Fast Intra Algorithm Based On Texture Feature And Cart Decision Tree For Virtual Reality 360-degree Video

Virtual reality(VR)360-degree video is now widely used in various fields.However,its high resolution increases the coding complexity.While the existing video coding framework,such as HEVC,is not optimized according to the characteristics of virtual reality video.In order to reduce coding complexity and shorten coding time,this paper proposes a fast intra coding algorithm based on texture feature and Cart decision tree.The algorithm is divided into two aspects as follows.(1)This paper designs a fast CU partitioning algorithm based on texture contrast,in view of the complexity of the segmentation process of the virtual reality 360-degree video coding unit in the ERP format.Texture contrast(TC)is defined as an index describing the texture characteristics of the coding tree unit(CTU)and coding unit(CU)The depth range is predicted by calculating the texture contrast of the CTU,and then the depth of the CU is calculated within the depth range.The texture contrast is used to realize the depth prediction of the CU,avoiding the traversal process of all depths in the reference model,and shortening the encoding time.The algorithm uses statistical methods to obtain the division decision thresholds of CUs of different depths in each area.Experimental results show that,compared with the HM16.20 reference model,the algorithm saves 33% of the coding time on average,while the BD-rate only increases by 0.80%.(2)A fast intra-mode selection algorithm based on texture features and decision tree is proposed in view of the high complexity of the virtual reality 360-degree video mode selection process.The algorithm achieves rapid decision-making of prediction modes by reducing the number of candidate modes in the intra-prediction process: for polar regions,the selection probability of the candidate modes in the horizontal direction is significantly greater than that in the vertical direction,and the candidate modes in the vertical direction are greatly reduced;For the equatorial region,the cart decision tree is constructed by using the texture contrast of the four directions of the prediction unit as attributes,and the classification results of the decision tree are used to filter out the most likely prediction mode of the current coding unit.Experimental results show that compared with the HM16.20 reference model,the proposed algorithm saves an average of 28% of the coding time,and the BD-rate only increases by 0.61%.