A new lossless image compression algorithm based on arithmetic coding

Abstract

We present a new lossless image compression algorithm based on Arithmetic Coding. Our algorithm selects appropriately, for each pixel position, one of a large number of possible, dynamic, probability distributions, and encodes the current pixel prediction error by using this distribution as the model for the arithmetic encoder. We have experimentally compared our algorithm with Lossless JPEG, that is currently the lossless image compression standard, and also with FELICS and other lossless compression algorithms. Our tests show that the new algorithm outperforms Lossless JPEG and FELICS leading to a compression improvement of about 12% over Lossless JPEG and 10% over FELICS.