An Efficient Image Retrieval Based on Fusion of Fast Features and Query Image Classification

An Efficient Image Retrieval Based on Fusion of Fast Features and Query Image Classification

Vibhav Prakash Singh (Department of Computer Science & Engineering, Indian Institute of Technology (BHU), Varanasi, India), Subodh Srivastava (Department of Computer Science & Engineering, Indian Institute of Technology (BHU), Varanasi, India) and Rajeev Srivastava (Department of Computer Science & Engineering, Indian Institute of Technology (BHU), Varanasi, India)
Copyright: © 2017 |Pages: 19
DOI: 10.4018/IJRSDA.2017010102
OnDemand PDF Download:
List Price: $37.50


Content Based Image Retrieval (CBIR) is an emerging research area in computer vision, in which, we yield similar images as per the query content. For the implementation of CBIR system, feature extraction plays a vital role, where colour feature is quite remarkable. But, due to unevenly colored or achromatic surfaces, the role of texture is also important. In this paper, an efficient and fast CBIR system is proposed, which is based on a fusion of computationally light weighted colour and texture features; chromaticity moment, colour percentile, and local binary pattern (LBP). Using these features with multiclass classifier, the authors propose a supervised query image classification and retrieval model, which filters all irrelevant class images. Basically, this model categorizes and recovers the class of a query image based on its visual content, and this successful classification of image significantly enhances the performance and searching time of retrieval system. Descriptive experimental analysis on benchmark databases confirms the effectiveness of proposed retrieval framework.
Article Preview

1. Introduction

Technology and information are growing fast with time, and information database is getting large. This information may contain textual data, images and videos. In this information database, images are everywhere, from medical domain to professional conferences people used to take digital images, and save them for future applications, therefore size of image databases are increasing rapidly. Consequently, image search and retrieval tasks are getting difficult with the time. Currently, machine learning based approaches play a vital role for image classification and retrieval in different fields; especially in computer added diagnosis system where these approaches easily detect and categories the different medical disease and images (Cheriguene et al., 2015; Dey & Ashour, 2016; Kausar et al., 2016; Kriti et al., 2015; Zemmal et al., 2016).

Content based image retrieval is a wide research area in which many researcher are working meticulously from past two decades. CBIR methods use numerous features of the image, most of them are low level features, based on colour, texture, shape, spatial orientation etc. (Hiremath & Pujari, 2007; Jalab, 2011; Lin et al., 2009; Manjunath et al., 2001; Walia & Pal, 2014; Yue et al., 2011). In feature extraction methods, local patterns with spectral content have made their place because of their efficiency and simplicity (Paschos et al., 2003). Usually, humans recognize an image from its colour, texture, and shape based appearances. Therefore, it is natural to use all these features for image retrieval.

Colour is a basic feature, has strong correlations with the underlying objects in an image, and usually used in image retrieval and object recognition. Colour features are relatively easy to extract and match, and have been found to be effective for indexing and searching of colour images in image databases. Also one of the main aspects of colour feature extraction is the choice of a colour space. There are many colour features that have been solely used for image retrieval. These are Colour histogram (CH) (Swain & Ballard, 1991), Colour difference histogram (Liu et al., 2013), Colour correlogram (Huang et al., 1997), Colour moments (Ghosh et al., 2015; Stricker & Orengo, 1995), and colour coherence vector, represents colour with different coherence & incoherence bin etc (Acharya & Ray, 2005).

In addition, texture is also an important visual feature which captures neighbouring pixels and their inter-relationship with surrounding. It is a structure of surfaces, formed by repeating a particular pattern or several patterns in different relative spatial positions. Many statistical, non-statistical and spectral techniques have been proposed to describe texture features of an image, such as Edge histogram descriptor (EHD), wavelet coefficients, Gabor filter, Tamura, Local binary pattern (LBP), and Gray level co-occurrence features (GLCM) etc (Haralick et al., 1973; Han & K, 2007; Mandal et al., 1996; Ojala et al., 1996; Palm, 2004; Samanta et al., 2015; Won et al., 2002). In spite of all these colour and texture features, shape based features are also important for relevant image retrieval, because human perception is based on shape of an object and they are recognized mainly from their shapes (Amanatiadis et al., 2011).

Complete Article List

Search this Journal:
Open Access Articles: Forthcoming
Volume 5: 4 Issues (2018): 1 Released, 3 Forthcoming
Volume 4: 4 Issues (2017)
Volume 3: 4 Issues (2016)
Volume 2: 2 Issues (2015)
Volume 1: 2 Issues (2014)
View Complete Journal Contents Listing