Research On Screen Content Video Coding Based On Inter Frame Correlation

Screen content refers to the desktop content of electronic devices that are randomly arranged by graphics,text,and natural images.It is one of the main data contents in games,animations and new media applications.Compared with the content of natural images collected by photographic equipment which has been widely studied,screen content shows many different features,such as no sensor noise,flat areas,repeated patterns,highly saturated and limited number of different colors.Effectively use of these features can enhance the coding efficiency of screen content video sequence at the basis of existing video coding standards which based on natural content.The new generation of high performance video coding standard H.265/HEVC introduced screen content coding(SCC)and developed a series of new coding technology targeted on features of screen content.In this paper,based on HEVC-SCC,reordering schemes are designed targeted on inter frame correlation to improve the coding efficiency of screen content video,which mainly includes the following two parts:Firstly,by analyzing the influence of inter frame correlation on coding efficiency of screen content video,an optimal sorting scheme is proposed to eliminate the time domain redundancy of video sequence more completely by using HEVC-SCC inter prediction technology.The optimal sorting scheme uses the inter frame correlation of the video sequence to reorder the sequences so that the sorted frames can obtain better reference frames,and then improve the coding efficiency of screen content video.The optimized sorting scheme mainly includes four parts: hash feature extraction,prediction cost calculation,optimal sequence generation and reference frame management.In video coding,the influence of inter frame correlation is mainly reflected in the inter frame prediction.The calculation of inter frame correlation is also based on the basic idea of inter frame prediction.In this scheme,the inter frame correlation is defined as the prediction cost,that is,the prediction cost of a frame with another frame played as a fixed reference frame for simulation prediction.The smaller the prediction cost is,the stronger the inter frame correlation between the two frames.We extract the hash feature of the screen content at the block level to measure inter frame correlation of video sequence and optimize sorting of video sequence according to inter frame correlation.Finally,the best reference frame is provided for the coding frame by reference frames management.The experimental results show that compared with HEVC-SCC,this scheme can improve the coding efficiency of screen content video.Secondly,on the basis of optimal sorting scheme,we propose two kinds of expansion schemes,namely,adaptive packet sorting scheme and sorting scheme based on coding structure.By analyzing the influence of different grouping lengths on the efficiency of the optimal sorting scheme,we extend the grouping strategy in the optimal sorting scheme and improve it into an adaptive packet form to solve the problem of local optimum.In the optimal sorting scheme,the calculation of the prediction cost is based on the same reconstruction quality of each frame.In the sorting scheme based on the coding structure,reordering of sequences not only considers the influence of inter frame correlation on the coding efficiency but also the effect of the reconstruction quality of the reference frame on the coding efficiency.The experimental results show that the adaptive packet sorting scheme and the sorting scheme based on the coding structure are more efficiency coded screen content video sequence than HEVC-SCC.