The Research And Implementation Of Entropy Decoding And Loop Filter Algorithm Based On H.264

As the H.264 brought codec efficiency the same time has also brought increased computational complexity. So how to reduce the computational complexity of H.264 to meet the common embedded platform by software method and implement the H.264 applications ultimately on mobile devices has became the research focus in many research institutions and institutions. The study goal in this paper is to achieve in real-time decoding on ARM9 platform through algorithm improvements and code optimization for the time-consuming modules of the H.264 decoding process and the main work as follows:The second chapter in the paper is about the research of the CAVLC decoding algorithm. This chapter first analyzed the main reason what affecting the entropy decoding speed in CAVLC entropy decoding process-finding variable-length code table, at the same time test software has been used to analyze time-consuming to identify the main time-consuming module. Then this chapter investigated the method about the related literature's author how to optimize the variable-length code table, on this basis, the paper put forward an improved algorithm for CAVLC decoding——based on data storage format and the code table address mapping algorithm. By changing the data storage format in this chapter, the storage space was saved. We can determine the mapping between the code word and the array subscript by analyzing the code word characteristics. By using the subscript to find matching information can reduce the lookup operation and the algorithm's computational complexity. Finally, the chapter realized and proved the algorithm.The third chapter is about the research of the loop filter algorithm. This chapter first analyzed the filter determining conditions in loop filter algorithm. Because the encoder use many forecasting models, to lead to a harsh judge conditions for filtering. Then the article proposed a effective way by judging the image macro block boundary is smooth. The new algorithm first calculated the entire border of the gradient, if the gradient thresholds in a given context, to be determined as smooth, use a strong filter, or points of judgment. The macro block within the block has been divided into intra and inter-frame to discuss. the intra frame use of moderate intensity filter and the inter frame use weak intensity filter or not according to prediction block size. Finally, the chapter realized and proved the algorithm.Chapter four selected JM11.3 decoding code for the study object and verified the effectiveness and feasibility of improved algorithm in this paper by verifying the decoding process is real-time implementing on ARM+WINCE platform.This chapter first introduced the transplant of the decoding process, then discussed the decoding process optimization and code optimization based on C and ARM assembly language. Finally, we edited a player application under EVC and verified the feasibility of real-time decoding of the optimal decoding process by decoding side by side in the form of play.