Abstract: Nowadays, the service of network video is increasing explosively. But the quality of experience (QoE) model of network video quality is not stable. The video quality may be impaired by many factors. This paper proposes QoE models for network video quality. It consists of two components: 1) the perceptual video quality model considering the impair factors related to video content as well as distortion caused by content and transmission. Next the model is built through a decision tree using a set of measured features form the network video. This proposed model can qualitatively give the grade of video quality and improve the accuracy of prediction. 2) Based on the above model, another model is proposed to give the concrete objective score of video quality. It also considers original impair factors and predicts the video quality using fuzzy decision tree. The two models have their own advantages. The first model has a good computational complexity; the second model is more precise. All the models are simulated by actual experiments. They can improve the accuracy of objective model. The detail results are shown.

Key words: perceptual video quality, QoE, objective model, fuzzy decision tree, accuracy