Abstract: As the theory of quotient space granularity comput is imported in image retrieval, a novel image retrieval method using multi-granularity color features is proposed. According to different behaviors under different granularities, one can first obtain color features under different granularities, obtaining different quotient spaces. Second, the obtained quotient spaces are combined based on the quotient space granularity combination principle to achieve image retrieval. Compared with other methods using a single color feature, the image retrieval method based on multi-granularity color features utilizes the image’s own information more effectively. Experimental results demonstrate feasibility and validity of the proposed method.

Key words: multi-granularity, color feature, image retrieval