Analysis, Improvement And Application Of Web Conferencing Video Quality Assessment

Today, IP-based streaming service has penetrated into various fields of people’s lives, in which video conferencing with its intuitive, vivid, high efficiency characteristics have been widely used. However, IP network is a structure of heterogeneous and time-varying accompanied by lack of quality of service control. Video conferencing systems running on it often undergo video impairments. Traditional video quality assessment is a highly subjective and time-consuming process, making it hard to be used widely. By video quality modeling, we introduce an algorithm evaluating video quality based on its coding features and network quality.1. The article analyze the existing video conferencing technologies, then study the evaluation system of streaming video, factors affecting the streaming video quality and its corresponding methods of measurement.2. We analyze the features sensitive to video impairments, and abstract these features to spatial gradients, luminance, contrast and absolute temporal gradient characteristics of video. Then, based on these features, the video quality is described as a characteristic value gain or offset. Subsequently, the HVS simulation function introduced characteristics of the human eye. Finally, according to linear fit of particular concerned features of video conferencing systems, we obtain HVS based video conferencing video quality assessment methods.3. We simulate video conference based on H.264 encoding using PlanetLab network. The simulation results show that, parameter model has a large deviation from actual score, which cannot manifest the characteristics of video quality. With comprehensive loss index model, the relationship between network performance and video quality can be determined by the parameter model in probability distribution way, but is of poor accuracy.4. Temporal and spatial reference system are modeled separately from H264 encoding characteristics. By modeling the network degradation using the error value as a measure of the video quality impairment, deviation model which reflect the influence on video conference video image caused by package loss is obtained. Finally, according to comparison with the subjective method, we verify the effectiveness of this method.