Ordering predictive coding of digital image

An efficient coding method for multi-level digital image data with up to about 16 levels is described, which can attain data compression nearly equal to the theoretical bound given by a two-dimensional Markov model entropy, without any loss of picture information. The method, ordering predictive coding, is composed of an ordering prediction followed by a run-length coding of a serial prediction error bit stream. It is proved that the ordering prediction is the optimized prediction scheme and has a lower prediction error entropy than that of DPCM and bit plane, and also proved that the run-length coding reduces the entropy almost near to that of the two-dimensional Markov model entropy.