Research And Implementation Of Inter Prediction Optimization Based On Pre-Analysis

Video can give users a large amount of rich information in the most intuitive way,and users also have a great demand for video related applications and services.However,because the original video is composed of frames of pictures,the huge amount of data brings great difficulties to the storage and transmission of the video,so the original video needs to be compressed and encoded before it is put into use.Due to the strict correspondence between encoder and decoder,video encoding standard provides a unified stream format and corresponding technical specifications,which also makes the video encoding standard become the basis of the development of the video industry.The improvement of video resolution and the emergence of new video applications such as live broadcast also bring new challenges to video encoding standards.In order to meet the needs of encoding and compressing ultra high-definition video,a new generation of video encoding standards AVS3 and VVC are being formulated.Compared with the previous generation,the new generation of video encoding standards will achieve higher compression performance.However,compared with the previous-generation standard AVS2,the new-generation video encoding standard AVS3 adds a large number of new encoding technologies under the original encoding framework.Although these encoding technologies have greatly improved the coding performance of the encoder,they have also greatly increased the coding time of the AVS3 encoder.How to design fast algorithms for encoder has always been a hot topic in the field of video encoding.In this paper,based on AVS3 encoder,the related technologies of inter prediction coding are intensively studied.The main contributions and innovations are as follows:This paper proposes a block split prediction algorithm based on spatial correlation.Aiming at the spatial correlation of the coding block and its spatial neighboring blocks,a split complexity calculation method is introduced,and the split complexity of the current coding block is predicted according to the split complexity of the spatial neighboring blocks.Finally,according to the split complexity of the current coding block and its prediction value,whether to code current coding block and whether to skip subsequent split is decided and thereby skipping the non-optimal split mode.This paper also proposes a compensation mechanism based on rate-distortion cost,which can effectively improve the performance of the algorithm.The algorithm proposed in this paper can reduce the coding time of the SVT-AVS3 encoder by 25.83%,with only 0.31%performance loss.This paper also proposes an inter-prediction optimization algorithm based on pre-analysis.The AVS3 standard has newly adopted two split modes,binary-tree partition and extended quad-tree partition.As a result,the same image area will be encoded multiple times,which greatly increase the coding time.In this paper,based on the existing fast algorithm of HPM reference software,the inter prediction information of the firstly coded split mode is used to optimize the split decision process,so as to improve the coding speed.In the algorithm proposed in this paper,the split condition of current coding block is first decided based on the information of the inter prediction of the firstly coded split mode.According to the split condition,all available split modes are decided,and the non-optimal split modes are skipped.The fast algorithm proposed in this paper can improve the coding speed with a certain performance loss.