Filetype PDF | Posted on 30 Jan 2023 | 2 years ago
Authentication
digital resources
149x Filetype PDF File size 0.63 MB Source: www.ijser.org
International Journal of Scientific & Engineering Research, Volume 7, Issue 12, December-2016 304
ISSN 2229-5518
SEGMENTATION TECHNIQUES IN IMAGE PROCESSING
Preeti Panwar
Department of Computer Science & Applications, Kurukshetra University, Kurukshetra
preetipanwar102@gmail.com
Girdhar Gopal
Department of Computer Science & Applications, Kurukshetra University, Kurukshetra
girdhar.gopal@kuk.ac.in
Rakesh Kumar
Department of Computer Science & Applications, Kurukshetra University, Kurukshetra
rakeshkumar@kuk.ac.in
ABSTRACT
Image processing is a form of signal processing. One of the mostly used operations of image processing is
image segmentation. Over the last few year image segmentation plays a vital role in image processing .
The goal of image segmentation is to partition the pixels into silent image segments i.e., these segments
corresponding to individual objects, natural parts of objects, or surface. The problems of digital image
segmentation represent great challenges for computer vision. Segmentation techniques which are used in
image processing are edge based, region based, thresholding, clustering etc.In this paper, different image
segmentation techniques have been discussed.
Keywords: Image, Digital Image processing, Image segmentation, Thresholding.
1. Introduction common example is the television image. Analog
image processing is mainly used for the hard copies
Image processing is the general issue in today’s like printouts and photographs.
world, in the field of computer vision. Image Digital image processing: - The term digital image
processing is the form of signal processing where processing generally refers to processing of a two-
both the input and output signals are images. An dimension image by a digital computer. The
image may be defines as a two-dimensional principle advantages of digital image processing
IJSER
functional, f(x, y), where x and y are spatial methods are its repeatability, versatility, and the
coordinates, and f is called the grey level or preservation of original data precision. The three
intensity of the image at that point. Images can be general phases of digital image processing are pre-
two-dimensional signals via a matrix representation, processing, enhancement, display information
and image processing can be understood by employ extraction.
one-dimensional signal processing techniques to
two-dimensional signals. Applications of image 2. Digital image processing:
processing are satellite imaging, medical imaging,
photography, and image compression etc[1] Digital image processing uses computer algorithm
to perform image processing on images to improve
Image processing basically includes the following the quality of the image by removing noise and
two steps: other unwanted pixels and also to obtain more
• Importing the image via image acquisition information on the image. There are fundamental
tools; steps in digital image processing. These steps are
• Analysing and manipulating the image. image acquisition, image enhancement, image
restoration, color image processing, wavelets and
multi resolution processing, compression,
1.1 Methods of image processing:- morphological processing, segmentation,
representation and description, object recognition.
There are two type of methods used for image [2] The fundamental steps of image processing are
processing namely, analog and digital image as follows:
processing.
Analog image processing refers to the modification 1. Image acquisition: Image acquisition is
of image through electrical means. The most refer as to acquire the image. Acquisition is
IJSER © 2016
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 7, Issue 12, December-2016 305
ISSN 2229-5518
as simple as an image that is already in
digital form. Generally, image processing 3. Image segmentation
is involves by the image acquisition.
2. Image enhancement: Image enhancement is Among the various image processing
among the simplest and most appealing techniques, image segmentation is very
areas of digital. The enhancement important step to analyse the given image and
techniques are used to enhance the detail extract data from them[4]. Segmentation is a
that is simply to highlight certain features process to subdivide the image into small image
of an image. region and that region corresponding to
3. Image restoration: Image restoration is individual surfaces, objects, or natural parts of
used to improving the appearance of an objects. The level of subdivision is depending
image. Image restoration techniques tend to on the problem being solved. That is,
be based on mathematical or probabilistic segmentation should stop when the objects of
models of image degradation. interest have been achieved.
4. Color image processing: Color image The goal of segmentation is to change and
processing is an area that has been widely simplify the representation of an image into
used now days because of rapidly use of something that is more useful and simple to
digital image over the internet. analyse. However, it is the process of assigning
5. Wavelets: Wavelets are used for a label to every pixel in an image such that
representating images in various degrees of pixels with the same label have some
resolution. Basically this is used for predefined characteristics. All of the pixels in a
pyramidal representation and image data region are similar with respect to some property
compression in which images are such as colour, intensity, or texture. Some
subdivided successively into smaller applications of image segmentation are image
regions. processing, medical imaging, computer vision,
6. Compression: Compression is a technique digital libraries, face recognition, image and
that is used for reducing the storage video retrieval, satellite image. [5].Based on
required for saving an image or the different technologies, image segmentation
bandwidth required for transmitting it.
IJSER
7. Morphological processing: Image approaches are currently divided into following
components that are useful in categories, based on two properties of image.
representation and description of shape are Detecting Discontinuities:-It divide an image
extracted from morphological processing based on short change in intensity, this includes
and it deals with image extraction tools image segmentation algorithms like edge
8. Segmentation:Segmentation subdivides an detection.[1]
image into its essential parts or objects.
The level of subdivision is depends on the Detection Similarities:-It divides an image into
problem being viewed. regions that are similar according to a
9. Representation and description: predefined criterion, this includes image
Representation and description follow the segmentation algorithm like region growing,
output of a segmentation stage, which usual and region splitting and merging,
is row pixel data, constituting either the thresholding.[6]
boundary of a region or all the points in the
region itself.
Description also called feature selection,
deals with extracting the information that
result in some qualitative information of
interest. It differentiate the one class of
objects from another.
10. Recognition: - Recognition is the process
of assigning a label (e.g. “vehicle”) to an 3.1 Edge-based techniques:
object based on its descriptors.[3]
IJSER © 2016
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 7, Issue 12, December-2016 306
ISSN 2229-5518
Edge detection is a major tool for image 1. Select a group of seed pixels within an
segmentation. Edge defines the boundaries between image.
regions in an image. Edge detection of an image 2. Select a set of similarity criterion such as
significantly reduces the amount of data and filters grey level intensity or color and setup a
out unusable information, while keep the important stopping rule.
structural properties in an image. 3. Grow regions by attaching each seed those
have predefined properties similar to seed
It could detect the variation of grey levels, but it is pixels.
sensitive to noise. It is main tool in pattern 4. Stop region growing when no more pixels
recognition, image segmentation, and scene met the criterion for inclusion in that
analysis.[7] Edge are local changes in the image region.[10]
intensity edge typically occur on the boundary
between two regions. The main features are Region splitting and merging: - This method
extracted from the edges of an image. Edge segment the image based on homogeneity criteria.
detection has major features for image analysis. This method works on the basis of quadtrees. It
These features are used by advanced computer considers the entire image as a single region and
vision algorithm. Edge detection is used for object then divides the image into four regions based on
detection which gives various applications like certain predefined criteria. It checks the regions for
biometrics, medical image processing etc.[8] the same defined criteria and divides it further into
four regions if the test result is negative and the
There are three different types of discontinuities in process continues till the criteria is defined. [4]
the grey level like line, point and edges. Spatial
masks can be used to detect all the three types of 3.3Thresholding:
discontinuities in an image. All the edge detection Image segmentation by thresholding is a simple but
operators are grouped under two groups are:- powerful approach for segmenting images. It is
st st useful in select foreground from background.
1. 1 order derivative: - 1 order derivatives
are Sobel operator, Canny operator, Prewit Thresholding operation convert a multilevel image
operator, Test operator. into a binary that is it chooses a proper thresholding
nd nd T, to divide image pixels into several regions and
2. 2 order derivative: - 2 order derivatives
IJSER
are Laplacian operator, Zero-crossing.[6] separate objects from background. Any pixel (x, y)
is belong to object if its intensity is greater than or
3.2Region based techniques: equal to threshold value i.e., f(x, y) ≥T, else pixel
Region based methods are based continuity. This belong to background. [11]
technique divide the entire image into sub regions Based on the selection of threshold value, there are
depending on some criterion like all the pixels in two type of thresholding method:-
one region must have the same grey level. The
simplest approach to segment image based on the 1. Global thresholding: - Global thresholding
similarities assumption is that every pixel is is used when the intensity sharing between
compared with its neighbour for similarity check the objects of foreground and background
(for grey level, texture, color, shape). are very distinct, a single value of threshold
can simply be used to separate both objects
Region based technique is relatively simple and apart. In this type of thresholding, the value
more immune to noise as compare to edge detection of threshold T depends on the property of
method. Edge based method divide an image based the pixel and the grey level value of the
on changes in intensity near edge whereas region image. Some of the common used global
based methods, divide an image into region that are thresholding methods are Otsu method,
similar according to set of predefined criteria.[9] entropy based thresholding, etc.
Region growing: - Region growing is a techniques 2. Local thresholding: -This method divides
for extracting a region of image based on predefined an image into several sub regions and then
criterion .Region growing can be prepared in four chooses different thresholds Ts for each
steps:- subregion respectively. Some common
used local threshold techniques are 2-D
IJSER © 2016
http://www.ijser.org
International Journal of Scientific & Engineering Research, Volume 7, Issue 12, December-2016 307
ISSN 2229-5518
entropy-based thresholding histogram Neural network segmentation includes two
transformation, simple statistical important steps:
thresholding.[12] (1) Feature extraction: - The input data of
3.4Clustering: neural network is determines in this step.
Some important features from images are
Clustering is an unsupervised learning task. extracted.
Clustering is defined as process of grouping object (2) Image segmentation:- The features that are
based on attributes, so that object with similar extracted from the image are segmented in
attributes lies in same cluster. Clustering is used for this step.
the purpose of pattern recognition, image
processing, data analysis, and more.Clustering Neural network have fast computing and highly
algorithm is classified as hard clustering, k-means, parallel computing ability makes it suitable for real
fuzzy clustering, etc.[13] time application. It improves segmentation results
when the data deviated from a normal situation. It is
Hard clustering: - Hard clustering assumes that a high robustness that makes it immune to noise.[12]
pixel can only belong to a single cluster. Thereexist
sharp boundaries between clusters. One of the most 3. Conclusion
popular and well used clustering algorithms is k-
means clustering algorithm. K-means clustering Digital image processing uses computer algorithm
group n pixels of an image into k number of cluster, to perform image processing on the image. The
where k
no reviews yet
Please Login to review.
