IMAGE SEGMENTATION

1
IMAGE SEGMENTATION

• By
• SAVITHA EMANI

2
Overview

• Introduction
• Types of Segmentation
• -Thresholding
• -Region based
• -Edge based
• -Hough Transform
• -Watershed approach
• -Other Methods
• Conclusion

3
INTRODUCTION
4
Segmentation

• Image segmentation is the process of partitioning
  the digital image into multiple regions that can
  be associated with the properties of one or more
  criterion
• It is an initial and vital step in pattern
  recognition-a series of processes aimed at
  overall image understanding.

5
Region/Segment

• Region an aggregation of pixels
• Properties like gray level, color, texture, shape
  help to identify regions and similarity of such
  properties, is used to build groups of regions
  having a particular meaning.

6

• In mathematical sense the segmentation of the
  image I, which is a set of pixels, is the
  partition of I into n disjoint sets R1,R2, . . .
  , Rn, called segments or regions such that their
  union of all regions equals I. 
• I R1 U R2 U.. U Rn

7
SEGMENTATION TECHNIQUES
8

• Thresholding
• Clustering
• Region based
• Edge based
• Model based
• Watershed approach
• Other Methods

9
Region Based Segmentation

• Region growing
• Split Merge
• Pyramid linking

10
Region growing

• Begin with a set of seed points and from these
  grow regions by appending to each seed those
  neighboring pixels that have properties similar
  to initial seed.

11
Split and merge

• Follows the Quad tree method to partition into
  segments

12
Pyramid linking
13
Edge Based Segmentation

• Edge Image Thresholding
• Edge Relaxation
• Border tracing
• Simple border tracing
• Border tracing in gray-level images
• Border detection as graph searching

14
Edge Relaxation

• e has a vertex at each of its ends and three
  possible border continuations can be found from
  both of these vertexes.
• All three possible edge positions at the end of
  the edge e must be included to cover all the
  possible ways the border can continue from both
  ends of e.

15

• The type of edge e can then be represented using
  a number pair i-j describing edge patterns at
  each vertex, where i and j are the vertex types
  of the edge e.
• By symmetry, we only list the cases where iltj.
  The following context situations are possible
• 0-0 isolated edge -- negative influence on the
  edge confidence
• 0-2, 0-3 dead end -- negative influence on edge
  confidence
• 0-1 uncertain -- weak positive, or no influence
  on edge confidence
• 1-1 continuation -- strong positive influence on
  edge confidence
• 1-2,1-3 continuation to border intersection --
  medium positive influence on edge confidence
• 2-2,2-3,3-3 bridge between borders -- not
  necessary for segmentation, no influence on edge
  confidence.

16
(No Transcript)
17
Border tracing

• Simple border tracing
• Algorithm Inner boundary tracing
• Search the image from top left until a pixel of a
  new region is found. This pixel P0 then has the
  minimum column value of all pixels of that region
  having the minimum row value. Pixel P0 is a
  starting pixel of the region border. Define a
  variable dir which stores the direction of the
  previous move along the border from the previous
  border element to the current border element.
  Assign
• (a) dir 3 if the border is detected in
  4-connectivity
• (b) dir7 if the border is detected in
  8-connectivity.
• Search the 33 neighborhood of the current pixel
  in an anti-clockwise direction, beginning the
  neighborhood search in the pixel positioned in
  the direction
• (a) (dir3)mod 4 if the border is detected in
  -connectivity
• (b) (dir3)mod 8 if dir is even (dir6)mod 8 if
  dir is odd
• The first pixel found with the same value as the
  current pixel is a new boundary element Pn.
• Update the dir value.
• If the current boundary element Pn is equal to
  the second border element P1, and if the previous
  border element Pn-1 is equal to P0 , stop.
  Otherwise repeat step 2.
• The detected inner border is represented by
  pixels P0 Pn-2 .

18

• Algorithm Outer boundary tracing
• Trace the inner region boundary in -connectivity
  until done
• The outer boundary consists of all non-region
  pixels that were tested during the search
  process if some pixels were tested more than
  once, they are listed more than once in the outer
  boundary list.
• Note that some border elements may be repeated in
  the outer border up to three times. See the
  following figure.
• The outer region border is useful for deriving
  properties such as perimeter, compactness, etc.

19
Border detection as graph searching

• Two nodes ni and nj corresponding to two
  8-connected adjacent pixels xi and xj and are
  connected by an arc if the edge directions ø(xi)
  and ø(xj) match the local border direction in the
  following sense. We can apply the following rules
  to construct the graph To connect a node ni
  representing the pixel xi with a node nj
  representing the pixel xj,
• Pixel xj must be one of three existing neighbors
  of xi in the direction d? ø(xi)-p/4 ,
  ø(xi)p/4
• s(xi) ands(xj) must be greater than T, where T is
  some preset threshold of edge significance.
• Another common requirement is to connect two
  nodes only if the difference of their edge
  direction is less than p/2.

20

• A-algorithm graph search
• Expand the starting node nA and put all its
  successors into an OPEN list with pointers back
  to the starting node nA . Evaluate the cost
  function for expanded node
• If the OPEN list is empty, fail Determine the
  node ni from the OPEN list with the lowest
  associated cost f(ni) and remove it. If ni nB,
  then track back through the pointers to find the
  optimum path and stop
• If the option to stop was not taken in step 2,
  expand the specified node ni, and put its
  successors on the OPEN list with pointers back to
  ni. Compute their costs f. Go to step 2.

21
Hough transform
22
(No Transcript)
23
Watershed approach
24
(No Transcript)
25
Other Segmentation Methods

• Texture segmentation
• Curve fitting

26
Conclusion

• Hence in this paper we have seen a brief
  introduction to image segmentation and several
  segmentation methods. Image segmentation is used
  in several of image processing.
• Applications of image segmentation include
  identifying objects in a scene for object-based
  measurements such as size and shape ,identifying
  objects in a moving scene, identifying objects
  which are at different distances from a sensor,
  computer graphics, medical imaging etc.

27
References

• Digital Image Processing (Gonzalez and Woods)
• Digital Image Processing(Kenneth R. Castleman)
• Digital Image Processing(Bernd Jahne )
• Digital Image Processing(Jain)
• Digital Image Processing(William K. Pratt)
• http//iria.pku.edu.cn/jiangm/courses/dip/html/no
  de125.html
• http//noodle.med.yale.edu/alums/chakrab/pap1/sect
  ion3_1.html
• http//en.wikipedia.org/wiki/Image_segmentation
• http//www.prip.tuwien.ac.at/yll/papers/Unpublish
  ed/segmentation2pageintro.pdf?PHPSESSID9d9dd44c5e
  441480ad1331a4ef0af136
• www.eee.bham.ac.uk/spannm/Teaching20docs/Computer
  20Vision20Course/Image20Segmentation.ppt
•  

28
Thank you