Research On Color Feature Based Image Retrieval

After introduced the bottom visual characteristic of content based image retrieval — color,the shape,the texture characteristic's extraction method comprehensively,the measure method of similarity and the index of retrieval algorithm, the paper studied the color histogram,and the circular color histogram which contained the image space information emphatically.But this method still had the color space which used does not conform to the person's visual perception,caused the characteristic dimension for the guarantee quantification precision to be high, various colors channel could not effectively unify together and soon. In view of the above question, the improvement algorithm is: to adopt HSV color space which has the visual uniform ,gives the different quantification progression to the different color channel, and H,S,V three components carries on no laddered quantification, H,S,V three components weighting summation composes the row of the circular color histogram construction.