Fast Intra-Prediction Mode Decision in H.264/AVC Based on Macroblock Properties

Intra-prediction is a widely used technique in intra coding. H.264/AVC adopts rate-distortion optimization (RDO) technique to obtain the best intra prediction, while maximizing visual quality and minimizing the required bit rate. However, the RDO reduces the encoding speed via the exhaustive evaluation of all candidate modes. In this paper, we decrease the encoding time by reducing the number of candidate modes. The proposed algorithms use spatial properties of macroblock (MB) (such as directional edge information), and some of statistical properties (such as similarity of reference pixels and adjacent blocks information) to select a subset of all candidate modes. Subsequently, the RDO procedure uses the reduced subset of all candidate modes for extracting the final mode. Experimental results show that our algorithm, compared to the RDO and some other fast algorithms, reduces the total encoding time with negligible loss in PSNR and a slightly increased bitrate.