Research On Adaptive Intra Prediction Algorithm For High Efficiency Video Coding Based On Video Features Analysis

With the high speed development of science and technology,the way people access to information has become more convenient and faster than ever before.And the video has become one of the main means for people to obtain information based on its outstanding performance such as vivid viewing effect and high efficiency on information transfer.On the other hand,the progress and development of the technology of video capture and image display has promoted the application of High Definition(HD)video or Ultra High Definition(UHD)video,which also create a huge pressure on video transmission and storage.In this context,a new video coding standard namely High Efficiency Coding(HEVC)has been developed.Compared to the last generation of video coding standard,HEVC can save about 50% bit rate while achieve the same video quality by adding lots of advanced video coding technology.However,the highly efficient coding compression has led to the computational complexity dramatically increased,which seriously hinders the popularization of various real-time video applications based on wireless mobile communication.Therefore,it become one of the hot topics in the research of video coding standards that how to effectively reduce the complexity of coding time on the premise of ensuring the quality of video coding.As one of the core technologies of video coding,intra prediction is the initial part of video coding,which plays an important role in the whole process of video coding.Thus,the fast coding algorithm of HEVC intra coding becomes an important issue in the field of video coding.Firstly,this paper briefly introduces the development of video coding standards and the overall technical framework,and makes a brief introduction to the encoding and decoding framework and key technologies of HEVC video codec standard.Then,the intra prediction process of HEVC is introduced in detail,and the reasons for the increase of coding time complexity are analyzed in detail.At last,this paper makes a comparative analysis and systematic summary of the current research on the fast algorithm of intra prediction in HEVC.In this paper,a fast adaptive decision algorithm based on video features is proposed for HEVC intra prediction.Based on the encoding structure of HEVC intra prediction and the video features,the proposed algorithm extracts a variety of image features and encoding parameters as the initial variables to establish the complexity classification model of the coding unit(CU)and predition unit(PU)by using machine learning.If a CU/PU is judged as simple image block,its partition process will be early terminated.On the contrary,if a CU/PU is judged as complex image block,its mode selection process on PU will be skipped.The proposed dalgorithm effectively judge the complexity of the coding block,which can reduce the unnecessary intra prediction process and achieve remarkable results in reducing the coding time complexity.The experimental results show that the proposed algorithm could achieve around 60% encoding time reduction for various test video sequences on average,while maintaining the encoding performance efficiently with only 1.62% Bjontegaard distortion-rate increase compared to the test model HM15.0 of HEVC.By making full use of the video information and the HEVC intra coding structure information,the proposed algorithm makes an accurate judgment for the image blocks complexity of HEVC intra prediction process based on machine learning,which effectively optimize the quadtree partition frame and finally achieve efficiently encoding time saving.Thus,the proposed algorithm has a certain reference value for the research and video coding and the application of real-time video based on vireless communication network.