An image retrieval algorithm based on dynamic weights and features combination

Because single feature lacks in the universality in Content-based image retrieval, this paper presented a dynamic weights adjustment strategy, which combined features in image retrieval. First, utilizing the spatial distribution of color of local color histogram and scale-invariant of SIFT, feature vector contained more information. After that, utilizing the users´ relevant retrieval, a new formula of calculating the weights of features was presented, which doesn´t increase the quantity of computation and takes into account the preference of uses synchronously. Experimental results show that the method improves the results of retrieval comparison with the method of using single feature and permanent weights.