Research On Key Techniques Of Video Coding

With the emergence of communication networks, along with the rapid growth of multimedia service, the role played by multimedia communication has become increasingly important and essential in next generation communication system. Therein, video coding is the fundamental and kernel tech in emerging multimedia/video communication. With the sponsor of National 863 Project –"Digital Video and Audio Coding, Transmission, Testing and Application System"(No. 2002AA119010 ), we focus on the key technologies of video coding framework in this dissertation. Simultaneity, we play a positive role to push video and audio standardization as a member of Digital Audio and Video Coding Standard workgroup of China (AVS). This dissertation focus on the certain parts of video compression framework, including enhancement of coding efficiency, computational complexity control, bit rate control, and discrete wavelet transform (DWT), etc. In the mobile communication environment, transmission bandwidth is limited. To transmit video over a limited transmission channel with video quality un-degraded, video codec need higher coding efficiency than before. A novel enhanced intra prediction is proposed based on experimental statistics, which applies statistical model to improve mode prediction precise. It is obvious that more precise prediction comes with lower bit rate while holding the video quality and improving the coding efficiency. The newest video coding standard outperforms any other counterpart since it adopts many new and creative technologies. However, it also holds a terrific computational complexity. It is very essential to control computational complexity in mobile communication device with limited computation capability. Traditionally, video codec is optimized in target platform in instruction level to speedup its execution. Here, a scalable complexity control strategy for intra prediction coding is designed. Based on video spatial and temporal correlation, all intra prediction modes are divided into two categories: strong correlation set and weak correlation set. Sequentially, all modes in strong correlation set are computed to get a sub-optimal mode, then partial modes in weak correlation set is selected to compute the final coding mode based on computational redundancy. The proposed method can precisely control computational complexity of intra coding while holding almost coding performance. The strategy should be useful and essential in mobile communication system. Rate control is not a normative part of a video coding standard, but it is very important in video codec implementation. There are different rate control algorithms adopted in individual coding standard for various applications. The key step of rate control is to adjust quantization parameter based on bit rate predictive model QP=f (R). In this dissertation, we propose a frame-based rate predictive model which is a function of actually coded bits and valid bits of DCT coefficients in a macroblock. Then, a novel rate control strategy is developed, which holds a good performance in low bit rate application environment. Compared with Discrete Cosine Transform (DCT), DWT has its own advantages, e.g. Wonderful coding scalability and small block artifacts, etc. DWT is adopted in still-image coding standard-JPEG2000 and applied to video coding standard-Motion JPEG2000. Here, we propose a DWT-based post-process method to enhance or recover details of lossy coded image. Bigger compression ratio comes with more effective result with this method. Especially, the proposed method is very easy to implement without more computational complexity. Simultaneously, parts of productions in this dissertation are proposed to AVS as technical proposals and adopted in normative body of AVS-M.