Design And Algorithm Optimi-Zation Of AVS2 Video Player Based On FFmpeg On Android Platform

With the rapid development of mobile communication technology and the popularization of hand-held mobile terminal equipment,the medium of information dissemination and acquisition is transferred from text to video,people's demand for HD and UHD video is increasing.AVS2 is a codec standard with independent intellectual property rights in China.Its main application target is ultra-high definition video,which supports efficient compression of 4K and high dynamic range video.At present,there are few video applications that can support AVS2 standard in mobile terminals on the market.There are two common problems in video applications that support AVS2:First,because the hardware performance is generally weaker than that of PC,the computation of hand-held mobile terminal in the process of decoding and playing HD and UHD AVS2 video is so large that it affects the real-time decoding and playing;Second,the unreliability of the communication channel will lead to the loss or damage of some video data.International codec standards such as H.264/AVC and H.265/HEVC have more research and application of error concealment algorithm in the decoder,but AVS2 has little research in this aspect.To solve these problems,this paper designed an AVS2 video player based on FFmpeg on Android platform,and optimizes IDCT algorithm and error concealment algorithm in decoding algorithm of the player.In this paper,there are three innovations in the optimization of AVS2 video player decoding algorithm:First,the IDCT optimization algorithm is proposed for TU with a few high frequency components and all 0 high frequency components in AVS2 decoding process,which reduces the computation of IDCT algorithm and improves the decoding speed.Secondly,LCU error concealment block partition algorithm is optimized.According to the characteristics of block partition in AVS2,MVEP algorithm is used to replace MVC algorithm,which improves the effect of error concealment.Thirdly,the operation sequence of error concealment of the remaining PU is optimized,from raster scanning sequence to determining the order of error concealment of PU according to the weights of adjacent blocks,so as to improve the overall error concealment effect.In the process of completing this project,the main works and achievements of the project include the following four items:First,FFmpeg is modified and clipped,and the core layer library and underlying code are cross compiled and ported through NDK.Secondly,according to the analysis of Android platform architecture and player function structure,we design three-layer structure and divide each layer structure into function modules,and implement it according to the design.Thirdly,the proposed IDCT optimization algorithm is implemented in the engineering code,and the experimental results show that the decoding speed can be increased by 6%?9%for various resolutions and contents of video.Fourthly,the error concealment optimization algorithm is implemented and tested in RD17.0 code.The results show that the PSNR of the algorithm is 0.19?1.75db higher than the original algorithm.