摘要
结合帧内模式选择主客观评价方法,提出了一种新的帧内模式选择算法,提高了主观质量,同时兼顾了算法处理效率。传统AVS帧内模式选择率失真优化使用绝对误差和(SAD)作为失真度度量,方法简单,但不能很好符合人眼视觉系统(HVS);基于梯度幅度值的结构相似度图像质量评价方法(GSIM)符合HVS的特性,且能评价严重模糊的降质图像,但计算复杂,不利于实时编码。本算法结合SAD和GSIM的优点,采用两者的结合作为失真度度量,并利用预测模式间的相关性进行帧内预测模式选择。取SAD的最优值和次优值,计算其差值,根据差值和阈值的比较判断是否需要GSIM计算,其中阈值的选择根据预测模式间相关性自适应确定。需要计算GSIM值时,根据GSIM的最优值和次优值的情况判断是否需要对帧内模式进行修正。实验结果表明,本算法较传统算法有较好的主观质量,编码时间增加很少。
The rate distortion optimization for AVS intra mode decision uses the sum absolute difference （SAD） as the distortion metric. The method is simple, but is not consistent with human vision system （HVS） quite well. The structural similarity （SSIM） proposed recently accords with HVS much more well, but it has some deficiencies in assessing badly blurred images. The model of gradient-magnitude-based structure similarity （GSIM） can solve the problem. The rate distortion optimization using GSIM can be consistent with HVS quite well, but it is complex and not conducive to real-time coding. The paper uses the combination of SAD and GSIM as the distortion metric, calculates the difference of optimal value and suboptimal value for SAD, compare the difference value with the threshold value to determine whether the calculation of GSIM is needed, in where the threshold is adaptive-adjusted based on the intra mode correlation. If GSIM value needs to be calculated, the GSIM optimal values and suboptimal values determine whether correction for intra mode is needed. Experimental results show that the subjective quality of the proposed algorithm is better than that of traditional algorithms, with little increase in encoding time.
出处
《太原理工大学学报》
CAS
北大核心
2014年第3期376-379,共4页
Journal of Taiyuan University of Technology
基金
山西省自然科学基金资助项目(2010011034-1)
