A lossless image compression system using a binary arithmetic coder

In this paper we incorporate a binary arithmetic coder into a predictive lossless image compression scheme. Its idea arises from the observation that a binary arithmetic coder needs much less computation than a multi-symbol one and therefore should run faster. To map the prediction errors to binary symbols which are fed into the arithmetic coder, Rice code has been adopted because of the Laplacian distribution for prediction errors. Experiments show that the compression performance of this scheme is better than that of LOCO and S+P and is close to that of CALIC.