International Journal of Computer Theory and Engineering

Abstract

Color histogram is one of the common techniques used in image retrieval systems. However, the main problem with color histogram indexing is that it does not take the color spatial distribution into consideration. In this paper, we introduce the local histograms to describe the spatial information of colors, even though, a single local histograms is not enough for efficient and robust image retrieval system. We propose the use of color local histograms on multiresolution images. The multiresolution color local histograms give much better retrieval efficiency. The multiresolution images are generated using the median filter. to measure similarity between two images using the local histograms, the traditional approaches use a metric distance. To improve performance of the CBIR system, the non metric measure (PMM) is used to measure similarity of images instead of classic metric. Experiment results prove that the CBIR using our new measure has better performance and perceptually relevant result.