Optimization Of Entropy Coding In HEVC With CUDA

There are more and more high-definition video and ultra high definition videoin peoples’ life. The compression efficiency of H.264/AVC can’t adapt to thebandwidth of the network. JCT-VC was established to formulate the new video codingstandard-HEVC. The compression efficiency of HEVC aims to double the efficiencyof H.264/AVC. But the improvement of compression sacrifices the complexity ofcomputation. Because of the high complexity of HEVC codec, the compression speedis slow. So the optimization of HEVC codec is essential. The paper optimizes theHEVC codec based on CUDA.The paper studied the feature of HEVC codec. HEVC uses the WPP technologyand the entropy coding can be paralleled. The paper optimizes the HEVC codec inCPU and CUDA based on this feature. The paper studied the key features of HEVCcoding algorithm and compared it with H.264/AVC, such as larger coding block, moreprediction directions and so on. And modified the framework of HEVC encoder tomake it can operate parallelly. In addition, the paper introduce the advantage ofCUDA in parallel computation. And the paper use CUDA to optimize the modulewhich needs heavy computation and takes long time in HEVC, such as DCTtransform and entropy coding.The experimental result demonstrates: modifying HEVC to parallel frameworkcan decrease the encoding time and improve the encoding efficiency. And theoptimized DCT transform algorithm and entropy coding algorithm can improveefficiency significantly. The optimization is useful in practical application of HEVC.