Research And Optimization On Mode Decision And Rate Control Algorithm Of H.264

Nowadays, there are more demands of multimedia services as the progress of information technology and the increase of network bandwidth. As an important part of multimedia technology, video encoding is a hot topic in research and engineering domain. There are many international video coding standards such as MPEG-2, H.263 and MPEG-4. In 2003, a new video standard called H.264/AVC is published by ISO and ITU-T. In order to improve encoding performance, H.264 adopts many advanced encoding technology such as inter/intra prediction, sub-pixel motion estimation, loop-filter and multiple reference frame. Unfortunately, such excellent coding efficiency comes with extremely high complex computations. How to reduce the computational complexity of H.264 is a hot topic in video coding field.We do a deep research on H.264 standard and illustrate some key technology such as inter/intra prediction, transform and quantization, entropy encoding and loop-filter.Then, we analyze mode decision algorithm of H.264 and find the defect of default algorithm. Based on summary of existing fast mode decision algorithm and research status, we propose a novel inter mode decision algorithm. This algorithm includes early SKIP mode detection and unnecessary RDO computation exclusion. Based on our algorithm, only a small number of intermodes are selected in the RDO computation. Experimental results show that our algorithm can averagely reduce 71% coding time and 1.94% bitrate compared with the reference software JM11.0 only with 0.0875dB PSNR loss.Finally, we do research on rate control algorithm of H.264 and make improvement. First, we introduce theory foundation of rate control and summarize rate control algorithm of existing video encoding standards. Then, we give and elaborate introduction of rate control algorithm of H.264 and find the defect this algorithm. We proposed a new rate control algorithm with optimal initial QP selection for very low bitrate. Our algorithm is based on frame complexity prediction.Experimental results show that our algorithm can prevent buffer overflow and underflow with average PSNR improvement by 0.31dB compared with traditional rate control algorithm under low bitrate condition.