Comparative Study and Optimization of Feature-Extraction Techniques for Content based Image Retrieval

Abstract

The aim of a Content-Based Image Retrieval (CBIR) system, also known as Query by Image Content (QBIC), is to help users to retrieve relevant images based on their contents. CBIR technologies provide a method to find images in large databases by using unique descriptors from a trained image. The image descriptors include texture, color, intensity and shape of the object inside an image. Several feature-extraction techniques viz., Average RGB, Color Moments, Cooccurrence, Local Color Histogram, Global Color Histogram and Geometric Moment have been critically compared in this paper. However, individually these techniques result in poor performance. So, combinations of these techniques have also been evaluated and results for the most efficient combination of techniques have been presented and optimized for each class of image query. We also propose an improvement in image retrieval performance by introducing the idea of Query modification through image cropping. It enables the user to identify a region of interest and modify the initial query to refine and personalize the image retrieval results.

Feature extraction methods for batik pattern recognition: A review

“…Features resulted in a variety of research on the color feature are histograms, color moments (CM), color coherence vector and the color correlogram [5,8,9,10,11,12]. There is dominant color descriptor (DCD), color structure descriptor (CSD), and Scalable color descriptor respectively for color feature extraction [13]. Utilization of color features has advantages and disadvantages, for example, the histogram feature that makes easier the computing process due to intuitive nature but still has large dimensions without the spatial info and sensitive to noise.…”

Features Analysis In CBIR Systems

“…Semantic image retrieval [7] begins by request made by a person, e.g. "find pictures of Mahatma Gandhi".…”

A Contextual Approach for Modeling Activity Recognition

“…The test instance labels which deviated from normal activities were detected as anomalies. Feature extraction and Optimization Techniques for detecting Region Of Interest (ROI) is addressed in [7]. Human Action Recognition using Hidden Markov Model is discussed in [8].…”

“…Since we want the focus to remain on human in order to recognize the human activity, we avoid unwanted computation that would have been performed on the non-ROI region. Comparative Study of Optimization techniques in feature extracton is discussed in [7]. One of the techniques discussed, namely cropping, helps in marking desired Region Of Interest (ROI).…”