Fast Algorithm Of Intra Prediction For VP9

VP9 is the next generation video coding standard that developed by Google in 2013. It still utilizes the traditional hybrid video coding framework and also adopts the newest video coding technology in recent years. As a result, the coding performance of VP9 has a significant increment relative to VP8, the previous video coding standard developed by Google, and it achieves a bit rate reduction of about 50% compared with the VP8 coder while keeping the same perceptual video quality. However, the improvement of compression ratio sacrifices the coding speed at the same time. Based on the former problem, this paper mainly focuses on the fast algorithms for intra prediction with expecting to reduce the coding time and thus make VP9 standard more suitable for real time services.In order to reduce the coding complexity of VP9, the key technologies of intra prediction in VP9 is firstly studied and analyzed in this paper. Based on which, two fast algorithms of intra prediction are then proposed for VP9, they are the fast algorithm based on mode selection and the fast algorithm based on block partition, respectively.To reduce the computational complexity of mode selection in intra coding in VP9, a fast algorithm of intra prediction based on mode selection is proposed. During the intra coding process, each super block(64×64) has to do 6810 times of mode selection so as to determine the optimal partition and the corresponding intra prediction mode. Thus it can greatly reduce the time of mode selection by reducing the number of candidate modes. In this paper, the number of candidate intra-prediction modes is reduced based on the texture direction of the current block and the optimal intra predict ion mode of the neighbor block. Experiment results show that the proposed method can reduce the coding time by 55% in average with a negligible degradation on coding performance.To fasten the process of intra prediction block size decision, a fast algorithm of intraprediction based on block partition is proposed. The main idea of this algorithm is to determine the optimal partition directly or skip some unnecessary partition. Firstly, the factors which influence the optimal partition are analyzed based o n experiment results,they are spatial complexity, smoothness and the quantization parameter. Based on which the intra prediction can be done in two separate method. For the flat and smooth areas, only large blocks are used, and on the contrary, only small blocks are tested for the rest areas. Experiment results indicate that the proposed method can greatly reduce the encoding time under the identical perceptual video quality.