Research On Optimization Algorithm Of Intra Prediction Based On HEVC

In the past ten years,the frequency of multimedia technology has increased rapidly.In order to meet the demand for high-definition video in the Internet age,the previous video coding standards are constantly optimizing their own coding algorithms,but it is difficult to make progress in compression performance.In this case,HEVC(High Efficiency Video Coding)gradually gains the favor of a large number of video customers with its excellent coding capabilities and compression effects.In HEVC,on the one hand,a large recursive coding unit CU(Coding Unit)is introduced,including PU(Prediction Unit)and TU(Transform Unit),but it will consume more coding time when recursively dividing blocks.On the other hand,HEVC has more types of prediction modes than H.264,and more detailed prediction modes will make the predicted value more accurate,thereby improving compression performance;but traversing a large number of prediction modes during the encoding process will also increase the computational complexity and consumes more coding time.Compared with H.264,the bit rate of HEVC is expected to be reduced by 50%under the condition of ensuring the same objective quality of the video.Therefore,the algorithm is researched from the two perspectives of CU segmentation and prediction mode decision-making,simplifying unnecessary calculation processes,further shortening encoding time,improving encoding performance and ensuring video quality.the main works are:1)Firstly,give a brief introduction to the overall process of HEVC,and in-depth analysis of the specific process of intra prediction,and then understand the optimal segmentation method of CU from the CU segmentation process,and analyze the conditions for selecting the optimal prediction mode in PU.Finally,the selection of the distortion value in the rate-distortion optimization(RDO)process is measured,and select the rate-distortion optimization model to be used.2)An optimization algorithm for intra prediction is proposed from the aspects of ending CU segmentation and reducing prediction modes.On the one hand,the average absolute error(MAD)method is used to determine the texture of the CU block to terminate the segmentation of CU early.On the other hand,the angular prediction(Angular)mode is classified into 4 categories according to the general direction,and the improved gradient calculation method is used to obtain the general direction of the prediction mode in the PU,thereby reducing the number of prediction modes participating in the rough selection(RMD)process.3)Use the difference between the reference model HM and the optimization algorithm in terms of Bitrate,time,and PSNR to analyze the overall experimental results.In All?Intra mode,the optimized CU segmentation and the reduced prediction mode reduce the coding time by 28.61% and 18.66% respectively,both two methods increase only a small bit rate,improve the coding performance of HEVC.