Color Image Retrieval Using Visual Weighted Blocks

This paper proposes a color image retrieval method based on visual weighted blocks in order to further enhance the efficiency of image retrieval. The color image histogram is such an effective method that it is used widely in various image retrieval systems. But the color image histogram has many limitations because it does not contain spatial information in its retrieval scheme, so the retrieval results are not satisfactory. In this paper, the color images are divided into blocks in order to make the color image histogram more distinguished. And human visual characteristics and local features are also utilized in this paper. Firstly, images are divided into blocks and the sub-block color histograms are calculated. Second, each pixel is arranged a saliency score which could reflect the local variation of each pixel, and then the visual weighted value for sub-blocks are also calculated. Finally, all the histograms of sub-blocks are combined into one in order to compute the similarity between color images. To prove the capability of this paper, we evaluate our framework on the standard dataset and the experimental results have shown that the proposed method is more accurate and efficient.