传统编码器虽然具有可控的码率,但却无法有效控制编码视频的质量,存在随图像内容变化而产生抖动的缺陷.针对当前互联网带宽特性,如自适应码率(Adaptive Bitrate,ABR)网络带宽控制技术中带宽固定的限制条件比传统广播网松弛,在以失真为约束的条件下,提出了一种新的率失真优化的失真分配方案,根据每个编码单元的拉格朗日乘子与图像组(Group of Pictures,GOP)级别的乘子之间的相互关系模型,设计了以帧级为单元的失真分配策略.基于高效率视频编码(High Efficiency Video Coding,HEVC)模型随机编码结构的默认配置下,对通用测试条件中规定的标准测试序列,实验结果显示质量一致性限制的编码器率失真性能Bjøntegaard Delta-Peak Signal to Noise Rate(BD-PSNR)提升了0.057 dB,编码后的图像组失真的方差减小了 50％,能有效地减少编码视频的质量抖动,具有更加平稳的编码质量.
Coding Model Design with Quality Consistency Constraints
Although the bit rates of traditional encoders are controllable,the quality of the encoded video cannot be effectively controlled,and jitters may be produced as the image content changes.Considering the current Internet bandwidth characteristics,such as the fixed bandwidth restriction in Adaptive Bitrate(ABR)adaptive network bandwidth control technology that is looser than traditional broadcast networks,a new rate-distortion optimized distortion allocation scheme is proposed under the condition of distortion as a constraint.Based on the relationship model between the Lagrange multipliers of each coding unit and the multipliers at the Group of Pictures(GOP)level,a frame-level distortion allocation strategy is designed.Based on the default configuration of the random encoding structure of High Efficiency Video Coding(HEVC)model,the sequence is tested using the criteria specified in the general test conditions.The results show that the encoder rate-distortion performance Bjøntegaard Delta-Peak Signal to Noise Rate(BD-PSNR)under the quality consistency limit is improved by 0.057 dB,and the variance of the encoded image group distortion is reduced by 50％,which can effectively reduce the quality jitter of the encoded video and provide a more stable encoding quality.
video codingrate-distortion optimizationdistortion allocationdistortion controlencoder design
万方数据
维普
