A comparison of two methods for retrieval of medical images in the compressed domain

Retrieval of images based on features derived directly from the images provides a useful way of accessing medical imagery. While most of these techniques operate in the pixel domain, a more efficient approach is to perform retrieval directly in the compressed domain of images. In this paper we look at two different methods for such compressed-domain retrieval of medical images. Both algorithms are based on wavelet image compression such as the one used in the JPEG2000 standard. We compare the effectiveness of the two techniques based on a database of 400 medical infrared images.