See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/291975278

A Review on Edge Detection Techniques for Image Segmentation

Article · August 2014

CITATIONS READS

47 9,780

1 author:

Shubhashree Savant
Maharashtra Institute of Technology
6 PUBLICATIONS 57 CITATIONS

SEE PROFILE

All content following this page was uploaded by Shubhashree Savant on 27 January 2016.

The user has requested enhancement of the downloaded file.

 Shubhashree Savant / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 5 (4) , 2014, 5898-5900

A Review on Edge Detection Techniques for

Image Segmentation
Shubhashree Savant
Assistant Professor, MCA Department, MIT(E),
Dr. Babasaheb Ambedkar Marathwada University, Aurangabad
Maharashtra, India

Abstract— Image segmentation is an important step of the application have been isolated.
digital image processing. Segmentation is a process of Segmentation of nontrivial images is one of the most
subdividing an image into the constituent parts or objects in difficult tasks in image processing. Segmentation accuracy
the image. Segmentation is either discontinuity based or region determines the eventual success or failure of computerized
based. Edge detection is useful for discontinuity based image
segmentation technique. This paper presents a review on
analysis procedures.
different edge detection techniques for image segmentation. Image segmentation algorithms generally are based on one
of two basic properties of intensity values:
Keywords— Image segmentation, Edge detection, Gradient,
1. Discontinuity
Laplacian, Canny 2. Similarity
In the first category, the approach is to partition an image
based on abrupt changes in intensity, such as edges in an
I. INTRODUCTION
image.
The edges of image have characterized the boundaries and
The principal approaches in the second category are based
regions of the image. Edges in images are areas with strong
on partitioning an image into regions that are similar
intensity contrasts – a jump in intensity from one pixel to
according to a set of predefined criteria.
the next. Edge detection is one of the most important tasks
Detection of Discontinuities
in image processing and image analysis.[1] Edge detection
There are three basic types of discontinuities in an digital
process significantly reduces the amount of data and filter
image:
out the useless information while preserving important
o Points
structural properties in an image. According to review work
o Lines
there are many ways to perform edge detection.
o Edges.
In general the edge detection techniques are grouped into
two categories:
IV. EDGE DETECTION
o Gradient
Edge detection is the most common approach for detecting
o Laplacian
meaningful discontinuities in gray level. Different
approaches are given for implementing first-and second-
order digital derivatives for the detection of edges in an
II. LITERATURE SURVEY
The edge detection process serves to simplify the analysis image. Various types of edges are shown in Fig.1
of images by drastically reducing the amount of data to be
processed while at the same time preserving useful
structural information about object boundaries.[2] The
filters are used in the process of identifying the image by
locating the sharp edges which are discontinuous. These
discontinuities bring changes in pixels intensities which
define the boundaries of the object.[3] The edge detection Fig1. Types of edges (a) Step edge (b) Ramp edge (c) Line
aims to identify points in a digital image at which the image edge (d) Roof edge
brightness changes sharply or abruptly. Image edge
detection mainly deals with extracting edges in an image by Edges are represented in images as sudden disparities. An
identifying pixels where the intensity variation is very ideal edge would be a unit step from one pixel to the next.
high.[4] Edges are used to measure the size of objects in an However, many factors must be taken into consideration:
image; to isolate particular objects from their background; o Noise
to recognize or classify objects.[5] o Brightness of the image as a whole
o Corners
III. IMAGE SEGMENTATION o Edge-like features
Segmentation subdivides an image into its constituent
regions or objects. The level to which the subdivision is A problem of fundamental importance in image analysis is
carried depends on the problem being solved. That is, edge detection. Edges characterize object boundaries and
segmentation should stop when the objects of interest in an are therefore useful for segmentation, registration, and

www.ijcsit.com 5898
 Shubhashree Savant / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 5 (4) , 2014, 5898-5900

identification of objects. A typical application where edge by using a gradient operator. Such operators, also called
detection techniques are involved is automatic character masks, provide finite-difference approximations of the
recognition. orthogonal gradient vector fx and fy.

Sobel Edge Detector

The Sobel edge detector uses the masks shown in Fig. 2 to
approximate digitally the first derivatives Gx and Gy and
finds edges using the Sobel approximation to the
derivatives.

Gx Gy

Original Image Edge Detection

The three fundamental steps of edge detection[6]

o Image Smoothing: suppress as much noise as
possible, without destroying the true edges. Fig. 2
o Detection of edge points: determine which edge The general calling syntax for the Sobel detector is
pixels should be discarded as noise and which
should be retained (usually, thresholding provides [g, t] = edge(f, ’sobel’, T, dir)
the criterion used for detection).
o Edge localization: determine the exact location of where f is the input image, T is a specified threshold, and
an edge (sub-pixel resolution might be required for dir specifies the preferred direction of the edges detected.
some applications, that is, estimate the location of ‘horizontal’, ‘vertical’ or ‘both’ (the default)
an edge to better than the spacing between pixels).
Edge thinning and linking are usually required in Robert Edge Detector
this step. The Roberts edge detector uses the masks in Fig. 3 to
approximate digitally the first derivatives Gx and Gy.
V. EDGE DETECTION TECHNIQUES
Edge detection techniques are grouped into two categories:
o Gradient Gx Gy
o Laplacian

First-order derivatives in an image are computed using the

gradient. Second-order derivatives are obtained using the
Laplacian Fig. 3
Its general calling syntax is
Edge Detection using function edge
[g, t] = edge(f, ‘Roberts’, T, dir)
The basic idea behind edge detection is to find places in an
The parameters of this function are identical to the Sobel
image where the intensity changes rapidly, using one of two
parameters. The Roberts detector is one of the oldest edge
general criteria:[7]
detectors in digital image processing
1. Find places where the first derivative of the intensity is
greater in magnitude than a specified threshold.
2. Find places where the second derivative of the intensity Prewitt Edge Detector
has a zero crossing. The Prewitt edge detector uses the masks in Fig. 4 to
The general syntax for edge function is: approximate digitally the first derivatives Gx and Gy.

[g, t] = edge(f, ’method’ ,parameters) Gx Gy

where f is the input image, method is one of edge detection

technique and parameters are additional parameters.

Gradient
Edges in an image are pixel locations with abrupt changes
in gray levels. In a continuous image (i.e., assuming Fig. 4
continuous values), the derivative of the image f(x,y)
assumes a local maximum in the direction of the edge. Its general calling syntax is:
Therefore, one edge detection technique is to measure the
[g, t] = edge(f, ‘Prewitt’, T, dir)
gradient of f in a particular location. This is accomplished

www.ijcsit.com 5899
 Shubhashree Savant / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 5 (4) , 2014, 5898-5900

The parameters of this function are identical to the Sobel Canny Edge Detector
parameters. The Prewitt detector is slightly simpler to Canny operator is based on three criteria.[2]The basic idea
implement computationally than the Sobel detector, but it uses a Gaussian function to smooth image firstly. Then the
tends to produce somewhat noisier results maximum value of first derivative also corresponds to the
minimum of the first derivative. In other words, both points
Laplacian with dramatic change of gray-scale (strong edge) and points
The gradients work best when the gray level transition is with slight change of gray-scale (weak edges) corresponds
quite abrupt. For smoother transitions, it is more to the second derivative zero crossing point. Thus these two
advantageous to also compute the second-order derivative thresholds are used to detect strong edges and weak edges.
and to see when this second derivative crosses zero. If the
second derivative crosses zero, it means that the location VI. CONCLUSIONS
does indeed correspond to a maximum and therefore this This paper presents a review on various edge detection
pixel location is an edge location. This technique is called techniques like gradient-based and laplacian-based
the zero-crossing edge location technique. A common mask techniques. Edge detection is a significant task for image
operator for the estimation of the second derivative is the segmentation used for object detection and many other
Laplacian operator applications. First-order derivatives in an image are
computed using the gradient. Second-order derivatives are
Laplacian of Gaussian (LoG) obtained using the Laplacian Result of edge detection
The Laplacian is a 2-D isotropic measure of the 2nd spatial techniques vary from set of images.
derivative of an image. The image highlights regions of
rapid intensity change and is therefore often used for edge
detection. The Laplacian is often applied to an image that REFERENCES
has first been smoothed with something approximating a [1] A. Koschan and M. Abidi. “Detection and Classification of Edges
in Color Images”. IEEE Signal Processing, 22(1):64–73, January
Gaussian Smoothing filter in order to reduce its sensitivity 2005
to noise. The operator normally takes a single gray level [2] J. Canny, “A Computational Approach to Edge Detection,” IEEE
image as input and produces another gray level image as Trans. Pattern Analysis and Machine Intelligence, vol. 8, pp. 679-
output. [7] The general calling syntax for the LoG detector 698, 1986
[3] G.T. Shrivakshan, Dr.C. Chandrasekar, “A Comparison of various
is Edge Detection Techniques used in Image Processing” IJCSI
[g, t] = edge(f, ’log’, T, sigma) International Journal of Computer Science Issues, Vol. 9, Issue 5,
No 1, September 2012
where sigma is the standard deviation. [4] Er. Komal Sharma, Er. Navneet Kaur “Comparative Analysis of
Various Edge Detection Techniques” IJARCSSE Volume 3, Issue
12, December 2013
Zero Crossing [5] R.Jayakumar, B.Suresh “A REVIEW ON EDGE DETECTION
It uses second derivative and it includes Laplacian operator. METHODS AND TECHNIQUES” IJARCCE vol. 3, Issue 4, April
This is sensitive to noise. Haralick proposed the use of zero 2014
[6] Gonzalez, R.C., and Woods, R. E. “Digital Image Processing” . 3rd
crossing of the second directional derivative of the image ed., Pearson Education (Singapore) [1992]
intensity function. The general calling syntax is: [7] 34.Maar, D., Hildreth E., “Theory of edge detection” , Proceedings
Royal Soc. London, vol. 207, 187-217, 1980
[g, t] = edge(f, ’zerocross’, T, H) [8] S.Lakshmi, Dr. V.Sankaranarayanan “A study of Edge Detection
Techniques for Segmentation Computing Approaches” IJCA
Special issue on CASCT, 2010
where H is the filter function.

www.ijcsit.com 5900

View publication stats