Fast Intra-prediction Of HEVC Based On Support Vector Machine

Video is one of the most popular forms of media.Video has a wide range of applications,but a single video occupies a large storage space.Therefore,the video coding technology with good compression performance and low complexity becomes the current research hotspot.The paper takes HEVC/H.265,the latest generation video coding standard,as the research object.Firstly,it introduces the requirements and research status of video coding technology,expounds the basic principles of coding technology and the development history of the international mainstream coding standards,and analyzes the HEVC coding standard.New technology,looking for breakthroughs in coding optimization.Based on this,the following studies have been done on the HEVC standard:1.Observing the main technical features of HEVC and analyzing the main reasons for high coding complexity;2.For the problem of long decision-making time of quadtree structure in intra-prediction in the new standard,pre-pruning of quadtrees is performed using a support vector machine classification algorithm.3.Through the parameter screening optimization algorithm and experiments,compared with the existing intra-frame fast algorithm.Experimental results show that the encoding time of the intra-frame prediction algorithm based on the support vector machine has little difference with other algorithms,and the higher the video resolution,the more time is saved;in terms of prediction accuracy,compared to other intra-frame fast algorithms,this paper's algorithm has a lower coding image bit rate and the accuracy of the prediction algorithm is higher.