Multi Directional Wavelet Filter Based Region of Interest Compression for Low Resolution Images

Abstract

An increase in the demand for storing the large archrivals’ of medical image data bases and the image data base for surveillance applications paved way for Region of Interest (ROI) based image compression techniques. Different ROI based coding techniques identify fixed shaped regions for compression. However, the real world images are having irregular shaped ROI. So, extraction of directional information along with identification of relevant information along multiple resolutions is very important for performing ROI based compression. Therefore, in this paper a Multi Directional Wavelet Based ROI Compression for low resolution images is proposed. Furthermore, the ROI is encoded using lossless encoding techniques for obtaining good resolution and the rest of the image is coded with lossy image compression techniques for obtaining high compression ratio. The proposed algorithm is compared with JPEG2000 standard which uses ROI for compression of images, with arithmetic encoder which is a lossless image compression method and also with SPIHT encoder which is a lossy image compression method.