A CONTEXT-BASED ADAPTIVE FAST INTRA_4x4 PREDICTION MODE DECISION ALGORITHM FOR H.264/AVC VIDEO CODING

In this paper, we propose a context-based adaptive fast Intra_4x4 prediction mode decision algorithm for H.264/AVC. Firstly, a 3-order Markov random field model is introduced to describe the spatial distribution of RD optimal Intra_4x4 prediction modes at picture level. Secondly, in a neighboring context specified by this model, only a small set of candidate modes are chosen for RDO calculation according to a constrained error probability criterion. Thirdly, to reduce candidate modes further, thresholds which can be estimated by linear regression are adopted to perform early terminations. Finally, the proposed algorithm can adapt itself well to diverse video sequences due to an ability of automatic adjustment, while the model and thresholds are initialized off-line by statistics. No apparent computational overheads are involved throughout the algorithm. Experiments show that the novel fast algorithm searches only 2 6 Intra_4x4 modes per block and reduces up to 55% encoding time with a penalty on coding efficiency less than 0.1 dB. Index Terms H.264, intra-prediction, fast algorithm