An Efficient Statistical Intra-mode Prediction Method for H.264

H.264 is a new codec that achieves roughly half of the bit-savings over MPEG2 and MPEG4 for a given quality. One of the features of H.264 is that allows intra frames to be predicted from previously encoded macroblocks in the same frame. This prediction is currently based on a simple algorithm that utilizes the encoded modes of the adjacent macroblocks. While this method is fast, it is often inaccurate, especially at high bitrates. In this paper a new adaptive statistical method is proposed that increases prediction accuracy by over 100% in certain situations. This improvement in accuracy results in the reduction of bits required for each intra frame

[1]  Ajay Luthra,et al.  Overview of the H.264/AVC video coding standard , 2003, IEEE Trans. Circuits Syst. Video Technol..

[2]  F. Pan,et al.  Fast Mode Decision for Intra Prediction , 2003 .

[3]  Qing Li,et al.  Fast intra-prediction model selection for H.264 codec , 2003, SPIE ITCom.

[4]  Yo-Sung Ho,et al.  A Fast Mode Decision Algorithm for H.264/AVC Intra Prediction , 2007, 2007 IEEE Workshop on Signal Processing Systems.

[5]  Gary J. Sullivan,et al.  Rate-constrained coder control and comparison of video coding standards , 2003, IEEE Trans. Circuits Syst. Video Technol..

[6]  C.-C. Jay Kuo,et al.  Feature-based intra-prediction mode decision for H.264 , 2004, 2004 International Conference on Image Processing, 2004. ICIP '04..