期刊名称:International Journal of Computer Science and Information Technologies
电子版ISSN:0975-9646
出版年度:2015
卷号:6
期号:3
页码:2645-2652
出版社:TechScience Publications
摘要:Because of content based techniques, image retrieval is becoming efficient. The goal of CBIR is to extract features of images and to compare them to retrieve similar images. We are focussing on features of images such as colour, shape and texture. In our proposed system, the efficiency is improved by use of different efficient feature extraction algorithms – RGB to Lab conversion for colour feature extraction, Canny edge detection algorithm for shape feature extraction and Framelet transform for texture feature extraction. We applied these feature extraction algorithms on sub blocks of images. For similarity matching of extracted features, Manhattan distance method is used. Also, to improve the speed of image retrieval, indexing and serialization techniques are used. The proposed method is compared with existing systems [1], [2] and we can see an increase in average precision from 45.06% to 64.1%.