An efficient image indexing algorithm in JPEG compressed domain

Considering the fact that majority of images stored in computers, such as those from the Internet, are in JPEG compressed format, it would be highly desirable to have the content-based image indexing and retrieval conducted directly in the compressed domain. In this paper, we present a partial decoding algorithm to index images directly in the compressed domain for all JPEG compressed images. Following that, an empirical study is also carried out to compare the performance of such approaches in DCT domain and that of original images in the pixel domain. This technology will prove invaluable in applications where fast image key generation is required such as on the Internet or embedded in future imaging equipment. This is necessary as not only the amount of media available on the Internet will continue to grow but also the amount stored in homes-on digital televisions, computers, cameras and media-will grow.