Research And Implementation Of Video Encoding Algorithm Based On Human Visual System

With the progress of science and technology and economic development,people's demand for digital video related services is growing,the video encoding technology as the basis of the core technology of video business,and gradually become the focus of attention of the industry.The mainstream of video coding standards,such as H.264,H.264,AVS series have followed traditional hybrid coding framework,along with the related technology is gradually mature and perfect,this kind of traditional coding technology end calculation complexity is increasing,and coding efficiency enhance more and more slowly.In order to break through the traditional technology framework,based on the human visual system(HVS)video encoding algorithm came into being.This technology uses HVS visual perception characteristics,through the removal of video information in the perception of redundancy to achieve compression encoding.Combined with the research status,this paper firstly proposes a comprehensive consideration of HVS characteristics,which can be applied to the visual perception model of encoding.The model outputs a visual perception coefficient VPI to each macro block,which reflects the importance of the macro block to the HVS.The calculation of VPI,considering the characteristics and mechanism of the persistence of vision and the ability to distinguish the limited HVS visual attention.The experimental results show that the VPI can accurately reflect the subjective feelings of the human eye.This paper put forward a based on visual perception of the rate distortion optimization algorithm,algorithm to structural similarity as the distortion metric,and the visual perception coefficient VPI of Lagrange coefficients are adaptively adjusted,so as to choose the best coding mode.Experimental results show that the proposed algorithm can improve the efficiency of 7.1%in the case of almost no increase in the complexity of the algorithm.