Implementation For H.264Vedio Coding Based On GPU

In recent years, video application has already been more and more popular. With the development of the processors, display devices and network, high-resolution video is becoming a trend in video. With the growth of the image size, the computational complexity of video compression is growing rapidly in order to pursue higher video compression performance. From H.264to the HEVC, the computational complexity increases2-4times. It’s very difficult for us to achieve the real-time encoding for HD video based on the exiting platform. And this has restricted the development of the business of the HD video. Therefore, the compression and transmission of HD video is becoming the focus of research. On the other hand, with the development of the GPU, parallel computing is becoming popular for its low costs, flexible programming and powerful ability in parallel processing. Thence, accelerating the video encoding based on GPU has an extensive application future.In this paper, we choose the mainstream H.264video standard as the target to optimize based on the GPU platform. And the results of this paper are as follows:1. In this paper, we design and improve the H.264encoding framework based on CPU+GPU. This can reduce the data transfer times between CPU and GPU to improve the efficiency of the program.2. In this paper, we design and improve the H.264intra prediction encoding parallel algorithms to reduce the prediction encoding time.3. In this paper, we design and improve the entropy coding parallel algorithm. This algorithm use the properties of the memory and the thread of the GPU improve the efficiency of the entropy coding by processing the LumaAC, LumaDC, ChromaDC and ChromaAC independently.The experimental results show that the integral encoding time is4-5times faster than x264. And this algorithms increase the speed of the HD video encoding greatly.