Lossless Image Compression using Topological Pixel Re-ordering and Switched Predictor

摘要

This paper presents an improved lossless image compression of gray-scale images in terms of entropy (bits/pixel). The given image is re-ordered using various topologies based on uniformity measure, if required. The entropy of the given image/re-ordered image is calculated for the residues, i.e., the difference between the actual pixel intensity value and estimated pixel intensity value, obtained using different estimation procedures. The estimation of the pixel intensity is based on the uniformity measure, calculated using pixel intensities in the given image. In this work, the method of estimating pixel intensity is presented. Superior performance is observed in this approach over three standard techniques available in the literature.