
Segmentation using k‐means clustering
CSE 166, Fall 2017 18
Input Segmentation using k‐means, k = 3
Superpixels
CSE 166, Fall 2017 19
Image of 4,000 superpixels with
boundaries
Image of 4,000 superpixels
Input image of 480,000 pixels
Superpixels
CSE 166, Fall 2017 20
500 superpixels 250 superpixels1,000 superpixels
Superpixels for image segmentation
CSE 166, Fall 2017 21
Segmentation using k‐means, k = 3
Superpixel image (100 superpixels)
Input image of 301,678 pixels
Segmentation using k‐means, k = 3
Images as graphs
CSE 166, Fall 2017 22
Simple graph with edges only between
4‐connected neighbors
Stronger (greater weight) edges are darker
Graph cuts for image segmentation
CSE 166, Fall 2017 23
Cut the weak edges
Graph cuts for image segmentation
CSE 166, Fall 2017 24
Smoothed input Graph cut segmentation
Input