You are on page 1of 26

Submitted by

K.GAJALAKSHMI
Under the Guidance of
Prof.V.SRINIVASAN
Department of Computer
Science & Engg


Introduction
Segmentation Methods
Objective
Literature Review
Methodology
Implementation
Results and Future work



An image is defined as a two dimensional
function F(x,y),where x and y are special coordinates
and the amplitude of F at any pair of coordinates (x,y)
is called the intensity or grey level of the image at that
point. When x,y and the amplitude values of F are all
finite, discrete quantities ,then image is called a digital
image. There are three types of images. They are
1. Gray-scale images
2. Binary images
3. Color images

Electronic data processing on a 2-D array of numbers.
Techniques which identifies shades, colors and relationship.
Examples of image processing operations
Image preprocessing
Image enhancement
Image restorations
Edge Detection
Image segmentation
Image compression

Process of isolating objects in the image from the
background
Partitioning the image into disjoint regions, such
that each region is homogeneous with respect to
some property, such as gray value or texture.
Segmentation subdivides the given image into its
constitute parts of objects

1. Discontinutity
2. Similartity

Edge-based approach: Image edges are detected and then
linked into contours that represent the boundaries of
image objects
Clustering-based approach: Image pixels are sorted in
increasing order as a histogram according to their intensity
values.
Region-based approach: The goal is the detection of regions
that satisfy a certain predefined homogeneity threshold.
Region-based approaches are available because the
segmented contours are always continuous and one-pixel-
wide
Split/merge approach: Input image is first segmented into
homogeneous primitive regions using K-means as a Split
step. Then, similar neighboring regions are merged according
to a certain decision rule as a Merge step.
Watershed approach: watersheds being the divide lines of the
domains of attraction of rain falling over the region. An
alternative approach is to imagine the landscape being
immersed in a lake, with holes pierced in local minima.
Basins (also called `catchment basins') will fill up with water
starting at these local minima. The watershed approach is
simple yet powerful tool very suitable to solve the
segmentation.
Toboggan approach: This approach links each pixel with the
smallest of its neighbors. The only input to the tobogganing
algorithm is a measure of discontinuity at each pixel in the
image

To segment medical image.
To segment the images and extract the object of
interest from the background using marker controlled
watershed segmentation.
To avoid over segmentation.

SI.NO Authors Title of papers Journal/year Remarks
1. Marleen de
Bruijne, Wiro
J. Niessen, J. B.
Antoine
Maintz, and
Max A.
Viergever
Localization and
Segmentation of Aortic
Endografts Using Marker
Detection
IEEE-April 2003 graft contour
detection
Second order
derivative
Localized by the
position of its markers
2. K. K. Parvati,
B. S. Prakasa
Rao and
M. Mariya Das
Image Segmentation
Using Gray-Scale
Morphology and Marker-
Controlled Watershed
Transformation
Hindawi Publishing
Corporation
Discrete Dynamics
in Nature and
Society -2008
foreground and
background markers
using morphological
image
reconstructions.
extract desired parts
of only gray-scale
images
3. Sbastien
Lefvre

Knowledge from Markers
in Watershed
Segmentation
locate the markers,
which are used only as
the initial starting
positions of the
regions to be
segmented.
contents of the
markers through a
supervised pixel
classification
SI.N
O
Authors Title of papers Journal/year Remarks
Jong-Bae
Kim, Hang-
Joon Kim
Multiresolution-based
watersheds for efficient
image segmentation
Pattern
Recognition
(2003)
Pyramid
representation
Image
segmentation
Region merging
region projection
Grau, V.,
Mewes, A.,
Alcaniz, M.,
Kikinis, R.,
Warfield, S
Improved watershed
transform for medical
image segmentation
using prior information,
IEEE Transactions
on Medical
Imaging-April
2004
Normal
distribution for the
objects in the
image
Each marker or
class being
represented by the
mean and variance
of its pixel values
Erlend
Hodneland ,
Xue-Cheng
Tai , Hans-
Hermann
Gerdes
Four-Color Theorem and
Level Set Methods for
Watershed
Segmentation
The
topographical
distance function is
applied in a level
set formulation to
perform the
segmentation
Roberto
Rodrguez.,
Teresa E.
Alarcn,
Oriana
Pacheco
A new strategy to obtain
robust markers for blood
vessels segmentation by
using the watersheds
method
Computers in
Biology and
Medicine-2005
extracting an
approximate region
containing the
blood vessel and
part of the
background near
the blood vessel.
segmenting the
blood vessel from
the background
within this
region.
Maria Frucci
Giuliana
Ramella,
Gabriella
Sanniti di
Baja
Using resolution
pyramids for watershed
image segmentation
Image and Vision
Computing -2007
resolution
pyramid
Lower resolution
Watershed
transform
Issam Dagher
Kamal El
Tom
Water Balloons: A hybrid
watershed Balloon Snake
segmentation
Image and Vision
Computing -2008

Water balloons
provide the advantage
of reducing over-
segmentation.
Richard Beare A Locally Constrained
Watershed Transform
IEEE transactions
on pattern
analysis and
machine
intelligence, 2006
Border of the
regions is
constrained
Mariela A.
Gozalez,
Teresita R.
Cuadrado,
Virginia L. Ballarin
Comparing marker
definition algorithms for
Watershed segmentation
in microscopy images
October 2008 The WT is a
segmentation
method, two
algorithms
based on fuzzy
inference
systems and
clustering in
order to obtain
markers that
allow robust
segmentations
Fuzzy-time
high
Alessandro
Cristoforetti ,
Luca Faes , Flavia
Ravelli, Maurizio
Centonze ,
Maurizio Del Greco
, Renzo Antolini ,
Giandomenico
Nollo,
Isolation of the left arial
surface from cardiac
multi-detector CT images
based on marker
controlled watershed
segmentation
Medical
Engineering &
Physics -2008
Considered
pixel as a object.
O. R. Vincent,O.
Folorunso
A Descriptive Algorithm
for Sobel Image Edge
Detection
Proceedings of
Informing Science
& IT Education
Conference
(InSITE)-2009
Image edge
detection is a
process of
locating the edge
of an image
which is
important in
finding the
approximate
absolute
gradient
magnitude at
each point I of
an input
grayscale image
10
Yung-Chieh
Lin, Yu-Pao
Tsai, Yi-
Ping Hung
and Zen-
Chung Shih
Comparison
Between
Immersion-Based
and
Toboggan-Based
Watershed Image
Segmentation
IEEE
TRANSACTIONS
ON IMAGE
PROCESSING, VOL.
15, NO. 3, MARCH
2006
Starting from minima of
lowest altitude
Starts from low to high
Slow and more memory
requirement
High to low altitude
Riding a sled downhill
Fast and requires less
memory
Sbastien
Delest,
Romuald
Bon and
Hubert
Cardot
3d watershed
transformation on
connected vertices
structure
Image and Vision
Computing (2007)
watershed
transformation with two
complementary
methods to limit the
over segmentation.
Markers offer an
interesting solution to
force the region location
and hierarchical
merging allows
browsing the merging
tree to find the desired
level of merging.
Analysis ,processing and understanding of digital image
often involve many different algorithm.
Segmentation is crucial process.
Watershed is powerful tool to segment a image.
-over segmentation
-irreverent objects
-undesired regions.

Over come this problems marker controlled watershed
segmentation is considered.
The properties which will be used to mark the desired
region
Using object marker and background marker


The original watershed method produces severe over
segmentation
many local minima small basins a lot of
watersheds

Read the color image and convert it to gray scale
Gradient magnitude as the segmentation
function
Mark the foreground objects
Compute background marker
Compute the watershed segmentation.
Get color image as input and convert into gray scale
image


SOBEL operator consists of a pair of 3 3 convolution
kernel. One kernel is simply, other rotated by .
G
x
=

G
y
=

This operation can be described by the expression
x (i, j) = mask
Gx
* l (i, j)
y (i, j) = mask
Gy
* l (i, j)
Where l (i , j) is indicating some neighborhood of
pixel (i, j) and * denotes convolution
(
(
(


1 2 1
0 0 0
1 2 1
(
(
(

1 0 1
2 0 2
1 0 1




(ii) The gradient magnitude is given by:




Approximate magnitude is computed using:


2 2
Y x
G G G + =
y x
G G G + =
Image consisting of either single marker points or
larger marker regions.

Marker has a one-to-one relationship

Number of markers will be equal to the final number
of watershed regions.

Using drawing tool to draw

1. Initialize value of the pixels belonging to the watersheds to zero
2. Sort the pixels of input image in the increasing order of their gray
values.
3. Let h
min
and h
max
designate the lowest and highest pixel values in
the image
4. pixels are accessed directly through the sorted array. Check if
pixel distance = 1 then that pixel is added into queue.
5. This process repeated until level h
-1
.if the pixel distance less than
zero or distance greater than 1 then that pixel is marked as
watershed.
6. Compare each pixel (p' ) with neighborhood (p). Whether p'
belongs to already labeled basin or watershed.
7. check whether new minima have been discovered.
8. The distance associated with p is reset to 0.
9. The step 3 to 8 repeated until the queue is empty.

Proposed marker controlled watershed segmentation for
medical image.
The initial steps such as edge detection, creating marker has
been completed.
Watershed segmentation has been implemented and over
segmentation detected.
Further to implement the marker controlled segmentation in
order to avoid the over segmentation problem and to extract
parts of interest from the medical image are to be carried out in
the next phase.

1. Marleen de Bruijne, Wiro J. Niessen, J. B. Antoine Maintz, and Max A. Viergever, Localization and Segmentation of Aortic Endografts Using Marker Detection, IEEE Transactions On Medical Imaging,
Vol. 22, No. 4, April 2003 473
2. K. K. Parvati,1 B. S. Prakasa Rao, and M. Mariya Das, Image Segentation Using Gray-Scale Morphology and Marker-Controlled Watershed Transformation ,Hindawi Publishing Corporation Discrete
Dynamics in Nature and Society Volume 2008, Article ID 384346, 8 pages
3. Sbastien Lefvre Knowledge from Markers in Watershed Segmentation
4. Sbastien Delest, Romuald Bon and Hubert Cardot, 3d watershed transformation on connected vertices structure.
5. Beucher s, and lantuejoul c., Use of watersheds in contour detection, int. workshop on image processing ,real- time edge and motion detection/estimation ,CCETT/INSA/IRISA, IRIsa report n132 pp2.1-
2.12,rennes,france,17-21 sept1979
6. Meyer f. and beucher s., Morphological segmentation ,journal of visual communication and Image representation ,vol.1,n1,pp 21-46 ,sept 1990 Academic Press.
7. Jong-Bae Kim, Hang-Joon Kim Multiresolution-based watersheds for efficient image segmentation Received 1 November 2001; received in ,revised form 7 June 2002 ,Pattern Recognition Letters 24 (2003)
473488.
8. Grau, V., Mewes, A., Alcaniz, M., Kikinis, R., Warfield, S.: Improved watershed transform for medical image segmentation using prior information, IEEE Transactions on Medical Imaging 23(4) (April
2004) 447458.
9. Erlend Hodneland , Xue-Cheng Tai , Hans-Hermann Gerdes ,Four-Color Theorem and Level Set Methods for Watershed Segmentation
10. Roberto Rodrguez., Teresa E. Alarcn, Oriana Pacheco ,A new strategy to obtain robust markers for blood vessels segmentation by using the watersheds method , Received 2 April 2003; accepted 1 June
2004 Computers in Biology and Medicine 35 (2005) 665686.
11. Maria Frucci , Giuliana Ramella, Gabriella Sanniti di Baja, Using resolution pyramids for watershed image segmentation, Received 25 November 2005; received in revised form 26 June 2006; accepted 12
July 2006.
12. Richard Beare, A Locally Constrained Watershed Transform, IEEE Transactions On Pattern Analysis And Machine Intelligence, Vol. 28, No. 7, July 2006 1063.
13. Yung-Chieh Lin, Yu-Pao Tsai, Yi-Ping Hung and Zen-Chung Shih, Comparison Between Immersion-Based and Toboggan-Based Watershed Image Segmentation ,IEEE Transactions On Image Processing,
Vol. 15, No. 3, March 2006
14. Issam Dagher , Kamal El Tom WaterBalloons,A hybrid watershed Balloon Snake segmentation, Received 7 March 2006; received in revised form 3 September 2007; accepted 28 October 2007
15. Mariela A. Gonzalez, Teresita R. Cuadrado, Virginia L. Ballarin ,Comparing marker definition algorithms for Watershed segmentation in microscopy images ,October 2008
16. Alessandro Cristoforetti , Luca Faes , Flavia Ravelli, Maurizio Centonze , Maurizio Del Greco , Renzo Antolini , Giandomenico Nollo, Isolation of the left atrial surface from cardiac multi-detector CT
images based on marker controlled watershed segmentation ,Received 8 September 2006; received in revised form 28 December 2006; accepted 5 January 2007, Medical Engineering & Physics 30 (2008) 48
58
17. O. R. Vincent,O. Folorunso, A Descriptive Algorithm for Sobel Image Edge Detection, Proceedings of Informing Science & IT Education Conference (InSITE) ,2009

You might also like