A study of a lossless image compression algorithm using adaptive prediction and context-based entropy coding

In this paper a context based lossless image compression algorithm is presented. It consist of an adaptive median-FIR predictor, a conditional context based error feed back process and a new error representation. The prediction error is encoded by a context-based arithmetic encoder. Experimental results show that for a set of 18 images of different kinds, the compression performance of the proposed algorithm is very close to that of CALIC and is better than LOCO and S+P. This paper also presents an algorithmic study of the proposed algorithm. The contribution of each of the building blocks to the compression performance is studied. It has been shown that these building blocks can be incorporated into further development of lossless image compression algorithms.