An examination of color image clustering by using fractal signatures

Content-based image retrieval (CBIR) is more and more important in science and technology because there are huge numbers of visual information. Choosing proper images that we need is still a difficult task. This paper focuses the problem whether or not the fractal dimension is helpful for clustering images and also CBIR. We propose a method that applies fractal dimension to cluster images for the purpose of CBIR. The results suggest that our algorithm not only reduces the search time of the system but also raises the accuracy of the searching results significantly. It suggests that using fractal dimension for image clustering can improve the efficiency in CBIR.