Research On Mode Decision Algorithm For H.264/AVC Intra Prediction

The intra prediction technology of H.264/AVC selects the best prediction mode by rate-distortion optimization strategy, which improves the encoding efficiency of the I-frames. However, the computational complexity of intra prediction process is increased and the speed of intra encoding process is reduced at the same time. In order to solve this problem, two fast mode decision algorithms for intra prediction are proposed in this paper, which are derived from the thorough analysis and researching for the rate-distortion optimization intra prediction mode decision algorithm. The main work of this paper is as follows:1. The basic principle and main algorithms of H.264/AVC intra prediction technology are researched, the algorithm complexity of rate-distortion optimization intra prediction mode decision algorithm are analysed and some representative fast mode decision algorithms for intra prediction are researched in this paper.2. A fast mode decision algorithm for intra prediction is proposed. This algorithm based on the directional features of luma block could get the best intra luma prediction mode by rate-distortion cost calculation among a few of more probable prediction modes which could be obtained by a little of simple computation. Given the characteristics of chroma prediction mode decision process, the computation of SATD instead of the rate-distortion cost is used to selects the best intra chroma prediction mode at the same time. Experimental results show that compared with the full search algorithm, this algorithm can reduce the encoding time about 48% with the increase of bit rate less than 3% averagely, and the image quality of this algorithm is almost the same as that of the full search algorithm.3. A fast mode decision algorithm for luma prediction is proposed. This algorithm based on the smooth degree of luma macroblock selects the suitable luma prediction type for current luma macroblock by the predefined luma prediction type judgment rule and the value of smooth degree characteristic quantity SADLM . Only one type of luma prediction modes'rate-distortion cost are need to be computed when the characteristics of current luma macroblock is obvious, which could reduce a lot of unnecessary computation. Both of two types of luma prediction modes'rate-distortion cost are need to be computed when the characteristics of current luma macroblock is unobvious, which could ensure the accuracy of this algorithm. Experimental results show that compared with the full search algorithm, this algorithm can reduce the encoding time about 59% with very little increase in bit rate and very small losing in peak signal to noise ratio averagely and almost have no influence on image quality. In addition, the simpler the image needs to be encoded is, the better the optimization effect of this algorithm will be.Both of the two fast algorithms are the effective improvement for the full search algorithm of H.264/AVC reference model, have the similar rate-distortion performance as the full search algorithm and get the expected optimization effect.