Ijettcs 2014 10 25 105
Ijettcs 2014 10 25 105
Ijettcs 2014 10 25 105
ISSN 2278-6856
Abstract
There is a tremendous use of internet worldwide, users
probably deals with multimedia data. There are in all three
types of search engines such as text retrieval system, image
retrieval system and multimedia retrieval system. Image
retrieval system are very much in demand in todays era. The
existing image retrieval systems available on web are google
image retrieval, yahoo image retrieval, Altavista image
retrieval etc. Users are not satisfied with the old image
retrieval systems like the one with relevance feedback. Users
today prefers systems which has good speed, gives security and
most importantly users dont want to get involved in the
processing, they just want an accurate results on just one click.
In this paper we are going to focus on different methods which
are used by image retrieval systems so far.
1. INTRODUCTION
There are many image retrieval systems available which
follow different techniques. Traditionally image retrieval
was based on captions provided with the image. But using
captions of image in image retrieval was not appropriate
because anyone can alter the caption, and also the fact that
caption cannot cover all the information related to image.
Then came the image retrieval system which focuses on
the relevance feedback. In relevance feedback, every time
a result is displayed we have to give a feedback manually
that, out of the result displayed which are more matching
to our input image. In this way, this feedback is taken as a
new input and again enhanced result is displayed. But
users soon found out that giving such feedback every time
is very tedious and time consuming job. After this came
CBIR, which proved a boon in area of image retrieval.
CBIR system deals with the actual content of images rather
that metadata such as caption, tags, keywords etc. The
word content in CBIR refers to features of image like
color, shape, texture, edges etc., which can be extracted
from that image itself. Thus a system which uses features
2. TECHNIQUES
FOLLOWED
RETRIEVAL SYSTEMS
BY
IMAGE
Figure 3. CBIR.
Humans manually entering keywords for images in a large
database can be inefficient, expensive and may not capture
every keyword that describes the image. Thus a system that
can filter images based on their content would provide
better indexing and return more accurate results. Figure.3
shows working of CBIR. A CBIR consists of two main
parts as follows
1) Offline Part The offline part deals with building the
database for images. All the images are analyzed, their
features are extracted and according to their features
images are clustered and indexed.
2) Online Part In Online part, user gives the image as an
input. The image features are extracted and those
extracted features are matched with the features
present in database. If the features get matched then,
image corresponding to that feature is similar to the
input image. In this way matching images are
retrieved and they are displayed on output screen in
descending order.
ISSN 2278-6856
Page 247
B. NTSC formula
NTSC formula is written as,
(7)
NTSC formula takes 56% of green, 11% of blue, since
green color has more luminescence or brightness than
blue. After converting an image to grey scale we have to
perform edge detection on it. To perform edge detection
there are certain algorithms [5] available such as:
1. Canny Algorithm
2. Sobel Algorithm
Then moment is calculated as the moment is the feature
for shape content.
c) CBIR using Texture Feature Extraction
Texture is a repetitive tone in an image. Texture describes
visual patterns. It contains vital information about the
structural arrangement of surface such as clouds, leaves,
bricks, fabrics etc [5]. Texture feature can combined with
other features for increasing the effectiveness of result. For
calculating texture many methods are available such as
Texture based differenciation can be done by two different
ways :
1) Structural Method
Structural methods include morphological operator and
adjacency graph. They describes texture by identifying
structural primitives and their placement rules. They tend
to be most effective when applied to textures that are very
regular.
3) Statistical Method
Statistical methods include wavelet transform, co- currence
matrices,
shift-invariant
principal
component
analysis(SPCA), Tamura features, Wold decomposition,
Markov random field, fractal model. The co-occurrence
matrix, counts co-occurrence of pixels with grey values i
and j at a given distance d. The distance d is defined in
polar coordinate (d, ). The co-occurrence matrix can be
defined as follows:
(6)
ISSN 2278-6856
(8)
Where card{.} denotes the number of elements in the set
The Tamura features are coarseness, contrast, regularity,
roughness etc. Coarseness and contrast have earlier been
used in applications like photobook and QBIC. Wold
features defines texture in form of perceptual properties.
The three Wold components, harmonic, evanescent, and
indeterministic, correspond to periodicity, directionality,
and randomness of texture respectively. Periodic textures
have a strong harmonic component, highly directional
textures have a strong evanescent component, and less
structured textures tend to have a stronger indeterministic
component.
Page 248
(9)
If this value is high relatively to the total number of
interest point, we have a good similarity. If value is low we
have bad similarity.
Publication
and Author
Techniques
Conclusion
Yet
Another
Content
Based
Image
Retrieval
system
14th IEEE,
International
Conference
Information
Visualisation
Samy AitAoudia,
Ramdane
Mahiou,
Billel
Benzaid
Color - HSV
Texture
contrast,
entropy,
energy and
inverse
differential
moment
Point of
interest Harris
detector
This paper
attempts to
evaluate the
performance
of the
YACBIR
system on
sample
datasets of
images. The
system gives
good results
on the tests
conducted.
Content
Based
Image
Retrieval
Using
Color and
Shape
Features
IJAREEIE,
November
2012
Reshma
Chaudhari,
A. M. Patil
Color HSV
Shape
Mass,
Centroid and
Dispersion
In this they
have improved
the accuracy of
CBIR from
44% to 72%
Image
Retrieval
by Using
Colour,
Texture
and
Shape
Features
IJAREEIE,
April 2014
Prof. C. S.
Gode, Ms.
A. N. Ganar
Color Color
Histogram(H
SV)
Texture
Co-occurence
matrix
Shape
Canny
Algorithm,
moment.
Satisfactory
results
Sketch
Based
Image
Retrieval
System
Using
Wavelet
Transfor
m
IJIRD, April
2013
Sonal
Shinde,
Priya
Nanaware,
Godavari
Kudalikar,
Harpreetkaur
Nagi
Shape
Wavelet
Transform
Fuzzy CMeansClustering of
images
Lucene
Algorithm
Indexing
ISSN 2278-6856
Satisfactory
results
4.Conclusion
There are many CBIR systems developed across the world.
They all use features like color, shape, texture and point of
interest individually or in combination. We have focused
on some important feature aspects related to CBIR. If these
features are used individually they give results accordingly,
which means if only color feature is used, images having
same color features will be retrieved. From the above
research [3, 4 ,5] we have concluded that when we use this
features like color, shape, texture, point of interest etc., in
combination we yield better results.
References
[1] Steve Lawrence and C. Lee Giles, Giles, Text and
Image Metasearch on the Web, International
Conference on Parallel and Distributed Processing
Techniques and Applications , 1999.
[2] Dinakaran, J. Annapurna and Ch. Aswani Kumar,
Interactive Image Retrieval Using Text and Image
Content, CYBERNETICS AND INFORMATION
TECHNOLOGIES Volume 10, No 3, Sofia 2010
[3] Ms. K. Arthi and Mr. J. Vijayaraghavan, Content
Based Image Retrieval Algorithm Using Colour
Models, International Journal of Advanced Research
in Computer and Communication Engineering, Vol.
2, Issue 3, March 2013.
[4] Swati V. Sakhare and Vrushali G. Nasre, Design of
Feature Extraction in Content Based Image Retrieval
(CBIR) using Color and Texture, International
Journal of Computer Science & Informatics, VolumeI, Issue-II, 2011.
[5] Prof. C. S. Gode and Ms. A. N. Ganar, Image
Retrieval by Using Colour, Texture and Shape
Features,
International Journal of Advanced
Research
in
Electrical,
Electronics
and
Instrumentation Engineering, Vol. 3, Issue 4, April
2014.
[6] Senosy Arrish, Fadhil Noer, Ahmadu Maidorawa and
Naomie Salim, Shape-Based Plagarism Detection for
Flowchrt Figures in Texts, International Journal of
Computer Science & Information Technology, Vol
6, No 1, February 2014.
[7] Dr. H.B.Kekre, Sudeep D. Thepade, Tanuja K. Sarode
and Vashali Suryawanshi, Image Retrieval using
Texture Features extracted from GLCM, LBG and
Page 249
ISSN 2278-6856
AUTHOR
Prajakta Mahendra Ovhal received the
Bachelors degree (B.E.) in Computer
Engineering in 2013 MIT AOE, Pune. She is
now pursuing ME degree, in Computer
Engineering at P.E.S Modern College Of
Engineering. Her current research interests include Image
Processing.
Prof. B.D. Phulpagar is currently working as Assistant
Professor in Computer Engineering Department at P. E. S.
Modern College of Engineering Pune (India). He has
completed his postgraduate studies at Govt. College of
Engineering Pune. He is Pursuing Ph. D. in Computer
Engineering at Shri Guru Gobind Singhji Institute of
Engineering and Technology, Nanded, affiliated to Swami
Ramanand Teerth Marathwada University Nanded (India).
Page 250