Perceptual Video Quality Model and its Application in Wireless Multimedia Communications

With the exponential growth of video traffic over wireless networked and embedded devices such as mobile phones and sensors, mechanisms are needed to predict the perceptual quality of video in real time and with low complexity, based on which networking protocols can control video quality and optimize network resources to meet the quality of experience requirements of users. This thesis is composed of three related pieces of work. In the first piece of work, an efficient and light-weight video quality prediction model through partial parsing of compressed from the H.264/AVC compressed bitstream is proposed. A set of features were introduced to reflect video content characteristics and distortions caused by compression and transmission and were obtained directly in parsing mode without decoding the pixel information in macro-blocks. Based on the features, an artificial neural network model was trained for perceptual quality prediction. In the second piece of work, a perceptual video quality prediction model is trained based on massive subjective test results. Prediction of perceptual quality is achieved through a decision tree using a set of easily calculated features from the compressed bitstream and the network. Moreover, based on the prediction model, a novel Forward Error Correction (FEC) scheme is introduced to protect video packets by taking into consideration video content characteristics, compression parameters, as well as network condition. Given a perceptual quality requirement, the error control scheme adjusts the level of protection for different components in a video stream such that the network overhead needed for transmission is minimized. In the third piece of work, a study was conducted to examine whether the previous prediction model could provide a good confidence measure in a different domain of judgments. The accuracy of judgements demonstrated the predictive validity of confidence measure with respect packet loss ratio traits. The results of this study were consistent with the previous one and the experiments suggested that brief and evaluative thin slice judgments are made relatively intuitively. Present research represents a new entry into the domain of high level judgments, such as video confidence measure by the use of our existing perception quality model