An improved algorithm based on texture feature extraction for image retrieval

With the rapid growth of image database, content-based image retrieval (CBIR) technology with its high value in theory and application has become a hot topic in the field of image processing. Based on the texture feature of database Caltech256, this paper selects two kinds of texture retrieval method - Gabor and EHD, and an improved feature retrieval algorithm combines the EHD and Gabor algorithms is put forward. Experimental results shows that the proposed algorithm performances more excellent than the last two algorithms on the conditions of illumination, tailoring and rotation.