Shape Extraction Using Edge Detection Techniques

Divya Joshi Maulika S. Patel Nagma Ami Jyoti Gohel

G.H. Patel College of Department of G.H. Patel College of G.H. Patel College of
Engineering &Technology, Student Computer Engineering &Technology, Engineering &Technology,
Vallabh Vidhyanagar Engineering, Vallabh Vidhyanagar Vallabh Vidhyanagar
[email protected] G.H. Patel College of [email protected] [email protected]
Engineering &Technology,
Vallabh Vidhyanagar
[email protected]

ABSTRACT 1. INTRODUCTION
Content Based Image Retrieval (CBIR) is a system that searches
for similar images from a large image database according to users’
specification. Digital image database is growing rapidly in size as Content based image retrieval (CBIR) is also known as query by
a result of technological advancements. Traditional systems like image content (QBIC) and content-based visual information
text based image retrieval faces too many problems like human retrieval (CBVIR). It has been an active and advancing research
perception, deeper needs, image annotation and etc. Thus it is area since the 1990s [1].
difficult to support a variety of task-dependent queries for
traditional systems.
The Internet is one of the best places to find different types of data
CBIR systems comprehend a wide areas, viz. image segmentation, such as images, text documents etc. Digital image is growing
image feature extraction, representation, feature matching, storage rapidly in many fields like Online photo sharing of images, a basic
and indexing, image retrieval - making CBIR system development digital imaging education, medical imaging, in criminal
a challenging task. The visual content of an image is analyzed investigation.
features (i.e., color, shape, texture) extracted from the image.
CBIR system is used in Medical Imagery Retrieval, Finger Print
Retrieval, Satellite Imagery Retrieval, Internet, and Photo Due to variance of natural language, it is very difficult to get
Collections. relevant information. For example, if the user is looking for sunset
The paper focuses upon shape based image retrieval. With the images, he may get different types of images. Only a few of them
help of three different edge detection algorithms, they extract the satisfy the users’ specification because the existing software
shape features from a given input image. retrieve images on the basis of string match and many of these are
completely irrelevant to the user. Another main issue of traditional
Keywords system is human opinion. In certain cases, an image cannot be
Shape feature, Edge detection algorithms, Canny edge detection, describe properly in words. To handle such types of ambiguities,
Sobel operator, Prewitt operator. we develop feedback based techniques for customized content-
based image retrieval [1], [5].

A CBIR system automatically extract visual attributes like color,

shape, texture and any other spatial information of each and all
images in the database. CBIR system can be used in two ways.
Permission to make digital or hard copies of all or part of this First, exact image matching, i.e. matching two images and
work for personal or classroom use is granted without fee provided Second, approximate image matching, i.e. finding most closely
that copies are not made or distributed for profit or commercial
matched images to a query image [2], [3].
advantage and that copies bear this notice and the full citation on
the first page. Copyrights for components of this work owned by
others than ACM must be honored. Abstracting with credit is
permitted. To copy otherwise, or republish, to post on servers or to
redistribute to lists, requires prior specific permission and/or a fee.
Request permissions from [email protected].

ICTCS '14, November 14 - 16 2014, Udaipur, Rajasthan, India,

Copyright 2014 ACM 978-1-4503-3216-3/14/11…$15.00,
http://dx.doi.org/10.1145/2677855.2677927
 2.3 Shape
Another major image feature is the shape of the thing. It may be
defined as the typical surface pattern of an object. Shape
representations can be divided into two categories: Boundary-
based, and Region-based. The shape of an object is a binary image
representing the extent of objects. Boundary-based representation
only allows outer edge of the contour. However region-based
representation allows the entire shape region by describing the
considered region using its internal characteristics [2], [4], [5],
[8], [10], [12], [13].

3. METHODS AND DATA

In this paper, we proposed a shape based feature extraction [4],
[5]. The following three edge detection techniques are being
discussed in this paper:

3.1 Sobel Operator

Figure 1: General flow of CBIR This operator consists of a pair of 3×3 convolution kernels as
shown in Figure 2 and 3. One kernel is simply the other rotated by
90°.

2. FEARURE EXTRACTION
The first step in the CBIR process is extracting image features to a
distinguishable extent. Feature extraction is the important part of
the CBIR system. Color, Texture, Shape are features of an images
[6], [7],[ 9].

2.1 Color Feature Figure 2: Horizontal matrix Figure 3: Vertical matrix

We use color every day to tell the distinction between objects,
places, and the time of day. One of the most considerable feature
of image that make possible the recognition of images by humans
is color. Color is a property that depends on the reflection of light
These kernels are designed to respond maximally to edges running
to the eye and the processing of that information in the brain.
vertically and horizontally relative to the pixel grid. The kernels
Images characterized by color features have many advantages, like
can be applied separately to the input image, to produce separate
efficiency, strength, simplicity, low Storage Requirements [15].
measurements of the gradient component in each orientation (call
these Gx and Gy) [3], [16]. The gradient magnitude is given by:

2.2 Texture Feature

Texture is a core component of human visual observation. Texture |G|=√ +
recognition is easy, however it is complex to define. Unlike color,
texture occurs over a region rather than on a point. It is normally
Typically, an approximate magnitude is computed using:
defined purely by grey levels. We can say that texture is that
innate property of all surfaces that describes visual patterns,
having properties of homogeneity. It contains important |G| = |Gx| + |Gy|
information related to the structural arrangement of the surface
[2], [14].
Which is much faster to compute.

The angle of orientation of the edge (relative to the pixel grid)

giving rise to the spatial gradient is given by:

θ = arc tan(Gy/Gx)
3.2 Prewitt’s Operator Step 5:-

Prewitt operator is similar to the Sobel operator and is used for Hysteresis thresholding. Finally, hysteresis is used as a means of
detecting vertical and horizontal edges in images [16]. eliminating streaking. Streaking is the breaking up of an edge
contour caused by the operator output fluctuating above and
below the threshold.

4. RESULTS

In table 1, given the input image, the output of the various

operators is given.

Figure 4: Horizontal matrix Figure 5: Vertical matrix

5. CONCLUSION

3.3 Canny Operator There are various algorithms available for edge detection and best
The Canny Edge Detector is one of the most commonly used one has to be used according to requirements.
image processing tools [9], [11]. It is a multi-step process. Steps
are:
In this paper, we studied and implemented three different
techniques of edge detection. Compared with two techniques:
Step 1 & 2:- Sobel and Prewitt operators, Canny operator can detect edges of
an image better in noisy condition.

Convert to grayscale and reduction of noise. We can use any kind

of noise reduction filter.

Step 3:- 6. REFERENCES

1. Y. Rui, T. S. Huang, and S.-F. Chang, “Image retrieval:
Current techniques, promising directions, and open
Compute gradient magnitude and angle. A 3x3 convolution issues”, 2002
matrices are shown below: 2. S. Jain, A. Joglekar, and D. Bote, ISearch: A Content-
based Image Retrieval (CBIR) Engine, as Bachelor of
Computer Engineering final year thesis, Pune
University, 2002
3. Dr. Fuhui Long, Dr. Hongjiang Zhang and Prof. David
Dagan Feng. Content Based Image Retrieval, 2004

4. Raman Maini & Dr.Himanshu Aggarwal, “Study and

Comparison of Various Image Edge Detection
Techniques” , 2006
Figure 6: Horizontal matrix Figure 7: Vertical matrix 5. Ritika Hirwane, “Fundamental of Content Based Image
Retrieval” , 2012
6. Swati V. Sakhare & Vrushali G. Nasre, "Design
The magnitude, or edge strength, of the gradient is then of Feature Extraction in Content Based Image
approximated using the formula:
Retrieval (CBIR) using Color and Texture", 2011

|G| = |Gx| + |Gy| 7. Parulpreet singh & Kulvinder Singh Mann, “ An

Approach of Image Retrieval Using Content Based
Retrieval System”, 2013
Step 4:-
8. Hankyu Moon, Rama Chellappa and Azriel
Rosenfeld, " Optimal Edge-Based Shape
Non-maximum suppression. From the step 3, we can find that, Detection", 2002
using Sobel or Prewitt operator, edges it finds can be either very
thick or very thin depending on intensity. 9. P.S.Suhasini , Dr. K.Sri Rama Krishna and Dr. I. V.
Murali Krishna, “CBIR Using Color Histogram
Processing”, 2005-2009
10. Reshma Chaudhari & A. M. Patil, “Content Based 14. P. V. N. Reddy & K. Satya Prasad, ” Color and Texture
Image Retrieval Using Color and Shape Feature”, 2012 Features for Content Based Image Retrieval”, 2011

11. Mohamed Ali & David Clausi, “Using the Canny Edge 15. K Arthi & J. Vijayaraghavan, “Content Based Image
Detector for Feature Extraction and Enhancement of Retrieval Algorithm Using Colour Models”, 2013
Remote Sensing Images”, 2001
12. S.Arivazhagan, L.Ganesan & S.Selvanidhyananthan, “
16. Rafael C. Gonzalez, and Richard E. Woods, “Digital
Image Retrieval using Shape Feature”, 2007
image processing”, Prentice Hall, Upper Saddle River,
13. Hwei-Jen Lin, Yang-Ta Kao, Shwu-Huey Yen, and
N.J., (2008)
Chia-Jen Wang, “A Study of shape-Based Image
Retrieval”, 2004
Input image Canny image Sobel image Prewitt image

Table 1: Results of Edge Detection