Content Classification Based on Objective Video Quality Evaluation for MPEG4 Video Streaming over Wireless Networks.

User's perceptive video quality differs greatly with video contents hence, it is of practical importance to classify videos. Videos are most commonly classified according to their spatial and temporal features. In this paper, we classify videos into groups based on objective video quality evaluation. Video quality measured in terms of the Mean Opinion Score (MOS), obtained by the application and network level parameters is classified into groups using cluster analysis with good prediction accuracy. The QoS (Quality of Service) parameters that affect video quality considered in this paper are send bitrate, frame rate in application and packet error rate in the network level. We then find the degree of influence of each of the QoS parameter and analyze the relationship between QoS parameters and content types by using principal component analysis for streaming MPEG4 video over wireless networks. We compare our classified contents to the spatio-temporal dynamics of the content and establish the relationship between video contents and video quality by equations obtained by multiple linear regression. Finally, we apply the results to rate control methods. The proposed scheme makes it possible to apply priority control to content delivery, based on content types with similar attributes. [ABSTRACT FROM AUTHOR]