Research On Interframe Fast Coding Algorithm For HEVC Screen Content

The latest generation of High Efficiency Video Coding(HEVC)has doubled its compression performance based on the previous generation H.264/AVC standard.HEVC's excellent coding performance comes at the cost of increased coding complexity,while too high coding complexity greatly limits the real-time application of video.Screen Content Coding(SCC)as an extension of HEVC,also has high computational complexity.In order to reduce the computational complexity of SCC,this paper proposes a fast algorithm for coding unit(CU)partitioning and prediction unit(PU)mode selection process,and a fast algorithm for interframe motion estimation based on screen content attributes.Firstly,in the CU depth decision fast algorithm,using the current CU time domain correlation and the spatial domain correlation,the depth information of the neighboring CU and the average rate distortion cost of its neighboring CU are selected to predict whether the current CU needs to be divided.Thus,early decisions are made on the CU partition to skip the partial CU deep traversal process.Secondly,in the early termination fast algorithm of the PU mode,the PU mode selection process is terminated early according to the best PU mode in the coded CU,which is the rate distortion cost average of the Merge mode,and finally the computational complexity of PU mode selection process is reduced.The final experiment shows that the proposed algorithm saves an average of 35.4% coding time compared to the HM16.7+SCM6.0 standard encoder.The average(35)BR is only increased by 1.59%,and the average(35)PSNR is only reduced 0.05 dB,effectively reducing the coding complexity.This paper also proposes a fast algorithm for interframe motion estimation based on screen content attributes.First,the CUs in the screen content are divided into a natural image CU and a screen content CU according to the image gradient and pixel value difference of the encoded CU.For the natural image CU,the mean square error is calculated based on its motion vector information,and then the motion search range is adjusted using the mean square error magnitude.For the screen content CU,if the optimal division depth of its parent CU is 0 or 1,its motion search range is reduced to 8?8 size to accelerate the process of motion estimation.The final experimental results show that compared with the HM16.7+SCM6.0 standard encoder,the algorithm saves 10.4% of the coding time in the low delay(LD)mode,and the average(35)BR is only increased by increased 0.9%,the average(35)PSNR is only reduced by 0.04dB;in the random access(RA)mode,the average saving time is 9.7%,and the average(35)BR is only increased by 0.86%,the average(35)PSNR is only reduced by 0.03 dB.In both configuration modes,the algorithm effectively reduces the coding complexity.