Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based...
Transcript of Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based...
![Page 1: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/1.jpg)
Image Segmentation
Jianping Fan
Dept of Computer Science
UNC-Charlotte
Course Website: http://webpages.uncc.edu/jfan/itcs5152.html
![Page 2: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/2.jpg)
Outlines
Seeded Region Growing
K-Means Clustering
Graph Cut & Normalized Cut
![Page 3: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/3.jpg)
Definitions
Based on sets.
Each image R is a set of regions Ri. Every pixel belongs
to one region.
One pixel can only belong to a single region.
S
i
iRR1
ji RR
R1
R3R2
R4
R6
R5
R7
![Page 4: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/4.jpg)
VC 14/15 - TP9 -
Region-Based
Segmentation
R1
R3
R2
R4
R6
R5
R7
![Page 5: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/5.jpg)
Basic Formulation
Let R represent the entire image region. Segmentation partitions R into n subregions, R1, R2, ..., Rn, such that:
a)
b)
c)
d)
e)
a) Every pixel must be in a region
b) Points in a region must be connected.
c) Regions must be disjoint.
d) All pixels in a region satisfy specific properties.
e) Different regions have different properties.
n
i
i RR1
.21 region, connected a is , ..., n, i Ri
jijiRR ji , and allfor
.,...,2,1for )( niTRUERP i
.for )( jiFALSERRP ji
![Page 6: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/6.jpg)
Region growing
Groups pixels into larger regions.
Starts with a seed region.
Grows region by mergingneighboring pixels.
Iterative process
How to start?
How to iterate?
When to stop?
Initial
RegionsIterations
Stop
Condition
Finish
![Page 7: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/7.jpg)
Similarity Criteria
Homogeneity of regions is used as the main segmentation criterion in region growing.
gray level
color, texture
shape
model
etc.
Choice of criteria
affects segmentation
results dramatically!
![Page 8: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/8.jpg)
Region Growing
Inter-Pixel Similarity Calculation
(a) Pixel Neighborhood Similarity Calculation
),( yx
)1,1( yx )1,( yx )1,1( yx
),1( yx ),1( yx
)1,1( yx )1,( yx )1,1( yx
![Page 9: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/9.jpg)
Region Growing
Inter-Pixel Similarity Calculation
(1) Inter-Pixel Similarity
|)1,1(),(|)]1,1(),,[( yxIyxIyxyxD
(2) Binary Classification
TyxyxDdissimilar
TyxyxDsimilaryxS
)]1,1(),,[(,,0
)]1,1(),,[(,,1),(
![Page 10: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/10.jpg)
Region Growing
Threshold Determination for Decision Making
Relationships among neighboring pixels can defined as:
similar versus dissimilar
0
1
1T
similar dissimilar
![Page 11: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/11.jpg)
Region Growing
Threshold Determination for Decision Making
Entropy for similar and dissimilar pixels:
![Page 12: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/12.jpg)
Region Growing
Threshold Determination for Decision Making
![Page 13: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/13.jpg)
Gray-Level Criteria
Comparing to Original Seed Pixel
Very sensitive to choice of seed point.
Comparing to Neighbor in Region
Allows gradual changes in the region.
Can cause significant drift.
Comparing to Region Statistics
Acts as a drift dampener.
Other possibilities!
![Page 14: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/14.jpg)
VC 14/15 - TP9 -
Region-Based
Segmentation
![Page 15: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/15.jpg)
Region merging
Algorithm
Divide image into an initial set of regions.
One region per pixel.
Define a similarity criteria for merging regions.
Merge similar regions.
Repeat previous step until no more merge operations are possible.
![Page 16: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/16.jpg)
Region Growing
Region Growing Results
![Page 17: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/17.jpg)
Region Growing Region Growing Results
![Page 18: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/18.jpg)
Region Growing
Region Growing Results
![Page 19: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/19.jpg)
Integrating Multi-modal Features for Region Growing
Image Segmentation
![Page 20: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/20.jpg)
Image Segmentation
Integrating Multi-modal Features for Region Growing
![Page 21: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/21.jpg)
Image Segmentation
Integrating Multi-modal Features for Region Growing
![Page 22: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/22.jpg)
Image Segmentation
Integrating Multi-modal Features for Region Growing
![Page 23: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/23.jpg)
Image Segmentation
Integrating Multi-modal Features for Region Growing
![Page 24: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/24.jpg)
Observations
Only segmentation from visual information cannot support automatic image understanding & interpretation!
Image segmentation results may not make sense to human beings!
Integrating Multi-modal Features for Region Growing
![Page 25: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/25.jpg)
25
Seeded Image Segmentation
Color image segmentation polices:
Threshold
Boundary-based
Region-based
Hybrid techniques
![Page 26: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/26.jpg)
26
Hybrid techniques
Seeding region growing (SRG)
Different merging order possibility
SRG
![Page 27: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/27.jpg)
27
Automatic seed selection algorithm
Condition 1:
A seed pixel candidate must have the similarity higher than a threshold values.
Condition 2:
A seed pixel candidate must have the maximum relative Euclidean distance to its eight neighbors less than a threshold value.
![Page 28: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/28.jpg)
28
Automatic seed selection algorithm
Calculate the maximum distance to its neighbors as :
of the pixel, of its neighbors
Not on the boundary
0.05
ii
dd8
1max max
8...1,
222
222
i
CCY
CCCCYYd
rb
rrbbi
i
ii
rbCYCii rbi CCY
Condition 2:A seed pixel candidate must have the maximum relative Euclidean distance to its eight neighbors less than a threshold value.
![Page 29: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/29.jpg)
29
Automatic seed selection algorithm
Connected seeds are considered as one seed.
the detected seeds are shown in red colorOriginal color image
![Page 30: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/30.jpg)
30
Region growing
The pixels that are unclassified and neighbors of at least one region, calculate the distance:
222
222
ii
ii
rbi
rrbbi
i
CCY
CCCCYYd
![Page 31: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/31.jpg)
31
Region growing
1. red pixels are the seeds and the green pixels are the
pixels in the sorted list T in a decreasing order of
distances.
2. the white pixel is the pixel with the minimum distance
to the seed regions
3. check its 4-neighbors
![Page 32: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/32.jpg)
32
Region growing
1. If all labeled neighbors of p have a same label, set p to this label.
2. If the labeled neighbors of p have different labels, calculate the
distances between p and all neighboring regions and classify p to
the nearest region.
3. Then update the mean of this region, and add 4 neighbors of p,
which are neither classified yet nor in T, to T in a decreasing order
of distances.
4. Until the T is empty.
![Page 33: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/33.jpg)
33
Region merging
Consider the color different and size of regions :
Color different between two adjacent region Ri and Rj is defined as:
Size
select of the total number of pixels in an image as the threshold.
150
1
222222
222
,min
,
ijii
jiji
rbjrbi
rrbbji
CCYCCY
CCCCYYRjRid
![Page 34: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/34.jpg)
34
Region merging
We first examine the two regions having the smallest color differentamong others. If < threshold, merge the two
regions and re-compute the mean of the new region.
We repeat the process until no region has the distance less than the threshold.
Threshold=0.1
RjRid ,
![Page 35: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/35.jpg)
35
Region merging
If the size with number of pixels in a region is smaller than a threshold, the region is merged into its neighboring region with the smallest color difference.
This procedure is repeated until no region has size less than the threshold.
![Page 36: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/36.jpg)
36
Experimental results
JSEG algorithm.
![Page 37: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/37.jpg)
37
Experimental results
JSEG algorithm.
JSEG algorithm.
![Page 38: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/38.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 38
K-Means Clustering
1. Partition the data points into K clusters
randomly. Find the centroids of each cluster.
2. For each data point:
Calculate the distance from the data point to
each cluster.
Assign the data point to the closest cluster.
3. Re-compute the centroid of each cluster.
4. Repeat steps 2 and 3 until there is no further
change in the assignment of data points (or in
the centroids).
![Page 39: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/39.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 39
K-Means Clustering
![Page 40: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/40.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 40
K-Means Clustering
![Page 41: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/41.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 41
K-Means Clustering
![Page 42: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/42.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 42
K-Means Clustering
![Page 43: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/43.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 43
K-Means Clustering
![Page 44: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/44.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 44
K-Means Clustering
![Page 45: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/45.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 45
K-Means Clustering
![Page 46: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/46.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 46
K-Means Clustering
![Page 47: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/47.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 47
K-Means Clustering
Example
Duda et al.
![Page 48: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/48.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 48
K-Means Clustering
RGB vector
x j i2
jelements of i' th cluster
iclusters
K-means clustering minimizes
![Page 49: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/49.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 49
Clustering
Example
D. Comaniciu and P. Meer, Robust
Analysis of Feature Spaces:
Color Image Segmentation, 1997.
![Page 50: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/50.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 50
K-Means Clustering
Example
Original K=5 K=11
![Page 51: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/51.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 51
K-means, only color is used in segmentation, four clusters (out of 20) are shown here.
![Page 52: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/52.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 52
K-means, color and position is used in segmentation, four clusters (out of 20) are shown here.
Each vector is (R,G,B,x,y).
![Page 53: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/53.jpg)
Bahadir K. Gunturk EE 7730 - Image Analysis I 53
K-Means Clustering: Axis Scaling
Features of different types may have
different scales.
For example, pixel coordinates on a
100x100 image vs. RGB color values in
the range [0,1].
Problem: Features with larger scales
dominate clustering.
Solution: Scale the features.
![Page 54: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/54.jpg)
Spectral Clustering for Image Segmentation
![Page 55: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/55.jpg)
Graph-based Image Segmentation
Image (I)
Graph Affinities(W)
IntensityColorEdges
Texture
Slide from Timothee Cour (http://www.seas.upenn.edu/~timothee)
![Page 56: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/56.jpg)
Graph-based Image Segmentation
Image (I)
)(
1
)(
1),(),(
BvolAvolBAcutBANcut
Slide from Timothee Cour (http://www.seas.upenn.edu/~timothee)
Graph Affinities(W)
IntensityColorEdges
Texture
![Page 57: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/57.jpg)
Graph-based Image Segmentation
Image (I)
)(
1
)(
1),(),(
BvolAvolBAcutBANcut
Eigenvector X(W)
Aiif
AiifiX
DXXWD
A0
1)(
)(
Graph Affinities(W)
IntensityColorEdges
Texture
![Page 58: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/58.jpg)
Graph-based Image Segmentation
Image (I)
)(
1
)(
1),(),(
BvolAvolBAcutBANcut
Eigenvector X(W)
Discretization
Slide from Timothee Cour (http://www.seas.upenn.edu/~timothee)
Aiif
AiifiX
DXXWD
A0
1)(
)(
Graph Affinities(W)
IntensityColorEdges
Texture
![Page 59: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/59.jpg)
Graph-based Image Segmentation
V: graph nodesE: edges connection nodes
G = {V,E}
PixelsPixel similarity
Slides from Jianbo Shi
![Page 60: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/60.jpg)
Graph terminology
Similarity matrix:
Slides from Jianbo Shi
2
2
2)()(
,X
ji XX
ji ew
jiwW ,
![Page 61: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/61.jpg)
Affinity matrix
Similarity of image pixels to selected pixelBrighter means more similar
Reshape
N*M pixels
N*M pixels
M pixels
N pixels
Warningthe size of W is quadratic
with the numberof parameters!
![Page 62: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/62.jpg)
Graph terminology
Degree of node:
Slides from Jianbo Shi
j
jii wd ,
……
![Page 63: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/63.jpg)
Graph terminology
Volume of set:
Slides from Jianbo Shi
Ai
i VAdAvol ,)(
![Page 64: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/64.jpg)
Graph terminology
AjAi
jiwAAcut,
,),(
Slides from Jianbo Shi
Cuts in a graph:
![Page 65: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/65.jpg)
Representation
Partition matrix X:
Pair-wise similarity matrix W:
Degree matrix D:
Laplacian matrix L:
KXXX ,...,1
j jiwiiD ,),(
WDL
segments
pix
els
),(),( jiaffjiW
![Page 66: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/66.jpg)
Pixel similarity functionsIntensity
Texture
Distance
2
2
2)()(
),( I
ji II
ejiW
2
2
2)()(
),( X
ji XX
ejiW
2
2
2)()(
),( c
ji cc
ejiW
![Page 67: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/67.jpg)
Pixel similarity functionsIntensity
Texture
Distance
2
2
2)()(
),( I
ji II
ejiW
2
2
2)()(
),( X
ji XX
ejiW
2
2
2)()(
),( c
ji cc
ejiW
here c(x) is a vector of filter outputs. A natural thing to do is to square the outputs
of a range of different filters at different scales and orientations,
smooth the result, and rack these into a vector.
![Page 68: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/68.jpg)
Definitions
Methods that use the spectrum of the affinity matrix to cluster are known as spectral clustering.
Normalized cuts, Average cuts, Average association make use of the eigenvectors of the affinity matrix.
Why these methods work?
![Page 69: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/69.jpg)
Spectral Clustering
Data Similarities
* Slides from Dan Klein, Sep Kamvar, Chris Manning, Natural Language Group Stanford University
![Page 70: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/70.jpg)
Eigenvectors and blocks
Block matrices have block eigenvectors:
Near-block matrices have near-block eigenvectors:
1 1 0 0
1 1 0 0
0 0 1 1
0 0 1 1
eigensolver
.71
.71
0
0
0
0
.71
.71
1= 2 2= 2 3= 0
4= 0
1 1 .2 0
1 1 0 -.2
.2 0 1 1
0 -.2 1 1
eigensolver
.71
.69
.14
0
0
-.14
.69
.71
1= 2.02 2= 2.02 3= -0.02
4= -0.02
* Slides from Dan Klein, Sep Kamvar, Chris Manning, Natural Language Group Stanford University
![Page 71: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/71.jpg)
Spectral Space Can put items into blocks by eigenvectors:
Clusters clear regardless of row ordering:
1 1 .2 0
1 1 0 -.2
.2 0 1 1
0 -.2 1 1
.71
.69
.14
0
0
-.14
.69
.71
e1
e2
e1 e2
1 .2 1 0
.2 1 0 1
1 0 1 -.2
0 1 -.2 1
.71
.14
.69
0
0
.69
-.14
.71
e1
e2
e1 e2
![Page 72: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/72.jpg)
How do we extract a good cluster?
Simplest idea: we want a vector x giving the
association between each element and a cluster
We want elements within this cluster to, on the whole, have strong affinity with one another
We could maximize
But need the constraint
This is an eigenvalue problem - choose the eigenvector of W with largest eigenvalue.
WxxT
1xxT
![Page 73: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/73.jpg)
Criterion for partition:
Minimum cut
BvAu
BAvuwBAcut
,,
),(min),(min
First proposed by Wu and Leahy
A
B
Ideal Cut
Cuts with
lesser weight
than the
ideal cut
Problem! Weight of cut is directly proportional to the number of edges in the cut.
![Page 74: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/74.jpg)
Normalized Cut
)(
1
)(
1),(),(
BvolAvolBAcutBANcut
Normalized cut or balanced cut:
Finds better cut
![Page 75: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/75.jpg)
Normalized Cut
Volume of set (or association):
VtAutuwVAassocAvol
,),(),()(
A
B
![Page 76: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/76.jpg)
Normalized Cut
Volume of set (or association):
Define normalized cut: “a fraction of the total edge connections to all the nodes in the graph”:
),(
),(
),(
),(),(
VBassoc
BAcut
VAassoc
BAcutBANcut
VtAutuwVAassocAvol
,),(),()(
),(
),(
),(
),(),(
VBassoc
BBassoc
VAassoc
AAassocBANassoc
A
B
A
B
Define normalized association: “how tightly on average nodes within the cluster are connected to each other”
A
B
![Page 77: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/77.jpg)
01DyTSubject to:
Observations(I)
Maximizing Nassoc is the same as minimizing Ncut, since they are related:
How to minimize Ncut? Transform Ncut equation to a matricial form.
After simplifying:
),(2),( BANassocBANcut
Dyy
yWDyxNcut
T
T
yx
)(min)(min
Rayleigh quotient
jjiWiiD ),(),(
NP-Hard!y’s values are
quantized
![Page 78: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/78.jpg)
Instead, relax into the continuous domain by solving generalized eigenvalue system:
Which gives:
Note that so, the first eigenvector is y0=1 with eigenvalue 0.
The second smallest eigenvector is the real valued solution to this problem!!
Observations(II)
DyyWD )(
01)( WD
max y yTDW y subject to y
TDy 1 min
![Page 79: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/79.jpg)
Algorithm
1. Define a similarity function between 2 nodes. i.e.:
2. Compute affinity matrix (W) and degree matrix (D).
3. Solve
4. Use the eigenvector with the second smallest eigenvalue to bipartition the graph.
5. Decide if re-partition current partitions.
Note: since precision requirements are low, W is very sparse and only few eigenvectors are required, the eigenvectors can be extracted very fast using Lanczos algorithm.
DyyWD )(
2
2
2)()(
2
2
2)()(
,X
ji
I
ji XXFF
ji ew
![Page 80: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/80.jpg)
Discretization Sometimes there is not a clear threshold to binarize
since eigenvectors take on continuous values.
How to choose the splitting point?
a) Pick a constant value (0, or 0.5).
b) Pick the median value as splitting point.
c) Look for the splitting point that has the minimum Ncut
value:
1. Choose n possible splitting points.
2. Compute Ncut value.
3. Pick minimum.
![Page 81: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/81.jpg)
Use k-eigenvectors
Recursive 2-way Ncut is slow.
We can use more eigenvectors to re-partition the graph, however:
Not all eigenvectors are useful for partition (degree of smoothness).
Procedure: compute k-means with a high k. Then follow one of these
procedures:
a) Merge segments that minimize k-way Ncut criterion.
b) Use the k segments and find the partitions there using exhaustive search.
Compute Q (next slides).
1 1 .2 0
1 1 0 -.2
.2 0 1 1
0 -.2 1 1
.71
.69
.14
0
0
-.14
.69
.71
e1
e2
e1 e2
![Page 82: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/82.jpg)
Example
Eigenvectors
Segments
![Page 83: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/83.jpg)
Experiments
Define similarity:
for point sets.
for brightness images.
for HSV images.
in case of texture.
2
2
2)()(
2
2
2)()(
,X
ji
I
ji XXFF
ji ew
1iF
iIiF
)]cos(..),sin(..,[ hsvhsvviF
|]*|,|,*[| 1 nfIfIiF
![Page 84: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/84.jpg)
Experiments (I)
Point set segmentation:
(a) Pointset generated by Poisson process. (b) Segmentation results.
![Page 85: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/85.jpg)
Experiments (II)
Synthetic images:
![Page 86: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/86.jpg)
Experiments (III)
Weather radar:
![Page 87: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/87.jpg)
Experiments (IV)
Motion segmentation
![Page 88: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/88.jpg)
Other methods
Average association
Use the eigenvector of W associated to the
biggest eigenvalue for partitioning.
Tries to maximize:
Has a bias to find tight clusters. Useful for Gaussian distributions.
B
BBassoc
A
AAassoc ),(),(
A
B
![Page 89: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/89.jpg)
Other methods
Average cut Tries to minimize:
Very similar to normalized cuts.
We cannot ensure that partitions will have a a tight within-group similarity since this equation does not have the nice properties of the equation of normalized cuts.
B
BAcut
A
BAcut ),(),(
![Page 90: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/90.jpg)
Other methods
![Page 91: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/91.jpg)
Other methods
20 points are randomly distributed from 0.0 to 0.512 points are randomly distributed from 0.65 to 1.0
Normalized cut
Average cut
Average association
![Page 92: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/92.jpg)
Other methods
20 points are randomly distributed from 0.0 to 0.512 points are randomly distributed from 0.65 to 1.0
Normalized cut
Average cut
Average association
![Page 93: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/93.jpg)
Other methods
20 points are randomly distributed from 0.0 to 0.512 points are randomly distributed from 0.65 to 1.0
Normalized cut
Average cut
Average association
![Page 94: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/94.jpg)
Spectral Clustering for Image Segmentation
Good news:
Simple and powerful methods to segment images.
Flexible and easy to apply to other clustering problems.
Bad news:
High memory requirements (use sparse matrices).
Very dependant on the scale factor for a specific problem.
2
2
2)()(
),( X
ji XX
ejiW
![Page 95: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/95.jpg)
Examples
2
2
2)()(
,X
ji XX
ji ew
Spectral Clutering
Images from Matthew Brand (TR-2002-42)
![Page 96: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/96.jpg)
Spectral clustering
Makes use of the spectrum of the similarity matrix of the data to cluster the points.
Solve clustering for
affinity matrix
w(i,j) distance node i to node j
![Page 97: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/97.jpg)
Graph terminologySimilarity matrix: Degree of node:
j
jii wd , jiwW ,
Volume of set: Graph cuts:
![Page 98: Principal Video Shot: Linking Low-level Perceptual ... · VC 14/15 - TP9 - Region-Based Segmentation R 1 R 3 R 2 R 4 R 6 R 5 R 7. Basic Formulation Let R represent the entire image](https://reader035.fdocuments.us/reader035/viewer/2022071000/5fbc089aac762222d63bbc10/html5/thumbnails/98.jpg)
Normalized cuts results