Research On Fast Algorithm Of HEVC Intra Prediction

HEVC is the latest video coding standard,The compression performance of HEVC compared with the previous generation of video coding h.264/AVC raised nearly doubled.But a high compression ratio will inevitably increase the computational complexity of code,so the HEVC standard in the process of the commercialisation standard faces severe challenges.Therefore,it has become a research hotspot to reduce the computational complexity on the basis of the coding performance.Frame prediction is an important link of HEVC,predicting mode selection and divising coding unit CU decide the frame forecast computational complexity.Based on the research background of the basic financial resources of the central university,this paper makes an in-depth study on the division of the mode selection and coding unit CU for intra prediction,and proposes a fast algorithm for adaptive mode selection and a fast algorithm for CU partitioning based on Discrete Cosine Transform(DCT).The specific research contents are as follows:1)On the basic principle of video coding,we have compared HEVC with h.264 / AVC from two aspects of difference and advantage,then we have analyzed the computational complexity problem of HEVC,although the HEVC have the characteristics of flexible,reliable and high compression rate.2)With researching the principle of frame prediction mode selection and analying the cause of the high computational complexity mode selection process,this paper proposes a adaptive frame prediction mode decision algorithm.Then we have simulated the algorithm,found that the algorithm can save 22% encoding time averagely in the code rate and SNR almost invariant.3)We have analyzed the relations between DCT transform and image texture complexity,and accorded to the principle of statistical threshold offline training,then we have presented the algorithm base on the two aspects of coding unit CU dividing and pattern choice.Then we have simulated the algorithm,found that the algorithm can save 27% encoding time in the code rate and SNR almost invariant.4)We have simulated jointly two algorithm that frame prediction mode selection algorithm and CU fast classification algorithm,then demonstrated frame prediction encoding time being affected by the frame prediction of two key processes-mode selection and coding unit of the separation of CU.Experimental simulation tests show that the coding rate is reduced by an average of 44% when the bit rate increase is not significant and the signal-to-noise ratio is substantially constant.