An Improved Algorithm for Image Retrieval by Entropy
Zhou Li-hua · Infrared Technology · 2005
The dimension of a color histogram can be reduced to a number by information entropy, however, different histograms with the same probability distribution have the same entropy because of the symmetry of entropy. In order to remove the influence of symmetry on image retrieval, three methods, histogram sorting, histogram area and the combination of histogram sorting and histogram area, are presented . To the histograms with different visual sensation, though, they have the same or similar probability distribution, the location of similar element in histograms is different definitely. That is the location difference of the elements in the histograms, are our methods based on. The experiment results show that our proposed methods give good performance in terms of efficiency and retrieval effectiveness.