| With the rapid development of multimedia transmission technology and the improvement of player performance,users have higher requirements for the quality and mode of watching video.Virtual Reality(VR)technology or 360 degree video technology with its unique viewing experience to bring users a strong sense of presence and it is very popular.360 degree video is popular for its immersive experience to users,but with the accompanying ultra-high resolution and huge data volume,it faces the problems of large transmission bandwidth requirements and high coding complexity.Therefore,based on the High Efficiency Video Coding(HEVC)standard,this paper studies two core technologies to solve the problem of large amount of data and high coding complexity in the process of 360 degree video coding and transmission.The specific research contents of 360 degree video in this paper are as follows:(1)Aiming at the problem of large amount of data in 360 degree video transmission,a 360 degree video coding algorithm based on Region of Interest(ROI)is proposed.The ROI is extracted from 360 degree video by coding residuals.By allocating more bits to ROI and lower bits to non-ROI,the ROI can maintain better subjective quality even when the overall bit rate is reduced.The experimental results show that,compared with the original HEVC reference software,the proposed algorithm improves the video quality by an average of 0.15 d B at the same bit rate,saves the highest bit rate by 4.96% and save 2.46% on average under the same objective quality conditions,and maintains the video quality of ROI effectively.(2)To solve the problem of high complexity of 360 degree video coding,a fast inter coding algorithm for 360 degree video based on motion attributes and texture features is proposed.Firstly,based on the characteristics of 360 degree video,a motion attribute discrimination method is proposed to determine whether the Coding Unit(CU)to be coded is a static block or a moving block.The edge features of the current Largest Coding Unit(LCU)are calculated to establish a texture quad-tree to predict the partition depth and prediction mode.Then,when the motion attributes of the current LCU are known,the partition depth and motion attributes of the adjacent coded LCU in spatial domain are used to predict the partition depth range of the current CU.The motion and texture features of CU and the category of prediction modes of Prediction Unit(PU)that the current encoding CU belongs to.The experimental results show that compared with the original HEVC reference software,the proposed algorithm achieves an average coding time saving of 72.27% when BD-rate increases by only 1.74%.It achieves the goal of effectively reducing the complexity of inter-frame coding of 360 degree video.(3)Faced with the problem that the transmission method based on projection plane needs very high transmission bandwidth,a hierarchical coding method based on salient region is proposed.The basic layer transmits low-quality video.The viewport layer is divided based on Tile block in HEVC,and the salient region allocates larger bit rate,while the non-salient region allocates less bit rate.Compared with the contrast method based on projection plane,the hierarchical coding method based on viewport has the same performance as that based on projection plane in terms of bit-rate file storage and video quality.However,the transmission bandwidth of the scheme based on viewport is lower than that based on projection plane because of the small redundancy of transmission. |
PDF Full Text Request