You are on page 1of 32

Computao Visual e Multimdia

10504: Mestrado em Engenharia Informtica

Chap. 7 Region Segmentation


Region Segmentation

Outline

Chapter 7: Region Segmentation


Image segmentation:
a reminder

Chapter 7: Region Segmentation


Region segmentation:

Mean Shift: A Robust Approach toward Feature Space Analysis, by D. Comaniciu and P. Meer
http://www.caip.rutgers.edu/~comanici/MSPAMI/msPamiResults.html

http://robots.stanford.edu/cs223b/index.html

Image segmentation:
in pictures

Chapter 7: Region Segmentation

Edge segmentation:

Chapter 7: Region Segmentation


What is a region?

Chapter 7: Region Segmentation


Region-based approach

Chapter 7: Region Segmentation


Chapter 7: Region Segmentation


Region-based segmentation

n
i=1

Ri = I

Ri R j =
P(Ri ) = TRUE,

i = 1,2,n
i

P Ri R j = FALSE

Chapter 7: Region Segmentation


Comparison of histogram, region growing 


and deformable contour segmentations

3

2

3

5

2

3

4

5

4

5

6

3

6

6

7

9

12

11

11

8

7

3

10

11

9

12

10

8

4

22

15

17

18

18

3

2

9

10

19

16

9

3

1

3

3

1

2

5

1

3

2

3

5

2

3

4

5

4

5

6

3

6

6

7

9

12

11

11

8

7

3

10

11

9

12

10

8

4

22

15

17

18

18

3

2

9

10

19

16

9

3

1

3

3

1

2

5

1

3

2

3

5

2

3

4

5

4

5

6

3

6

6

7

9

12

11

11

8

7

3

10

11

9

12

10

8

4

22

15

17

18

18

3

2

9

10

19

16

9

3

1

3

3

1

2

5

1

3

2

3

5

2

3

4

5

4

5

6

3

6

6

7

9

12

11

11

8

7

3

10

11

9

12

10

8

4

22

15

17

18

18

3

2

9

10

19

16

9

3

7

9

12

11

11

8

7

3

10

11

9

12

10

8

4

22

15

17

18

18

3

2

9

10

19

16

9

3

1

3

3

1

2

5

1

3

2

3

5

2

3

4

threshold T11

threshold T10

3

2

3

5

2

3

4

5

4

5

6

3

6

6

1

3

3

1

2

5

1

3

2

3

5

2

3

4

5

4

5

6

3

6

6

7

9

12

11

11

8

7

3

10

11

9

12

10

8

4

22

15

17

18

18

3

2

9

10

19

16

9

3

5

4

5

6

3

6

6

7

9

12

11

11

8

7

3

10

11

9

12

10

8

4

22

15

17

18

18

3

2

9

10

19

16

9

3

1

3

3

1

2

5

1

threshold T12

1

3

3

1

2

5

1

3

2

3

5

2

3

4

5

4

5

6

3

6

6

7

9

12

11

11

8

7

3

10

11

9

12

10

8

4

22

15

17

18

18

3

2

9

10

19

16

9

3

1

3

3

1

2

5

1

3

10

11

9

12

10

8

4

22

15

17

18

18

3

2

9

10

19

16

9

3

1

3

3

1

2

5

1

region growing with variance of 2 in respect to value 11 with reference to threshold T11

3

2

3

5

2

3

4

5

4

5

6

3

6

6

7

9

12

11

11

8

7

3

10

11

9

12

10

8

4

22

15

17

18

18

3

2

9

10

19

16

9

3

1

3

3

1

2

5

1

3

2

3

5

2

3

4

5

4

5

6

3

6

6

7

9

12

11

11

8

7

3

10

11

9

12

10

8

4

22

15

17

18

18

3

2

9

10

19

16

9

3

1

3

3

1

2

5

1

3

2

3

5

2

3

4

5

4

5

6

3

6

6

7

9

12

11

11

8

7

3

10

11

9

12

10

8

4

22

15

17

18

18

3

2

9

10

19

16

9

3

deformable contour to meet threshold T11


1

3

3

1

2

5

1

3

2

3

5

2

3

4

5

4

5

6

3

6

6

7

9

12

11

11

8

7

Chapter 7: Region Segmentation


seed

http://uei.ensta.fr/baillie

Region growing segmentation


growing

final region

Seed-based region growing segmentation:


pixel aggregation

Chapter 7: Region Segmentation


The seed point can be selected either by a human or automatically by


avoiding areas of high contrast (large gradient) => seed-based method.

Problem: To isolate the strongest lightning region of the


image on the right hand side without splitting it apart.

Solution: To choose the points having the highest gray-scale
value which is 255 as the seed points shown in the image
immediately below.

original image

threshold = 255

returns multiple

seeds

threshold:

225~255

threshold:

190~225

threshold:

155~255

http://en.wikipedia.org/wiki/Region_growing

Chapter 7: Region Segmentation


Seed-based region growing 


segmentation: example

http://uei.ensta.fr/baillie

Chapter 7: Region Segmentation


Fast scanning algorithm:


a kind of unseeded region segmentation

If the criterion of homogeneity is local (compared to the value of


the candidate pixel and the pixel of the border) => linear method.

Threshold T:
P1 == P2 iff Diff(Col(P1),Col(P2)) < T

x

y

val=?

x==y:
x<>y:




val = x


boundary(x)y if |x-y|T


new region index if |x-y|>T



1
1

Chapter 7: Region Segmentation



http://uei.ensta.fr/baillie

Region growing segmentation:


advantages & disadvantages

Region splitting and merging segmentation


original image

splitting & merging


Chapter 7: Region Segmentation


thresholding seg.

http://uei.ensta.fr/baillie

Chapter 7: Region Segmentation


Region splitting: example

In this example, the criterion of homogeneity is the variance of 1.


original image

split 1

split 2

split 3

Splitting & merging: data structures


Chapter 7: Region Segmentation


RAG with adjacency relations (in


red) for big black region.

RAG with adjacency relations (in


red) for big black region.

http://astro.temple.edu/~siddu

Splitting & merging segmentation algorithm


Chapter 7: Region Segmentation


watershed crest line


watersheds

http://en.wikipedia.org/wiki/Watershed_(image_processing)#cite_note-1

Watershed segmentation

Chapter 7: Region Segmentation


Watershed segmentation
by flooding

Chapter 7: Region Segmentation


original image

This technique aims at identifying all the third type of


points (i.e., points of watershed lines) for segmentation

3D topographic
surface

Watershed segmentation by flooding



255

255

0
255

0
255

http://euclid.ii.metu.edu.tr/~ion528/demo/lectures/6/4/index.html

Chapter 7: Region Segmentation


Watershed segmentation algorithm


Chapter 7: Region Segmentation


Watershed segmentation algorithm:


dam construction

Chapter 7: Region Segmentation


original image

watershed of the
gradient image

gradient image

final contours

http://cmm.ensmp.fr/~beucher/wtshed.html

Flooding-based watershed segmentation  Chapter 7: Region Segmentation



applied to gradient image

Marker-controlled watershed 


segmentation (gradient image)

http://cmm.ensmp.fr/~beucher/wtshed.html

Chapter 7: Region Segmentation


original image

markers of the blobs



and of the background

over-segmented image

marker-controlled watershed
of the gradient image

Inter-pixel watershed segmentation



http://en.wikipedia.org/wiki/Watershed_(image_processing)#cite_note-2

Chapter 7: Region Segmentation


More complex segmentation methods


Chapter 7: Region Segmentation


Snakes

Copyright G.D. Hager


Chapter 7: Region Segmentation


Images taken from http://www.cs.bris.ac.uk/home/xie/


content.htm

Level sets

Copyright G.D. Hager


Chapter 7: Region Segmentation


Images taken from http://www.cgl.uwaterloo.ca/~mmwasile/


cs870/

Graph cuts

Copyright G.D. Hager


Chapter 7: Region Segmentation


Images taken from efficient graph-based segmentation paper


Generalized PCA
(Rene Vidal)

Human

Copyright G.D. Hager


Chapter 7: Region Segmentation


GPCA

Summary:

Chapter 7: Region Segmentation

You might also like