Research On Fast Intra Coding And Reconstruction Quality Optimization For HEVC Screen Content Coding

Screen content coding is an important extension of HEVC (High efficiency video coding). On the basis of HEVC coding framework, screen content coding add some new coding tools, such as ACT (Adaptive Color Space Transform) and so on. These new tools significantly improve the coding efficiency, but they also bring huge encode complexity and cannot apply in real-time application scenarios. So it is crucial to research how to lower the coding complexity and improve reconstruction video quality.The research content mainly includes two parts:1.For the problem of huge encode complexity, the fast intra CU partition algorithm based on entropy and coding bits was proposed by researching the recursive CU quad-tree partition process and its correlation with the regional information complexity. The proposed algorithm can greatly reduce the encoding time by skip many unnecessary recursive computation.2.For the problem of ACT cannot reduce the correlation of YUV video perfectly, the adaptive color space transform optimization algorithm was proposed by researching the decorrelation process of ACT and the unfriendliness of its decorrelation matrix for YUV video. The proposed algorithm can adaptively train decorrelation matrix for YUV video, which can reduce the correlation of YUV video perfectly and improve the reconstruction video quality.Experiment results show that the proposed fast intra CU partition algorithm can reduce the coding time by 32% with nearly unchanged coding efficiency at the "All Intra" coding profile. For the adaptive color space transform optimization algorithm, it can bring about 1.18%,1.02% and 0.70% average bit-rate reduction at the profile of "All Intra", "Random Access" and "Low Delay" respectively.