Research On Accelerating Algorithm Of New Generation Video Coding

With the continuous development of digital media,video technology has been increasingly used in various fields.However,due to the large amount of data carried by the video itself,the direct application is not realistic.Therefore,before transmitting the video information,it needs to be compressed.With the requirements for high resolution video,video coding technology has undergone a series of developments,such as H.264,HEVC,and so on.In view of the high complexity of these coding standards,related optimization algorithms have also been proposed.However,due to the introduction of many new technologies in the new generation of standard VVC,the coding time has also increased by more than twice that of HEVC.Therefore,we need to propose an improved optimization method.This article first introduces the research background and significance of video coding and introduces the development process of video coding standards and representative coding standards in detail.By researching and analyzing VVC and its corresponding reference software JEM,we have proposed relevant solutions that can increase the encoding speed of next-generation video.(1)A block partitioning decision algorithm based on content complexity is proposed.This algorithm is mainly for encoding acceleration for other frames except for the first frame.By analyzing the relationship between the content complexity of the video and the different partition types in advance,it is possible to reduce unnecessary attempts for partition types according to the scope of the current block complexity value.Compared with the original encoding process,this method achieves the purpose of reducing encoding complexity well.The average encoding time is reduced by 9.0%,while the encoding performance is only lost by about 0.55%.(2)A QTBT fast partitioning algorithm based on a priori object area is proposed.The algorithm firstly adaptively limits the two partition parameters at the CTU level to reduce the number of unnecessary divisions.Secondly,at the CU level,by analyzing the prior object information of the current frame,such as the prior depth information and the color information of the image,a large number of unnecessary division processes are reduced.The results show that this method reduces the encoding time by 35% on average,while the loss of encoding performance is only 1.5%.(3)A fast intra prediction mode selection algorithm is proposed.The algorithm first obtains an initial pattern in the MPM mode list,and then searches at different search steps depending on the situation: two modes are searched and selected in the range of 16 modes on the left and right sides of the mode,and one mode is obtained in the remaining ranges.Then perform a full search again for more detailed selection.Finally,by analyzing whether the current area is smooth,the final set of mode candidates for RD cost calculation is determined.The results show that this method is effective and can save an average of 23.28% of encoding time,while the loss of encoding performance is only 1.49%.Finally,the content of this paper is summarized,and the research directions and perspectives for the next stage are proposed.