Research On Fast Sample Adaptive Offset Algorithm For VVC

With the advent of the big data era and the 5G era,the applications of video images are increasing,which have put tremendous pressure on compressing and encoding video images.In order to improve the coding quality of the video and save the coding rate,Versatile Video Coding(H.266/VVC)came into being.Compared with the previous generation of High Efficiency Video Coding(H.265/HEVC),VVC can save nearly 50%of the bit rate while under the same quality.However,it still uses a block-based hybrid coding framework,which will inevitably cause the loss of high-frequency components of video signal and produce ringing effect after the process of transformation and quantization.Although Sample Adaptive Offset(SAO)is still used in VVC to adaptively add offset values to pixels in the reconstructed image during encoding,thereby reducing the ringing effect.However,the statistical process of SAO is complex and the computational complexity is high.In response to these problems,this thesis improves the standard SAO algorithm for VVC to reduce the coding complexity.The specific research content of this thesis is as follows:1.This thesis proposes a fast sample adaptive offset algorithm based on Histogram of Oriented Gradient(HOG)and depth information.This thesis first analyzes the relationship between the depth information and the offset mode of SAO in VVC,and uses the depth value to make predictions in Edge Offset(EO)mode and Band Offset(BO)mode in advance.Secondly,this thesis analyzes the relationship between HOG features and the direction of EO mode,and decides the best direction of EO mode in advance by extracting HOG features.Finally,the two optimization methods are integrated to effectively reduce the SAO filtering time.Experimental results show that the fast sample adaptive offset algorithm based on HOG features and depth information proposed in this thesis can effectively reduce the SAO filtering time by 67.79% with only 0.52% coding gain loss.2.This thesis proposes a fast sample adaptive offset algorithm based on the combination of time domain and spatial domain.In the time domain,the best matching CTU of the current Coding Tree Unit(CTU)is found in the reference frame through block matching,and the SAO mode of the best matching CTU in the reference frame is used as the reference SAO mode of the current CTU.In the spatial domain,the intra-frame prediction angle mode direction is mapped to the EO direction,and the mapped mode provides a reference for the SAO mode of the current CTU.Finally,this thesis combines the time domain and the spatial domain to optimize the SAO process together by depth.Experimental results show that the algorithm can reduce SAO filtering time by 57.39% in low-delay P frame coding mode,and the coding gain loss is only0.35%.