Font Size: a A A

Rate-distortion Optimization For Video Coding Under Constrained Computational Complexity

Posted on:2018-12-16Degree:MasterType:Thesis
Country:ChinaCandidate:J K FengFull Text:PDF
GTID:2348330518998549Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
As an effective method for balancing the coding rate and the coding distortion,rate-distortion optimization has been widely applied in the video coding.Conventionally,the rate-distortion optimization in the video coding does not take into account the coding complexity.However,due to the diversity of video applications,the video encoders in different application scenarios may have different requirements of or limitation on the computational complexity.At the same time,the temporal complexity,similar to the coding rate,is another important coding resource in the application scenarios where computational complexity is limited.Therefore,the rate-distortion optimization under the constrained computational complexity must take into account the coding rate,coding distortion and coding complexity,equally.In addition,due to the diversity of the video content,different coding units tend to have different rate-distortion performance gains under the same complexity increment.Hence,a better solution is to allocate the limited coding complexity differently to each coding unit in conjunction with the characteristics of their video content,and improve the overall rate-distortion performance.In this paper,a novel rate-distortion optimization scheme under constrained computational complexity is proposed for the latest H.265/HEVC standard.In order to reduce the complexity of coding and meet the requirements of fast coding for the complexity-limited applications,a fast prediction algorithm for coding unit(CU)partitioning mode based on the estimation of rate-distortion cost is proposed in this thesis.The proposed algorithm is implemented in two steps: the first step is to rapidly estimate the rate-distortion cost based on a pre-search process;And then use the estimated value to predict the CU partitioning mode,rapidly.The experimental results demonstrate that the CU partitioning mode predicted in this thesis is highly consistent with the result of the real rate-distortion optimization process.Compared to x265,the proposed algorithm can reduce,on average,the BD-rate by 12.7% at a similar coding speed,and reduce the BD-rate by 10.9% averagely under the same coding configurations.In order to make full use of the given coding complexity,and improve the overall coding performance,this thesis analyzes the relationship among the coding bitrate,coding distortion and coding complexity from the perspective of rate distortion theory,and defines the RD-C characteristics of coding tree unit(CTU).The diversity of the video content makes different CTU have different RD-C characteristics.In this thesis,the estimation of rate-distortion cost and the CU partitioning candidates are utilized to analyze the RD-C characteristics of each CTU.Based on such RD-C characteristics,a CTU-level complexity allocation algorithm is further proposed.This complexity allocation algorithm can allocate the limited coding complexity to each CTU properly,where the characteristics of its coding content is joined.The experimental results show that the differential allocation of complexity based on the RD-C characteristics can achieve 4.4% BD-rate reduction on average,compared with the uniform distribution of complexity at similar coding speed.Compared to the original algorithm in x265,the proposed algorithm can effectively improve the coding performance under the given coding complexity while controlling the coding complexity more accurately.
Keywords/Search Tags:H.265/HEVC, Rate-distortion optimization, CU partitioning mode, Complexity control
PDF Full Text Request
Related items