THE EFFECT OF DISTORTIONS AND FRAGMENTATION ON RESULTS OF CONTENT-BASED RETRIEVAL OF COLOR IMAGES

Abstract

This paper is dealing with the problem of content-based image retrieval using color co-occurrence matrices. It is well-known that there are a number of factors influencing the results of image retrieval. The specific contribution of this paper lies on quantitative assessment of the effect of spatial image distortions as well as on estimating the accuracy of retrieving original images using its fragments of different size as query examples. The presented study is capitalized on database containing 10 000 color RGB images of very different content. The necessary experimental results are provided.