By : Trusha Y.

Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

1

Image Segmentation
UNIT-7
Digital Image Processing
 Digital Image Processing 2

Image segmentation

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Subdivide an image into its constituent regions or
objects

 Stop when the objects of interest in an application

have been isolated

 The goal is to simplify and/or change the

representation of an image into something that is
more meaningful and easier to analyze
 Digital Image Processing 3

Image segmentation

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Used to locate objects and boundaries in image

 Accuracy determines the eventual success or failure of

computerized analysis procedures

 Control over the environments, wherever is possible

(industrial applications); not possible in autonomous
target acquisition
Digital Image Processing 4

Principal approaches

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

Discontinuity Similarity
 Digital Image Processing 5

Principal approaches

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

Discontinuity Similarity

• Partition an image based on abrupt

(sudden) changes in intensity
Digital Image Processing 6

Principal approaches

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

Discontinuity Similarity

• Partition an image into regions

• That are similar according to a set of
predefined criteria
Digital Image Processing 7

Principal approaches

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

Discontinuity Similarity

isolated point Thresholding

line Region growing

edge Region splitting

 By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa
8

DETECTION OF DISCONTINUITIES
Digital Image Processing
 Digital Image Processing 9

Basic

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 3 basic types of gray-level discontinuities
1. Points
2. Lines
3. Edges

 The common way is to run a mask through the image

w1 w2 w3

w4 w5 w6

w7 w8 w9
mask
 Digital Image Processing 10

Basic

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 The response of mask at any point is given as

 zi is the gray-level of the pixel associated with mask

coefficient wi
 Digital Image Processing 11

1. Point detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Laplacian mask is used

 Point can be detected at (x,y) location

If absolute value of response of mask at that point if
greater the threshold

-1 -1 -1 0 -1 0

-1 8 -1 -1 4 -1

-1 -1 -1 0 -1 0

Laplacian mask with diagonal Laplacian mask without diagonal

Digital Image Processing 12

1. Point detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

where
g(x,y) : output intensity at (x,y)
R(x,y) : mask response at (x,y)
T : threshold
13

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

1. Point detection
Digital Image Processing
 Digital Image Processing 14

2. Line detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

-1 -1 -1 -1 -1 2 -1 2 -1 2 -1 -1
2 2 2 -1 2 -1 -1 2 -1 -1 2 -1
-1 -1 -1 2 -1 -1 -1 2 -1 -1 -1 2

Horizontal +45o Vertical -45o

 If the 1st mask were moved around the image

It would respond more strongly to lines oriented
horizontally
 Digital Image Processing 15

2. Line detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 With a constant background
The maximum response would result when
Line passed through middle row of mask

 Same idea applied to other mas

2nd mask respond max for line at +45o
3rd mask respond max for vertical line
4th mask respond max for line at -45o
 Digital Image Processing 16

2. Line detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Apply every mask on the image
let R1, R2, R3, R4 denote the response of
the horizontal, +45o, vertical and -45o masks
respectively

 If at a certain point in the image

|Ri| > |Rj| for all j≠ i
that point is said to be more likely associated with a
line in the direction of mask i
 Digital Image Processing 17

2. Line detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 If we are interested in detecting all lines
simply run the mask through the image
and threshold the absolute value of the result
18

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

2. Line detection
Digital Image Processing
 Digital Image Processing 19

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 An edge is a set of connected pixels that lie on the
boundary between two regions

 ideal edge

 Set of connected pixels each of which is located at an

orthogonal step transition in gray level
 Digital Image Processing 20

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 edge having ramplike profile

 In practice because of the imperfection in image

acquisition the edges are blurred

 So they are modeled as ramplike profile

21

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

3. Edge detection
Digital Image Processing
 Digital Image Processing 22

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 The thickness of the edge is determined by the length
of the ramp

 Blurred edges tend to be thick

 Sharp edges tend to be thin

 Digital Image Processing 23

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Find the 1st and 2nd order derivative of ramp

 1st order derivative  2nd order derivative

 +ve at onset of ramp  +ve at beginning of ramp

 0 at constant intensity  -ve at end point of ramp

 0 at onset of ramp & at

constant intensity
24

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

3. Edge detection
Digital Image Processing
25

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

3. Edge detection
Digital Image Processing
 Digital Image Processing 26

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Magnitude of the 1st order derivative can be used to
detect the presence of an edge at a point in an image

 Computed using Gradient

 Sign of 2nd order derivative used to determine whether

an edge pixel lies on the dark or light side of an edge

 Computed using Laplacian

 Digital Image Processing 27

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Gradient operation

 The gradient of an image f(x,y) at location (x,y) is

 Digital Image Processing 28

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Gradient operation

 Magnitude of this vector is

 Direction angle of gradient with respect to x-axis is

 Digital Image Processing 29

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Gradient operation

 Direction of an edge at (x,y) is perpendicular to the

direction of the gradient vector at that point

 Gradient of image can be computed using

• Roberts cross-gradient operators

• Prewitt operators
• Sobel operators
 Digital Image Processing 30

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Gradient operation

 Roberts cross-gradient operators

-1 0 0 -1
0 1 1 0

+45o -45o
 Digital Image Processing 31

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Gradient operation

 Prewitt operators

-1 -1 -1 -1 0 1 0 1 1 -1 -1 0

0 0 0 -1 0 1 -1 0 1 -1 0 1

1 1 1 -1 0 1 -1 -1 0 0 1 1

Horizontal Vertical -45o +45o

 Digital Image Processing 32

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Gradient operation

 Sobel operators

-1 -2 -1 -1 0 1 0 1 2 -2 -1 0

0 0 0 -2 0 2 -1 0 1 -1 0 1

1 2 1 -1 0 1 -2 -1 0 0 1 2

Horizontal Vertical -45o +45o

 Digital Image Processing 33

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Gradient operation

 In practice Prewitt and Sobel mask are mostly used

 Mask is used to find Gx and Gy

 Once Gx and Gy is found, computation of magnitude is

complex it required squares and square root
 Digital Image Processing 34

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Gradient operation

 So an approach used frequently is

35

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

3. Edge detection
Digital Image Processing
36

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

3. Edge detection
Digital Image Processing
Digital Image Processing 37

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

Sobel +45 and -45 mask
 Digital Image Processing 38

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Laplacian

 Laplacian of image f(x,y) is 2nd order derivative

 Two forms of Laplacian are mostly used

 Digital Image Processing 39

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Laplacian

0 -1 0 -1 -1 -1

-1 4 -1 -1 8 -1

0 -1 0 -1 -1 -1
 Digital Image Processing 40

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Laplacian

 For several reason Laplacian is not used in its original

form for edge detection

• Laplacian is sensitive to noise

• Magnitude of Laplacian produces double edges
• Finally Laplacian is unable to detect edge direction
 Digital Image Processing 41

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Laplacian

 Role of Laplacian in segmentation

• Zero crossing property for edge location

• To know whether a pixel is on the dark or light side of an
edge
 Digital Image Processing 42

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Laplacian

 Laplacian is combined with smoothing as a precursor to

finding edges via zero-crossing

Where

σ is the standard deviation

r 2 = x 2 + y2
 Digital Image Processing 43

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Laplacian

 Laplacian of h is

 This function called Laplacian of Gaussian (LoG)

 Due to its shape, LoG is also called Mexican hat

44

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

3. Edge detection
Digital Image Processing

 Laplacian
 Digital Image Processing 45

3. Edge detection

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Laplacian

 The role of Gaussian in LoG is

to smooth the image
smoothing image reduce the effect of noise

 The role of Laplacian is

to provide an image with zero crossing used to
establish the location of edges
46

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

3. Edge detection
Digital Image Processing
Digital Image Processing 47

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

EDGE LINKING & BOUNDARY
DETECTION
 Digital Image Processing 48

Basic

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Edge detection algorithms are followed by linking
procedure to assemble edge pixels into meaningful
edges

 Approaches
1. Local pixel processing
2. Global processing via Hough transform
3. Global processing via graph-theory techniques
 Digital Image Processing 49

1. Local pixel processing

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Analyze the characteristics of pixels in a small
neighborhood ( 3⨯3 or 5⨯5 ) about every point (x,y) in
an image that has been labeled an edge point

 All points that are similar according to a set of

predefined criteria are linked
 Digital Image Processing 50

1. Local pixel processing

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Two principal properties used for establishing
similarity of edges pixels are

1. The strength of the response of the gradient operator

used to produce the edge pixel

Where E is nonnegative threshold

 Digital Image Processing 51

1. Local pixel processing

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Two principal properties used for establishing
similarity of edges pixels are

2. The direction of the gradient vector

Where A is nonnegative angle threshold

 Digital Image Processing 52

1. Local pixel processing

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Neighborhood of (x,y) is linked to (x0,y0)
If both magnitude and direction criteria are satisfied

 This process is repeated at every location in image

 Digital Image Processing 53

2. Global processing via Hough Transformation

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Points are linked by determining if they lie on a curve
of specified shape

 “n” points are there in image, recognized by edge pixel

 Want to find subset of these points that lie on straight

lines
 Digital Image Processing 54

2. Global processing via Hough Transformation

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 1st solution

 Find all lines determined by every pair of points

 Find all subsets of points that are close to particular

lines
 Digital Image Processing 55

2. Global processing via Hough Transformation

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 2nd solution is Hough Transformation

 Point (xi,yi)

 General equation for straight line

yi = axi + b

 Many lines pass through (xi,yi)

 All satisfy the line equation with different value of a

 Digital Image Processing 56

2. Global processing via Hough Transformation

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 2nd solution is Hough Transformation

 Write this equation as

b = -xia + yi

 Consider the ab-plane

 Another point (xj,yj) also has a line in this plane with

 With both line intersect at (a’,b’)

 Digital Image Processing 57

2. Global processing via Hough Transformation

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 2nd solution is Hough Transformation
 Digital Image Processing 58

2. Global processing via Hough Transformation

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 2nd solution is Hough Transformation

 Subdivide the parameter space (ab-plane)

into accumulator cells

 (amin,amax) and (bmin,bmax)

expected range of slope and intercept values

 Cell at (i,j) with accumulator A(i,j) corresponds to (a i,bj)

 Digital Image Processing 59

2. Global processing via Hough Transformation

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 2nd solution is Hough Transformation
 Digital Image Processing 60

2. Global processing via Hough Transformation

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 2nd solution is Hough Transformation

 For every point (xk,yk) take equal valur of “a” and solve
the equation to find corresponding “b”

 Resulting “b” then rounded off to nearest allowed value

in b-axis

 If ap result in bq then do A(p,q) = A(p,q) + 1

 Digital Image Processing 61

2. Global processing via Hough Transformation

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 2nd solution is Hough Transformation

 At the end value of Q in A(i,j) correspond to Q point in

xy-plane lying on the line y = aix + bj

 Number of subdivisions in ab-plane determines the

accuracy of the colinearity of these pints
 Digital Image Processing 62

2. Global processing via Hough Transformation

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 2nd solution is Hough Transformation
 Digital Image Processing 63

3. Global Processing via Graph-Theoretic Techniques

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Edge detection and linking based on representing edge
segments in the form of a graph

 Searching graph for low-cost paths that correspond to

significant edges

 Process is more complicated and required more

processing time
 Digital Image Processing 64

3. Global Processing via Graph-Theoretic Techniques

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Basic of graph theory

 A graph G=(N,U)

• N : nonempty set of node

• U : pair of distinct elements of N (set of arcs)

 Graph can be directed / undirected

 if graph is undirected then arc is unordered pair

 Digital Image Processing 65

3. Global Processing via Graph-Theoretic Techniques

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Basic of graph theory

 if graph is directed then arc is ordered pair

 In directed graph if arc from ni to nj

then ni called parent and nj called successor

 Process of identifying the successor of a node is called

expansion of the node
 Digital Image Processing 66

3. Global Processing via Graph-Theoretic Techniques

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Basic of graph theory

 Graph is divided in levels

• Single node on level-0 called root / start node

• Nodes on last level are called goal nodes

 A sequence of nodes n1 , n2 , … , nk called path from n1

to nk where each node ni is successor of ni-1
 Digital Image Processing 67

3. Global Processing via Graph-Theoretic Techniques

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Basic of graph theory

 Cost c(ni,nj) can be associated with each arc (ni,nj)

 The cost of path from n1 to nk is

 Digital Image Processing 68

3. Global Processing via Graph-Theoretic Techniques

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Let edge element is boundary between two pixels p
and q , such that p and q are 4-neighbors

 Edge element is defined by pair (xp,yp)(xq,yq)

 Digital Image Processing 69

3. Global Processing via Graph-Theoretic Techniques

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Edge detection in 3×3 image

 Find cost for each pair of p and q

 Draw a graph

 Select the graph with low-cost

 Edge detected for that path

Digital Image Processing 70

3. Global Processing via Graph-Theoretic Techniques

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

gray value
of p
Highest gray value
gray value of q
Gray-level value
ates
rd in 𝑐 ( 𝑝, 𝑞 ) =𝐻 −[ 𝑓 ( 𝑝 ) − 𝑓 (𝑞)]
oo
ixel c
P
Digital Image Processing 71

3. Global Processing via Graph-Theoretic Techniques

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

72

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

THRESHOLDING
Digital Image Processing
 Digital Image Processing 73

Thresholding

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Histogram for Light object in dark background

{
𝑎 𝑖 𝑓 𝑓 ( 𝑥 , 𝑦 ) >𝑇 2
{
𝑔 ( 𝑥 , 𝑦 )= 1
𝑖 𝑓 𝑓 (𝑥, 𝑦)≥ 𝑇
0 𝑖𝑓 𝑓 ( 𝑥 , 𝑦 )<𝑇
𝑔 ( 𝑥 , 𝑦 )= 𝑏 𝑖𝑓 𝑇 1< 𝑓 ( 𝑥 , 𝑦 )≤ 𝑇 2
𝑐 𝑖𝑓 𝑓 ( 𝑥 , 𝑦 ) ≤ 𝑇 1

Bimodel Multimodel
Histogram Histogram

1 light object 2 light objects

in in
Dark background Dark background
74

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

Role of illumination
Digital Image Processing
Digital Image Processing 75

Types of thresholding

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

Global Local Dynamic / Adaptive
Thresholding Thresholding Thresholding

Depends on
Depends only on Depends on
Spatial coordinates
f(x,y) f(x,y) and p(x,y)
x and y
 Digital Image Processing 76

Global thresholding

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

1. Select an initial estimate for T

3. Compute the average intensity values μ1 and μ2 for

2. Segment the image in G1 and G2 using T

G1 and G2
4. Compute a new threshold value

5. Repeat steps 2 to 4 until the difference in T n

successive integration is smaller than a predefined
parameter T0
77

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

Global thresholding
Digital Image Processing
78

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

Global thresholding
Digital Image Processing
 Digital Image Processing 79

Adaptive thresholding

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Subdivide image into small area

 Use different threshold for each subimage

 Since threshold used for each pixel depends on the

location of the pixel in terms of the subimage
80

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

Adaptive thresholding
Digital Image Processing
81

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

REGION BASED SEGMENTATION
Digital Image Processing
 Digital Image Processing 82

Basic

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 R : entire image
 View segmentation as process that partitions R into n
subregions R1 , R2 , … , Rn

a)
b) Ri is a connected region , i = 1 , 2 , … , n
c) Ri ∩ Rj = ∅ for all i and j , i ≠ j
d) P(Ri) = TRUE for i = 1 , 2 , … , n
e) P(Ri ∪ Rj) = FALSE for i ≠ j
 Digital Image Processing 83

Region growing

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Group pixels or regions into large regions based on
predefined criteria

 Start with seed point

 Growing is done by appending each seed

if its neighbors are having similar properties
 Digital Image Processing 84

Region growing

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Important points

 Selection of seed
 Selection of similarity criteria
 Formulation of stopping rule
85

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

Histogram
Region growing
Digital Image Processing
 Digital Image Processing 86

Region splitting and merging

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

 Image is subdivided into a set of arbitrary, disjoint
regions and then merge and/or split the regions to
satisfy the predefined condition

 The entire region is R

 Subdivided into smaller and smaller quadrant regions

so that for any region Ri
P ( Ri ) = TRUE
 Digital Image Processing 87

Region splitting and merging

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

1. Split into 4 disjoint quadrants for which
P (Ri) = TRUE

2. When n further splitting is possible, merge any

adjacent region Ri and Rj for which
P ( Ri ∪ Rj ) = TRUE

3. Stop when no further merging is possible

88

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

Region splitting and merging
Digital Image Processing
Digital Image Processing 89

Region splitting and merging

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

R1 R2

R31 R32 R41 R42

R3 R4
R331 R332
R33 R34 R43 R44
R333 R334
90

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

Region splitting and merging
Digital Image Processing
91

By : Trusha Y. Gajjar, CE Dept, CSPIT, CHARUSAT, Changa

UNIT - 7
END
OF
Digital Image Processing