The Study And Implement Of Rate Control Algorithm Based On H.264

With the development of the communication, the application of multimedia telecommunication is growing rapidly. However, the video must be compressed before transmission, because the video information is large and requires more bandwidth. Current encoders adopt predictive and transform mothods. Because the complexity and motility of images are not the same, it will lead to huge change in output bits. It is necessary that the change is controlled to a limited range by rate control. Rate control is one of the key techniques for video coding. It affects the bandwidth of encoded bits and the quality of reconstructed images. Every standard has its own rate control scheme. H.264/AVC is the latest video coding standards. However, the quantization parameters are involved in both the Rate-Distortion Optimization (RDO) and rate control, which makes rate control more difficult.Although current G012 rate control scheme can achieve better control in H.264, it doesn't consider the impact of scene change, which degrades the quality of the frames after scene change. In order to address this issue, G012 scheme is been investigated and a new rate control strategy is proposed.Firstly, the RDO and rate control algorithm are analysed and a new MAD prediction algorithm is proposed. By utilizing the similarity between the MVs in two adjacent frames, the new algorithm computes the MAD using current frame and the MVs in previous frame. The algorithm can achieve similar results with G012 scheme, and it can detect scene change because it utilizes the current frame information.Secondly, a new scene change detection algorithm is proposed based on the above mad prediction algorithm. The detection algorithm can detect the scene change accurately without more computation.Thirdly, the target buffer level is adjusted after scene change, which can improve the video quality after scene change.Lastly, the new rate control strategy is applied in current encoder.Simulation results show that the new strategy effectively alleviates visual quality degradation after scene change with little impact on the overall quality, and gives average 2db improvement in peak signal-to-noise ratio (PSNR) after scene-changed, compared with G012 scheme.