The Study And Optimization Of H.264 Real Time Encoder

With the rapid development of multimedia technology, video compression is widely applied in many fields. In order to raise the efficiency of compression, a new video compression standard H.264 was proposed by JVT in 2003. Compared with the former standards, the video compression efficiency of H.264 is much higher due to the adoption of some key technologies, such as intra-frame prediction, multi-reference-frame motion estimation and integer DCT transformation. However, using these above key technologies will increase the encoding complexity of H.264, at the same time limits its application scope in practice. x264 is a kind of open-source encoder that can support H.264 standard, which has already been utilized in many projects. Because of the adoption of many optimization methods both in algorithm and coding, x264 holds a leading position in the aspect of coding performance among many other open-source encoders. With the spread of open-source concept, x264 will have an expansive prospect in development and application.The purpose of this paper is to optimize the real-time encoder x264, increasing its real-time encoding performance. Firstly, this paper deeply studies the x264 encoder on the basic of a thorough knowledge about H.264 standard. Three parts are mainly analyzed, they are: encoding process, intra-frame prediction and inter-frame prediction. Secondly, considering the high complexity of multi-reference-frame motion estimation, this paper proposes an optimized encoding algorithm based on time and space relativity of the video sequence. According to the time relativity of the video sequence, this algorithm firstly predicts the pictures' motion situation and decides the number of maximum reference frame. Then it selects different motion searching algorithm after predicting the local motion situation according to the space relativity of the pictures, hexagon searching algorithm for intense motion area and diamond searching algorithm for slow motion area. Experiment results show that this algorithm can raise the encoding speed by 3-9fps in CIF video format on the condition that the video sequence quality is guaranteed. At last, the fixed frame structure is adopted instead of using the B-frame adaptive judging process. In addition, the multithread support in windows platform is also added, thus the optimization of this combination algorithm in coding level is realized in this paper.As the application field of video compression expands larger and larger, it is required by objective condition that video compression algorithm should have low computing complexity, at the same time, the compression efficiency should also be guaranteed; otherwise, it is not suitable in practice. This paper achieves the optimizing method in x264, decreasing the coding complexity. The encoding speed can be accelerated and the encoder's real-time ability can be enhanced if we can ensure the quality of the video sequence.