Image Retrieval Based on Composite of Mean Shift and Assignment Model

A new image retrieval method, which is based on composite of mean shift and assignment model and referred to CMSAM, is proposed. it matches each significant cluster of target image with one of retrieval image one by one and uses assignment model to seek a optimum cluster-pairs set between the two images. Based on the optimum cluster-pairs set, a similarity measure is defined. In comparison with most widely used methods based on color histogram, experimental results show that both recall and precision with CMSAM are improved significantly, in particular it can effectively retrieve such two images that their structures are similar but their colors are not identical, however, those methods canpsilat achieve such an retrieval result.