A New Near-Lossless Image Compression Algorithm

This paper proposes a new near-lossless image compression algorithm, namely empirical data decomposition (EDD), which can carry on adaptive analysis to the observation data. The algorithm suits to analyze non-stationary data and can effectively remove the relevance between of observation data. In EDD, analysis filter is automatically determined by observation data, and is able to realize the multi-resolution analysis. Performing compression on continuous tone image and hyper-spectrum images with the improved EBCOT based on EDD and JPEG2000, respectively, the simulation results indicate that, in the case of high or higher code rate, EDD can obtain higher image compression efficiency.