Research On Related Technologies For Video Performance Improvement

With the increasing requirements of people with the video experience, the process of super/high definition videos has become the development direction of the modern video techniques. However, the huge data volume has become the obstacle of their development. The Joint Collaborative Team on Video Coding (JCT-VC) is working on the next generation of video coding standard—high efficiency video coding (HEVC), which is developed on the base of H.264/AVC. This thesis investigates and tries to improve two of the technologies emploied in HEVC, which are the adaptive loop filter (ALF) technique and the advanced motion vector prediction (AMVP) technique for inter-view prediction.To alleviate the impact of quantization noise to video quality, a two step compensation method is implemented to the reconstructed pictures. Firstly, the DC compensation is carried out. Then, ALF is utilized to do the AC compensation. Since the quantization error of image data is mostly stationary and the correlation between images is currently symmetrical, the symmetrical ALF design methods are investigated. After that, an central symmetrical ALF design scheme is proposed. The method reduces the number of required ALF coefficients transmitted in coded bitstream, and then improves the video compression efficiency. Experimental results show that the two step compensation method can obviously improve the video quality. And at the mean time, the central symmetrical design scheme can decrease the data amount for ALF coefficients, which improves the coding efficiency while processing those pictures with symmetrical features.AMVP is one of the key technology for inter prediction in HEVC. This thesis introduces the development of AMVP, and try to do some improvements over it. Based on the distribution features of the motion vector predictors (MVPs) according to their conformance with the MV value, the amount and the order of MVP candidates are adjusted. The experimental results show that the proposed method can improve the coding efficiency for certain video sequences.