A lossless coding scheme for images using cross-point regions for modeling

Abstract

This paper presents CRIC (Cross-Point Regions for Lossless Image Compression), a scheme for losslessly encoding and decoding images, especially medical images, by optimizing on the probability of cross points that neighbor points of grey levels 2n. The base of this statement is the effect of Gray coding on cross points. At first, the effect of Gray codes is determined on an adjacent data set because images characteristically contain data that does not change much in a specific area; then this effect is generalized for real data without losing generality for their statistical properties. This is especially true for medical images that have many regions with the same grey levels. The Gray code transformation makes the bit states of cross points change from the original data bits, so first the probabilities of data bits on specific bit planes in cross point regions and then the entropies of the messages are changed. These probabilities are estimated and compared with the probabilities of the original data bits. This change of probability has important effects on the encoding and decoding processes in lossless medical image compression. ©2007 IEEE.