Image Retrieval Based on Improved Entropy and Moments

Entropy and moments have been widely used in image retrieval. However, perceptually dissimilar color histograms have similar entropy and moments by traditional methods if they have similar probability distribution. In order to remove such influence, three improved methods, histogram sorting, histogram area and the combination of the above two, are presented in this paper. Experiments show that the proposed algorithms are more effective in image retrieval than the traditional methods.