Research On HEVC Low Complexity Algorithm And Rate Control Algorithm

With the development of aerospace technology in China,the number of satellite services related to video is constantly increasing in application fields.At the same time,higher requirements are also put forward for video resolution,frame rate,and other aspects.The transmission of these high-quality videos requires a large amount of bandwidth resources,but the bandwidth resources on satellite are limited,so on satellite videos require a video encoding standard with a higher compression ratio.Compared to H.264/AVC,H.265/HEVC can reduce the bit rate by about half with the same reconstructed video quality.The high compression ratio depends on the introduction of many new coding tools,but also brings a huge amount of computation,which limits the real-time application of HEVC.In addition,in the actual application scenario,the bit rate used for video transmission fluctuates up and down.The goal of the ratecontrol algorithm is to select the appropriate encoding parameters so that the compressed stream can be transmitted smoothly under the target bandwidth and the distortion is as small as possible.Therefore,a better rate control algorithm can improve the practicability of HEVC.In view of the above two problems,based on the HEVC coding standard,this thesis makes a detailed study of the coding unit division process with high complexity and rate control process in HEVC.On this basis,a fast division algorithm of inter prediction coding unit is proposed,and the rate control algorithm in HEVC is improved.The main work of this thesis is as follows:(1)A fast division algorithm of inter prediction coding unit is proposed.The relationship between the Binary Sum of Absolute Difference(BSAD)of co-located Coding Tree Unit(CTU)of adjacent frames and the de pth of coding unit division is proved by experiments.At the same time,the relationship between the coding unit division depth of the current CTU and the coding unit division depth of the spatio-temporal adjacent CTU is proved by experiments.Then,the traversal depth range in the current CTU partition process is trimmed by combining these two relationships.experimental results show that the proposed algorithm can increase the coding speed by 29.40% on average,while the bit rate is only 1.3% higher.(2)A LCU layer code rate control algorithm based on visual saliance detection is designed.A new bit allocation weight for Largest Coding Unit(LCU)layer was designed,which considered both the visual saliency of the current LCU and the sum of the absolute values of the Discrete Cosine Transform(DCT)coefficients of the residual information of the reference LCU.At the same time,a new model parameter updating method is proposed based on the real required bit number,distortion and different rate-distortion characteristics between different frames after the current LCU encoding is completed.The experimental results show that compared with the rate control algorithm in HM-10.1,the proposed algorithm can improve the rate distortion performance by 2.89% on average,and the rate control error is reduced from 1.083% to0.709%.In addition,the region with higher visual significance in the reconstructed video also gives people a better viewing experience.