International Journal of Applied Engineering Research ISSN 0973-4562 Volume 13, Number 5 (2018) pp. 2432-2442
© Research India Publications. http://www.ripublication.com
2432
A Fuzzy Logic Based Soft Computing Approach in CBIR System Using
Incremental Filtering Feature Selection to Identify Patterns
1
V. Yadaiah and
2
Dr. R .Vivekanandam, .Rajaram Jatothu
3
1,3
Research Scholar, Department of Computer Science and Engineering,
Sri Satya Sai University of Technology and Medical Sciences, Sehore, M.P., India.
2
Professor, Director in Muthayammal Eng. College, Rasipuram, Namakkal,T.N., India
Abstract
Content based Image Retrieval (CBIR) may be a set of
techniques for retrieving semantically-relevant pictures from
an image database supported automatically-derived image
options. Generally, in CBIR systems, the visual features are
described at low-level. They are simply rigid mathematical
measures that cannot influence the inherent subjectivity and
fogginess of individual’s understandings and perceptions. As a
result, there is a niche between low-level features and high-
level semantics. We have a tendency to are witnessing the era
of massive information computing where computing the
resources is turning into the most bottleneck to handle those
massive datasets. With in the case of high dimensional data
where every view of information is of high spatiality, feature
selection is important for additional rising the clustering and
classification results.
In this paper, we have a tendency to propose a new feature
selection method is Incremental Filtering Feature Selection
(IFFS) algorithm that employs the Fuzzy Rough Set for
choosing best subset of features and for effective grouping of
huge volumes of data, respectively. We introduce a new system
of visual features extraction and matching by using Fuzzy
Logic (FL). FL is a powerful tool that deals with reasoning
algorithms used to emulate human thinking and decision
making in machines. An in depth experimental comparison of
the proposed method and other methods are done. The
performance of the proposed model yields promising results on
the feature selection, and retrieval accuracy in the field of
Content based Image Retrieval.
Keywords: Content Based Image Retrieval, Fuzzy Logic,
Fuzzy Color, and Incremental Filtering Feature Selection.
INTRODUCTION
Very massive collections of images are growing quickly
because of arrival of cheaper storage devices and also the
internet. Finding an image from a huge set of images is very
challenging task. One solution to this problem is to label images
manually. But it is too expensive, time consuming and not
feasible for several applications. Moreover, the labeling
process depends on the semantic accuracy in describing the
image. Therefore, many content based image retrieval systems
are developed to extract low levels features for describing the
image content [1].
A typical content-based retrieval system (as in Fig.1) is split
into 2 stages: off-line feature extraction and on-line image
retrieval [2]. In off-line stage, the system mechanically extracts
visual attributes of every image in the database based on its
pixel values and stores them in a different database inside the
system, known as a feature database. In on-line stage, the user
will submit a query example to the retrieval system. The system
represents this example with a feature vector. The distances
(i.e., similarities) between the feature vectors of the query
example and those of the media in the feature database are then
computed and ranked. The system ranks the search results and
then returns the results which are most similar to the query
examples.
Image data is vague in nature and in content-based retrieval this
property creates some issues like [3]:
1. Descriptions of image contents typically involve
inexact and subjective concepts.
2. Typically imprecision and vagueness exist in
descriptions of the images and in some of the visual
features.
3. User’s needs to image retrieval could also be naturally
fuzzy.
Fuzzy Logic (FL) is used in CBIR system because it is the
character of of image data, and also the nature of human
perception and thinking process. So, it can minimize semantic
gap between high level semantic and low level image features.
Also, it is robust to the noise and intensity modification in the
images. Finally, the users are interested in results according to
similarity (closeness) instead of equality (exactness).
In [4], a color histogram representation, called Fuzzy Color
Histogram (FCH), is presented by considering the color
similarity of each pixel’s color associated to all the histogram
bins through fuzzy-set membership function. An approach for
computing the membership values based on fuzzy-means
algorithm is developed. The proposed FCH is further exploited
in the application of image indexing and retrieval.
Konstantinidis et al [5] proposed a fuzzy linking system for
color histogram creation in L*a*b* color space. It contains 10
bins, and 27 rules used to derive the final histogram. Kucuktunc
et al [6] proposed a fuzzy linking system for color histogram
creation in L*a*b* color space. Their system contains 15 bins,
and 27 rules used to derive the final histogram.