基于改进SSIM的HEVC率失真优化  

Rate distortion optimization based on improved SSIM for HEVC

作　　者：班俊硕 赖惠成[1] 林宪峰[1] 杨敏[1] 董九玲 

机构地区：[1]新疆大学信息科学与工程学院,新疆乌鲁木齐830046

出　　处：《电视技术》2016年第7期11-14,31,共5页Video Engineering

基　　金：国家自然科学基金项目(61561048);新疆维吾尔自治区科学基金项目(2015211C257)

摘　　要：HEVC继承了传统的客观率失真模型,主要采用均方误差、平方误差和或绝对误差和等客观评价方法作为失真测度。然而,这些方法并不能得到很好的主观视觉感知质量。为了进一步解决HEVC的计算复杂度,提出了将结构相似度(Structural Similarity,SSIM)引入到HEVC的率失真代价函数中。该方法使用失真测度均方误差对SSIM进行改进,将改进后的SSIM对率失真代价函数进行优化,求解出拉格朗日乘子,从而实现率失真性能的提升。实验结果表明,该方法与HEVC参考测试模型相比,在全I帧(All Intra,AI)配置下,率失真性能平均提高了2.6%,同时可以获得很好的编码质量,以及可忽略的编码时间增长。Objective appraisal method, such as mean square error （ MSE ） , sum of the squared errors （ SSE ） or sum of absolute difference（SAD） , is adopted by HEVC, which inherits conventional objective rate distortion model. However, these methods fail to achieve high consistency ＇with subjective visual perception. In order to reduce the computational complexity of HEVC, in this pa- per, SSIM is considered as rate distortion cost function of HEVC. MSE is introduced into SSIM, so that a new rate distortion cost function can get. This problem can be solved by Lagrangian multiplier method. Experimental results show that compared our pro- posed algorithm with HM13.0 in AI configuration, the performance of rate distortion is average improved about 2.6%, as well as getting better coding quality. In addition, no significant runtime increases in the encoding process.

关 键 词：高效视频编码(HEVC) 率失真 结构相似度 拉格朗日乘子 

分 类 号：TN919.8[电子电信—通信与信息系统]