Image segmentation by clustering in the color space

1
Image segmentation by clustering in the color
space

• CIS581 Final Project
• Student Qifang Xu
• Advisor Dr. Longin Jan Latecki

2
Content

• Introduction
• Project Algorithm
• Project program
• Experiments and results
• Conclusion

3
Introduction

• Image segmentation is to find objects or
  meaningful part of objects.
• Two types of segmentation methods
• homogeneity
• contrast
• Image segmentation techniques
• region growing and shrinking
• clustering methods
• Boundary detection

4
Clustering Techniques

• Segment an image by grouping each elements based
  on some measure of similarity
• Domain spaces
• spatial domain (row-column (rc) space)
• color space
• histogram spaces
• other complex feature space

5
Clustering Algorithms

• Basic idea
• 1. Iteratively divide the space of interest into
  regions by median.
• 2. stop when the specific criteria is reached.
• k-means clustering
• Recursive region splitting algorithm
• standard technique
• 1. compute histograms for each component of
  interest (red, green, blue)
• 2. select a best threshold to split the image
  into two regions
• 3. Repeat 1 and 2, until no new regions can be
  created

6
Project Algorithm

• A combination of the k-medoid algorithm and
  classification trees techniques
• Feature space RGB
• Computation time is linear to the number of
  feature vectors
• Flow chart for the project

7
(No Transcript)
8
Project program (1)

• Main file rgbcluster.m
• firstauto3(inputFileName, mnp, percent)
• inputFileName image for segmentation
• mnp Minimal number of points in each cluster
• percent parameter that delays clustering of
  points in a margin region. Value 0.01-- 0.1.
  Normal value 0.05.
• output a set of clusters

9
Project program (2)

• Color index
• each entry R G B weights
• weight is the number of pixels for this color
• function colorWeights getStat(data, map)
• colorWeights map
• row, col size(map)
• for i 1row
• tmp find(datai)
• colorWeights(i, col1)
  length(tmp)
• end
• return

10
Project program (3)

• Means
• meanR
• meanG
• meanB

11
Codes for find rgb mean function mn
find_mean(data) row, col size(data) sumR
0 sumG 0 sumB 0 count 0 for i
1row sumR sumR data(i, 1) data(i,
col) sumG sumG data(i, 2) data(i,
col) sumB sumB data(i, 3) data(i,
col) count count data(i,
col) end thisMean(1) sumR /
count thisMean(2) sumG / count thisMean(3)
sumB / count mn thisMean return
12
Project program (4)

• Distance Euclidean distance between two points
• Codes
• function dist dist2pt(x, y)
• dist sqrt((x(1)-y(1))2 (x(2)-y(2))2
  (x(3)-y(3))2)
• return

13
Project program (5)

• Split2.m (provided by Dr. Latecki)
• lmainindex,rmainindex,ldata,ldist,rdata,rdist,cl
  uster,no,centroid split2(data,dist,a,mainindex,
  scale,mnp,percentage,cluster,no,centroid)

14
Project program (6)

• recurauto1.m recursively split data into
  clusters (provided by Dr. Latecki)
• 1. distance histogram (myhist)
• 2. threshold (evo2)
• 3. split (split2)
• 4. find left distance, go to left branch
• 5. find right distance, go to right branch
• 6. no new split, stop
• Unclustered points
• assigned to the clusters with closest distance
  to the centroids.

15
Experiments and results

• Rgb pictures
• Cherry flowers
• House
• tiger
• airplane
• car
• people

16
Resluts 1
Original pictures
segmented pictures
Mnp 30, percent 0.05, cluster number 4
Mnp 20, percent 0.05, cluster number 7
17
Results 2
Original pictures
Segmented pictures
Mnp 10, percent 0.05, cluster number 9
Mnp 50, percent 0.05, cluster number 3
18
Results 3
Original pictures segmented
pictures
Mnp 10, percent 0.05, cluster number 11
Mnp 30 Percent 0.05 Cluster number 4
19
Image size 2MB, mnp 30, cluster number 5
20
Original picture
Mnp10, cluster number 15
Mnp 30, Cluster number 4
21
Results movies
22
Conclusion

• Advantages
• no predefined cluster number
• user interactive
• computation time
• Disadvantages
• spatial information lost
• Cannot deal with noise or outliers