A novel image retrieval scheme in JPEG2000 compressed domain based on tree distance

Abstract

As the new generation standard for static image compression, JPEG2000 employs the wavelet transform instead of DCT in JPEG. JPEG2000 is thus advantageous over JPEG in terms of compression ratio, freedom of compression mode selection and adaptation to different kinds of images. Obviously, there is a tendency for JPEG2000 to take the place of JPEG, and large amount of images are saved in JPEG2000 format in image databases and on the Internet in the near future. So, research on image retrieval realized in JPEG2000 compressed domain has important real significance. In this paper, the data structure in JPEG2000 compressed domain is firstly analyzed. The concept of information tree and a novel tree-distance measurement are then proposed. Finally, we propose a new image retrieval method in JPEG2000 compressed domain. Experimental results show its potential in compressed domain image retrieval techniques.