注意力模型指导下的视频质量评价方法  

Video Quality Assessment Method with Attention Model Guided

作　　者：王淦[1] 宋利[1] 张文军[1] 

机构地区：[1]上海交通大学上海市数字媒体处理与传输重点实验室,上海200240

出　　处：《电视技术》2014年第7期5-8,共4页Video Engineering

摘　　要：在视频质量评价方法中,常常需要对人眼视觉系统做出合理的假设,其中注意力模型就是一个很重要的因素。提出了一种在注意力模型指导下的视频质量评价方法,在图像帧的质量评价中加入了显著性区域信息,使之更能符合人眼视觉特性,并兼顾了视频中的运动信息,在一定程度上提高了客观质量评价方法的性能。Among video quality assessment methods,assumptions about human visual system are made frequently,in order to accord with human psycho visual features and attention model plays an important role. In this paper, a video quality assessment method with attention model guided is proposed,in- corporating local image frame quality assessment with saliency information,also considering motion information in the video. It is testified that the pro- posed method enhances the performance of objective quality assessment.

关 键 词：视频质量评价 注意力模型 显著性区域 运动信息 

分 类 号：TN919.8[电子电信—通信与信息系统]