IMAGE

SEGMENTATION
 Image segmentation refers to the process of partitioning a
digital image into multiple segments (sets of pixels, also
known as super pixels).
 The goal of segmentation is to simplify and/or change the
representation of an image into something that is more
meaningful and easier to analyze
 SEGMENTATION TECHNIQUES

 POINT,LINE AND EDGE DETECTION.

 HOUGH TRANSFORM PEAK DETECTION.

 THRESHOLDING.

 REGION BASED SEGMENTATION.

 WATERSHED TRANSFORM.
 THRESHOLDING
 It is the process of separating an image into different
regions based on its grey level distribution.
 Consider an image containing a homogenous object and
background as shown in figure with grey levels 200 & 80
respectively
 Separation of the object pixels from the background pixels Is
accomplished by selecting a grey level value k such that all
pixels with in the image with f (x, y)>k will be classified as
pixels belonging to the object
 The goal of threshold is to select a threshold value that
separates an image into distinct grey levels
 Types of thresholding

1.Multilevel thresholding
2.Optimal thresholding
 REGION BASED SEGMENTATION

 It is a segmentation technique that is based on finding the

regions directly

 Various methods are adopted under the region based

segmentation are
1.BASIC FORMULATION
2.REGION GROWING
3.REGION SPLITTING AND MERGING
 BASIC FORMULATION
 The basic formulation for Region-Based Segmentation is:

(b)Ri is a connected region, i = 1, 2, ...,n

(d)P(Ri) = TRUE for i = 1,2,...,n.

 (a) means that the segmentation must be complete; that
is, every pixel must be in a region.

 (b) requires that points in a region must be connected in

some predefined sense. 

 (c) indicates that the regions must be disjoint.

 (d) deals with the properties that must be satisfied by the

pixels in a segmented region. For example P(Ri) = TRUE
if all pixels in Ri have the same gray level.

 (e) indicates that region Ri and Rj are different in the

sense of predicate P.
 P(Ri) is a logical predicate defined over the points in set
P(Rk) and is the null set.
 (a) means that the segmentation must be complete; that is,
every pixel must be in a region.
 (b) requires that points in a region must be connected in
some predefined sense.
 (c) indicates that the regions must be disjoint.

 (d) deals with the properties that must be satisfied by the

pixels in a segmented region. For example P(Ri) = TRUE if
all pixels in Ri have the same gray level.
 (e) indicates that region Ri and Rj are different in the sense
of predicate P.
 REGION GROWING

 This is a procedure that groups pixels or sub regions into

larger regions.
 The first step in region growing is to select a set of seed
points.
 Seed point selection is based on some user criterion.

 The initial region begins as the exact location of these

seeds.
 The regions are then grown from these seed points to
adjacent points depending on a region membership
criterion.
 Since the regions are grown on the basis of the criterion,
the image information itself is important
 ADVANTAGES

 1. Region growing methods can correctly separate the

regions that have the same properties we define.
 2. Region growing methods can provide the original images
which have clear edges the good segmentation results.
 3. The concept is simple. We only need a small numbers of
seed point to represent the property we want, then grow the
region.
 4. We can determine the seed points and the criteria we
want to make.
 5. We can choose the multiple criteria at the same time.

 6. It performs well with respect to noise.

 DISADVANTAGES

 1. The computation is consuming, no matter the time or

power.
 2. Noise or variation of intensity may result in holes or over
segmentation.
 3. This method may not distinguish the shading of the real
images.
 REGION SPLITTING AND MERGING
 Let R represents the entire image region and select a
predicate p one approach of segmenting R is to divide it
successively into smaller regions and smaller quadrant
regions so that for any region Ri,P(Ri)=TRUE if
P(R)=FALSE we divide the image into quadrants if P is
false for any quadrant we subdivide the quadrants into
sub quadrants
 This particular splitting technique has a convenient
representation on the form of a tree as shown
R R2

R41 R42
R3
R43 R44
 R

R1 R2 R3 R4

R41 R42

R43 R44
 Figure (a) shows a portioned image and figure (b) shows its
corresponding quad tree
 The root of the tree corresponds to the entire image and that
each node corresponds to the subdivision of a node into four
descend nodes
 If only splitting is used the final portion normally contain
adjacent region with identical properties.
 This drawback can be avoided by merging

 Two adjacent regions Rj and Rk are merging

 The steps are
Split into four disjoint quadrants any region Ri for
which P(Ri)=false
When no further splitting is possible, merge any
adjacent regions Rj &Ri for which P(RjURi)=true
Stop when no further merging is possible
 WATERSHED TRANSFORM
 Watershed is a ridge that divides areas drained by
different river systems
 A catchments basin is the geographical area draining into
a river.
 The watershed transform applies the ideas to gray scale
image processing can be used to solve a variety of image
segmentation problems.
 TYPES OF WATERSHED TRANSFORM

 Water shed segmentation using distance transform

 Watershed segmentation using gradients
 Marker controlled watershed segmentation