Research On Prediction And Filtering Algorithm Of Intra Video Coding

With the coming of the era “Big Data”,multimedia technology has taken unprecedented opportunities for development.Like the 3D video,VR video and interactive live system,many type of video technology has becoming a part of our life and brought great convenience for us.In order to meet the growing demand of the HD or UHD video resources of people,a bigger transmission bandwidth and memory resources are needed.Therefore,the next generation video coding standard H.266 has been put forward to minimize the bitrate for the same quality video sequences.The joint video experts team are mainly responsible for the research and development of H.266.Like the latest video coding standard HEVC,H.266 still remain the traditional hybrid coding structure based on block,and it propose a lot of new technology to reach the goal that the coding efficiency is double than HEVC.However,the existing coding technology is far away from the goal,further research are needed to improve the coding efficiency and reach a higher coding performance based on H.266.For the research of H.266,we mainly focus on the technology of intra prediction,transform and intra filter and propose two new methods: a new intra prediction method based on consistent luminance changes,a new intra filter method based on iterative updating.Together with the adaptive multiple transform technology,then we propose the corresponding optimization algorithm.The main academic contributions of this article are as follows:(1)As for intra prediction of H.266,we propose a new intra prediction method based on luminance changes.First,we use the neighbouring reference as the template to search the best matching reference template in the nonlocal reconstruction area.Then,we build a consistent luminance compensation model to prediction the current block.The experimental results are all based on the reference software JEM-5.0.1 and it show that the average coding performance has been improved about 0.58%.Finally,for the transform step,we analyse the residual's energy of intra mode combined with the adaptive multiple transform technology and propose a new transform scheme based on mode dependency to optimize the above technology,the experimental results show that it reduces the average encoding and decoding time of 5% and 1% respectively with the negligible loss for the coding performance.(2)As for the intra filter of H.266,we propose an intra filter technology based on iterative updating.First,we introduce the optimal coding theory for intra prediction.Then,we use the iterative filter scheme to approximate the optimal coding performance,the experimental results are all based on the reference software JEM-7.0 and it show that the intra iterative filter method has improved the coding performance average 0.47%.Finally,we propose the optimization algorithm with reducing the number of iterations and skipping some mode for intra prediction based on the statistical analysis of experimental results,the experimental results show that it only has little loss for the coding performance,but it greatly reduce the complexity of encoder and decoder.