H. 264 Study On The Quantitative Parameter Selection And Rate Control Algorithm

H.264 is the up to date international video coding standard which is jointly developed by ITU-T and ISO. Compared with other video compression standards, H.264 has made obvious progresses in coding efficiency and network adaptability. All range of services such as wireless (mobile) video, digital video broadcasting and online streaming media are supported. However, the increasing complexity of its algorithm imposes limits on its performance. In order to deploy the standard on different platform and in various environments, it should be quite meaningful to improve the algorithms and structures of H.264 standard according to the practical needs for academic and industrial purposes.Firstly, this dissertation briefly introduces the H.264 standard. Secondly, taking the rate-distortion technology as the central thread, the research on intra-frame prediction mode decision and rate control algorithm will be presented. The main content is as follows:(1) Rate-distortion theory and its correlative model will be elaborated at length while Lagrange multiplier method and hypothetical reference decoder are simply addressed.(2) Due to the numerous candidate modes for intra-frame prediction, the cost of complete rate-distortion optimization (RDO) is unbearable. So the latest fast mode decision methods based on the RDO algorithms are summarized and analyzed, after which some schemes are proposed to better the performance.(3) The three-layer rate control mechanism of H.264 is exploited. Aiming at the deficiency of rough initial quantization parameter (QP) decision of H.264, an adaptive initial QP selection algorithm is presented through defining the complexity measurement of I frame. Then an optimized linear model for Mean Absolute Difference (MAD) prediction combining the temporal and spatial domain MAD is brought forward to solve the chicken and egg dilemma. (4) The improved algorithm above is implemented by employing the H.264 reference software JM13.1. Intensive simulation is conducted to compare with the official resolution JVT-H017. The results validate that the proposed method calculates the initial QP more accurately and provides better rate control precision and rate-distortion optimization effect.