Reduced Reference Video Qoe Assessment Method Based On Image Feature Information

With the rapid development of Internet, in order to survive the intense competition and retain customers, the service providers have an urgent need for quality metrics which could help supervise and improve the services. In terms of video services, traditional service quality metrics are QoS metrics. But QoS metrics only focus on low layer network performance, which makes it inappropriate for measuring the actual user perceptual quality of video services. Therefor, recently more and more studies have turn to QoE. Not like QoS, QoE focus more on measuring the service quality from user’s view point, thus could directly represent the user opinion and also more suitable for service providers.This paper is about how to assess video QoE based on extracted image feature information. Firstly, the concept of image feature information is in-troduced. Based on the characteristic of HVS, the effectiveness of the image feature infrmation is also analysed. After that, wavelet transform and its appli-cation in reduced-reference assessment is discussed. Wavelet is a commonly used image processing method, which could help measure video distortion, which has a close relationship with MOS. In this paper, BP neural net is de-ployed for mapping the distortion metrics into MOS so that we could achieve precise video QoE assessment result.In summary, this paper begins with user perceptual characteristic, extracts feature information and then transmits using wavelet to build up the reduced-reference video QoE assessment method. The proposed method has not as-sumption about video format and has low bandwidth consumption, which makes it having wide availability and also easy to deploy.