Abstract:

Making use of the temporal correlations of video frames and human visual characteristics, a new algorithm for video-dynamic evaluation is proposed, which is used to quantify motion information and evaluate video quality. Motion rate, motion area, and visual-interest-feature are integrated to calculate video-dynamic. Experimental results show that the video-dynamic metric is reliable, and is well agreed with perceived quality. It can be utilized in video quality measurement.

Key words: video quality, video smoothness, video dynamic, visual interest