Video Quality Evaluation Model Based On Human Visual System

With the development of computers and communication network, the application of video and image data has already been raised .The popularity of video service has already become an irresistible trend. However, video and image datas are subject to a wide variety of distortions during acquisition, compression, storage, transmission and reproduction, and any of which may result in a degradation of visual quality. In order to optimize application systems and provide better visual quality with less resource cost, it is desirable to measure the quality of video and image. The business of protocol monitoring can not meet the market demand, instead the video quality assessment models have already become a new hot spot. The traditional subjective evaluation depends on observer's experiences and motivations, and therefore. Among them the video quality evaluation model based on human visual system (HVS) attracts more attentions for its objectivity, independence of original reference video and suitability for human perception.Traditional evaluation models only focus on the video's absolute errors, which have low correlation with the perceptual visual quality. These models do not make use of eye impact on video evaluation, so the results often deviate from the human feelings. Aiming at the defect, this paper presents a new model, which combines human visual system with absolute errors of video.There are two key parameters in this new model--blocking artifact and noise.First of all we introduce the blocking artifact and noise, then analyze its impact on video quality and finally obtain the video quality evaluation of new algorithms based on human visual characteristics according the combination of theory and human visual characteristics. The blocking artifact evaluation calculates not only the brightness gap between adjacent blocks, but also masking and nonlinearity impact, so the results are suitable for human perception very well. Noise evaluation arithmetic extracts the spatial characteristics of the video frame by using Roberts operator and temporal continuity of video sequences to calculate perceptual evaluation for noise. It also considered the change as a result of the lens, flash video and other factors caused by the impact of discontinous.Finally, the video quality evaluation model is given by the fitting of the studies and results of the block and noise effect. Then this paper studied the users' vision feel of video effects and block noise damage and the algorithm is verified through subjective and objective test of the correctness. The results show that the algorithm can correctly evaluate the video quality level and has superior performance.