Research On Rate Control Scheme Of HEVC

High Efficiency Video Coding (HEVC) is currently being prepared as the newest videocoding standard of Collaborative Team on Video Coding (JCT-VC). As the standard is beingmodified and completed, it will be widely used to encode video with high definition. In manycases, however, the network bandwidth is limited, how to allocate the bandwidth to guaranteethe quality of the video, in other words, the rate control techniques, is becoming an importantissue. In the HEVC codec, the rate control scheme of JCTVC-H0213proposal is used. Thisthesis analyzes the algorithm of the scheme and tested its performance, then, some improvedalgorithms are proposed based on the original rate control scheme.Some improvements are proposed to modify the pixel-based URQ (UnifiedRate-Quantization) model, then an optimization model is presented basing on the theoreticalanalysis and experimental results. The improvements include the modification of quantizationparameter in the model, the new parameter in the model and the method to predict thenon-texture bits. Predicting the non-texture bits and minus it from the target bits of a frame,the target texture bits can be used in the RQ (Rate-Quantization) model. The experimentalresults show that the improved model achieves more accurate results and the performance ofrate control is also better. Furthermore, the thesis proposes a bits allocation algorithm basedon the image content, including a frame-layer bits allocation algorithm in which the positionof frames are considered, and a picture complexity-based algorithm to predict theLCU(Largest Code Unit) which texture bits is zero. The proposed algorithms improve theaccuracy of bit allocation by considering the position of frames and the LCU with zero texturebits. The experimental results indicate that the improved bit allocation can enhance theperformance of the rate control scheme effectively.