ABSTRACT
ABSTRACT: In this paper, we propose a new image retrieval system based on representative regions. First, image is segmented to several categories using K-means and Affinity Propagation (AP) clustering methods, and the largest region of every category is found as the representative region. Then, color and texture features of the representative regions are obtained through the HSV color histogram and Local Binary Pattern (LBP). Representative Region Matching (RRM) algorithm is used for calculating the distance between the query image and database images combining color, texture features with location, and area weight. Experimental results show that the proposed method is more prominent than retrieval using some of the existing methods.
