Lossless compression of medical images using space-varying multiresolution multiplicative autoregressive models

This paper presents a new multiresolution predictive scheme for lossless image compression that consists of four successive steps; namely, spatial varying mean calculation, multiresolution decomposition, hierarchical interpolation and predictive coding using two-dimensional multiplicative autoregressive models. This approach has two advantages, namely, entropy reduction in the low resolution, and progressive transmission. The performance of the proposed technique is compared with three alternative techniques; namely, difference pyramid (DP), reduced difference pyramid (RDP) and hierarchical interpolation (HINT) method. The experiments demonstrate that the proposed technique achieves higher compression compared to the other techniques considered.<>