Research On The High Efficient Video Decoding Techniques Based On GPU

With the extensive application of high-definition video, The rapid increase of the video data has caused tremendous pressure to the computer central processor unit(CPU), the method that increasing the CPU clock frequency to solve the problem has not obvious effect or not. Thus that making full use of the computer’s graphics processor unit(GPU) to accelerate the video compression process has become a trend to reduce the computational burden of CPU.The main work of this paper is as follows:The paper analysis deeply the various of video compression standards, extracts common codec modules from them, and analysis the principles and methods of the modules; the paper brief analaysi the processing principles of GPU and numerical calculation techniques, and introduce the architecture of video compression on GPU and numerical calculation model.This paper uses Microsoft directx video acceleration based on GPU technology, design the architecture of high-definition video decoder, data structure and the call interface, achieve a high-definition H.264video decoder based on DXVA2.0. Then by controlling the use of resources in the decoder the author performance the H.264multi-channel decoder based on DXVA2.0. The experimental tests show that the H.264multi-channel decoder has better performance than traditional software decoder in the decoding speed and CPU occupancy rate.The paper unify the video post-processing module and video decoder module and get the very good video playback.This paper apply the multi-channel H.264decoder to the video recording and broadcasting system, used as multiple video decoder in monitoring center and the user terminal, The user can watch the hign-definition video playing and don’t increase the cost.