| Virtual reality technology is developing rapidly,and related applications have emerged,such as virtual reality video.Virtual reality video has a huge amount of data and is a big challenge for existing transmission and storage systems.The latest High Efficiency Video Coding(HEVC)coding standard can compress and process flat video,but cannot directly compress spherical virtual reality video.Therefore,the Joint Video Experts Team proposes a scheme for converting virtual reality video into two-dimensional video through projection and then using existing video coding standards to code.However,the high frame-rate and high resolution characteristics of virtual reality video make the compression process extremely computationally complex.In order to reduce the coding complexity and reduce the output bit rate,the paper focuses on the intra coding and quantization of virtual reality video in equirectangular projection(ERP)format,and a fast CU partitioning and an adaptive QP algorithm for virtual reality video are designed.The main innovations are as follows:(1)Aiming at the problem of high coding complexity in virtual reality video,this paper studies the projection and stretching of virtual reality video in ERP format,and statistics the relationship between coding unit depth and projection stretching degree.A CU fast partitioning algorithm based on partitioned depth interval prediction is designed.The current frame is divided into two regions according to the WS-PSNR weight.According to the different degree of stretching of the image of different regions in the ERP projection process,different adjacent reference blocks are used to predict the depth interval,which reduces the complexity of the segmentation.The paper further studies the relationship between PU depth,projection partition and intra prediction mode,and designs a fast decision method for intra mode,which reduces the number of candidate modes and reduces the coding complexity.The experimental results show that the proposed algorithm can reduce the coding time by 39.3%,while the BD-rate only increases by 0.8%.(2)The high frame-rate and high resolution virtual reality video code stream has a huge amount of data.In order to reduce the output bit rate and ease the transmission pressure,an adaptive QP algorithm is designed.The paper analyzes the relationship between WS-PSNR weight,quantization parameter and reconstruction quality in ERP projection format.QP compensation method is used to compensate positive for WS-PSNR with small weight and negative for WS-PSNR with big weight to achieve the effect of adaptive adjustment QP.The experimental results show that the proposed algorithm can reduce the BD-rate by 1.77%,and the WS-PSNR remains basically unchanged. |
