Improved arithmetic coding in H.264/AVC using Context-Tree Weighting and prediction by Partial Matching

In this paper, an improvement is proposed for coding efficiency of the context-based adaptive binary arithmetic coding (CABAC) in applications to the Advance Video Coding H.264/AVC. The improvement of coding efficiency is obtained by more accurate estimation of the conditional probabilities fed to the arithmetic codec. Better conditional probabilities estimation is achieved with well-known data modelling algorithms: Context-Tree Weighting (CTW) and Prediction by Partial Matching (PPM). The original method of incorporating CTW and PPM techniques into the H.264/AVC video codec is presented in this paper. The coding efficiency of the modified H264/AVC video codecs (with CTW and PPM techniques) have been tested with several test sequences. The coding efficiency of the modified H.264/AVC video codecs have been compared against the coding efficiency of the original H.264/AVC codec (with CABAC entropy coding method). Experimental results show that the proposed techniques of the conditional probabilities estimation allow up to 2% bitstream reduction in comparison to the standard CABAC algorithm.