This paper concerns the problem of assessing the quality of video over a packet network. Our objective is to present a methodology that is both reasonably accurate and simple enough to support the ever-increasing applications of large-scale streaming media over a packet network for their quality monitoring. Therefore we develop an approach which can evaluate the perceptual video quality based on network statistics, e.g., packet losses, delay jitter, and some available and important application-specific parameters, such as GOP structure and packetization. In the approach an algorithm is provided to compute the total packet loss rate according to delay jitter and playback buffer size, and a model which maps the packet loss rate to the perceptual quality of video is also given. The approach provides a lightweight video quality assessing solution suitable for real-time video quality monitoring. The simulations and experiments have demonstrated its feasibility and accuracy.
