摘要
为了有效地评价图像质量,该文提出一种应用人眼视觉特性的全参考图像质量评价方法。该方法主要考察了人眼的两个视觉特性,即韦伯定律和视觉注意机制,并利用这两个特性计算对应的差异激励图和视觉显著性图,将其作为能够反映图像失真的特征图,同时考虑了观察因素的影响,最后得到了失真图像的质量评价指标。实验结果表明,该方法在LIVE、CSIQ和LIVEMD三个图像库上有很好的表现。三个图像库的加权平均结果显示,本文方法的表现优于所有对比方法,包括近期提出的GMSD和VSI方法,说明本文方法的评价结果与主观感知不仅具有更好的一致性,而且具有很好的通用性和鲁棒性。
To assess image quality effectively,a new full-reference image quality assessment method sing human visual characteristics is proposed.This method mainly considers two characteristics of hum,visual system,i.e.,Weber＇s law and visual attention strategy,and the obtained differential excitati map and visual saliency map are used as indictors of image distortions to get the final image quality men which also incorporates the viewing conditions.The experiment results show that the proposed meth owns an excellent performance on three databases including LIVE,CSIQ and LIVEMD.The weighted a erage results indicate that the proposed method outperforms other methods,including the latest GMSD ai VSI,which means that the new algorithm occupies a better consistency with subjective evaluation and h good universality and robustness.
出处
《中国电子科学研究院学报》
北大核心
2015年第6期567-573,597,共8页
Journal of China Academy of Electronics and Information Technology
关键词
图像质量评价
人眼视觉特性
韦伯定律
差异激励
视觉显著性
Image quality assessment
Human visual characteristics
Weber＇s Law
Differential Excitation
Visual Saliency
