This paper proposes a novel image retrieval method integrating image region information and the corresponding texture features.Firstly

the method extracts the dominant color of each image block to search the color-connected regions in the image according to a new definition of correlated color.Afterwards

the texture features corresponding to each color-connected region are computed and represented as color co-occurrence matrices

for which a specific similarity measurement is designed to retrieve images based on their contents.Experimental results have shown that the proposed method has sound and robust retrieval performance by integrating color composition and color spatial information into image texture descriptors.