ROI Based Natural Image Retrieval Using Color and Texture Feature

Abstract

Region of interest(ROI) plays an important role in image analysis. In this paper, an efficient approach for content based image retrieval combining both color and texture features using three ROIs is proposed. Firstly, segment image to three parts using K-means algorithm. Secondly, select three ROIs from the three parts and then extract color features and texture features of ROIs. The similarity of two images will be determined by the similarities between pairs of ROIs. The experimental results demonstrate that the proposed method is encouraging with a successful retrieval rate.