The Research On Video Compression Algorithm Based On Subjective Perception

The new generation of coding standards,such as VVC,AV1,and AVS3,improve cod-ing efficiency mainly through more modes or more complex transformations.Essentially,it is trading huge time complexity for limited space complexity.This article explores the possibility of improving coding efficiency from the perspective of subjective visual qual-ity.Traditional video coding technology has always used objective evaluation indicators as a distortion measurement method,and it is still difficult to replace the subjective effect judgment of the human eye.This paper takes the subjective effect-based video coding optimization technology as the research theme,and focuses on the modeling of subjective perception indicators of the human eye.In addition,use subjective indicators to optimize the rate-distortion optimization process and rate control process of traditional encoding,so that the reconstructed video can achieve better subjective effects of the human eye.The main research content is divided into three parts:(1)Analyzed the traditional methods of extracting subjective perception indicators of the human eye and methods of optimizing the video encoding process.Combining the application of the free energy principle in human vision,three human subjective perception indicators are proposed,which can be calculated in a short time complexity,so that they can be applied in the field of video coding.The proposed subjective perception index can be closer to the subjective perception of the human eye,and it can be used to reduce the visual redundant information of the video.(2)Analyzed the principle of traditional coding rate-distortion technology,and op-timized the rate-distortion process by using subjective perception indicators.The rate-distortion process can be thought of as making a choice between distortion and bit rate,and the distortion index used in traditional rate-distortion has always been an objective index.Increasing the weight of the subjective index in the distortion allows the rate-distortion optimization to select a more appropriate encoding mode,thereby encoding images with higher human visual quality.(3)Analyzed the principle,advantages and existing problems of the rate control algo-rithm of HEVC encoder.By predicting motion intensity and scene switching,the process of frame-level bit allocation is optimized.Through the detection of salient regions and structural texture content,the process of CTU-level bit allocation is optimized.When the rate control encoding is turned on,the bits allocated at the frame level and CTU level can be more matched with the complexity of the image,thereby obtaining the improvement of subjective visual effects.The full text work is tested with HEVC standard HM16.17 reference software and x265₃.4 commercial encoder.Large-scale test results show that the proposed rate-distortion optimization algorithm based on subjective perception in the RA mode has increased the PSNR-Y and SSIM-Y index in the significant area by 2.00%and 0.40%,besides,the over-all area PSNR-Y index has decreased by 0.59%,and the SSIM-Y index has increased by 0.72%.In the proposed rate control algorithm based on subjective perception,the PSNR-Y index in RA mode is increased by 0.73%,the SSIM-Y index is increased by 1.38%,and the rate control error is 0.47%.In LD mode,the PSNR-Y index has increased by 0.30%,the SSIM-Y index has increased by 0.82%,and the code rate control error is 0.23%.