© 2018, IRJET | Impact Factor value: 7.211 | ISO 9001:2008 Certified Journal | Page 2486
An Approach of K-means Clustering Based SVM ensemble for Image
Retrieval
Shalini Soni
1
, Punit Kumar Johari
2
1,2
Department of CSE/IT, MITS Gwalior, MP, India
-------------------------------------------------------------------------***----------------------------------------------------------------------
Abstract - Content Based Image Retrieval is important for
retrieving the visual relevant image from the huge database.
Color, Texture, Shape information has been primitive image
descriptors in CBIR. In Content-Based Image Retrieval
System, the visual feature (Color, Texture, and Shape) are
represented at a low level. In this paper, we have extracted
the feature of technique with using K-means clustering and
use the SVM classifier.
Key Words: Content Based Image Retrieval, Feature
Extraction Technique, Color, Texture, and Shape, K-means
clustering, SVM.
1. INTRODUCTION
With the improvement of digital image and videos, CBIR
has become an important research area to search and
retrieval useful information. Content-Based Image
Retrieval is the process of the retrieving the image of the
huge database and using the extraction feature methods
[1]. CBIR includes the digital image, video, audio, graphics
and the text data. We have many ways to retrieve an
image. Feature Extraction is one function of CBIR its mean
mapping the image pixels into the feature spaces. Image
retrieval is a technique which is concerned with searching
& browsing digital image from the huge database. Image
database uses in many fields like as medicine, biometric
security, satellite image processing, military, and security
purpose.
2. CONTENT BASED IMAGE RETRIEVAL (CBIR)
The term Content-Based Image Retrieval (CBIR) was
originated in 1992. It was used by T. Kato to describe
experiments into automatic retrieval of an image from a
database, based on Color & Shape present [2]. Content-
Based Image Retrieval is an application of the computer
vision and it is used to retrieve the image in the huge
database. “Content-baseddz means searching of the contents
of an image rather than the metadata like as keywords, tag,
or characterization related with the image. Content-Based
Image Retrieval represent in the form of figure 1. Content-
Based Image Retrieval is defining the two type categorized.
(a) Text Query:
Image is defining the form of the Text such as keywords
and caption. Text features effectively as a query if suitable
text descriptions are given for descriptions in an image
database.
(b) Pictorial query:
An example of an adopted image is used as a query. To
restore related image like an image feature such as color, a
texture is used [3].
Fig -1: Architecture of Content Based Image Retrieval
3. LITERATURE REVIEW
In 2015, Suresh MB, Dr. Mohan Kumar Naik; Proposed
color space feature texture method for image retrievals,
such as RGB HSV and YCbCr. Texture feature is extracted
by applying GLCM [4].
In 2017, Miss Dhanshree S. Kalel, Miss Pooja M. Pisal, Mr.
Ramda P. Bagawade; Proposed the three visual feature
Color , Texture, Shape to retrieve query image from large
database color shape and texture extracted feature for
given query image and measure the similarity from
database image and retrieve the similar image as a query
image extraction features [5].
In 2016, Ms. V. Ausha, Ms. V. Usha Reddy, Dr. T. Ramashri;
Proposed an algorithm to retrieve the color image from the
Query
Image
Feature
Extraction
Similarity
Measure
Retrieve-
Image
Image
Database
Feature
Database
International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 07 | July 2018 www.irjet.net p-ISSN: 2395-0072