No-Reference Spatio-temporal Activity Difference PSNR Estimation

Abstract

Monitoring and maintaining acceptable Quality of Experience for end-users are main purposes for video service providers. Multicast network error transmission can critically degrade the perceived visual quality. This paper presents a novel no-reference video quality estimation using spatial and temporal perceptual activity over multicast IEEE 802.11n. This allows real-time monitoring and detection of visual degradations caused by network error transmission. The spatial and temporal perceptual activity features are extracted from each frame which subjected to network error transmission conditions. These features are implemented in the proposed method to produce video quality estimation and compared against the performance of full-reference objective video quality metrics. The proposed method is shown to be well correlated over a wide range of packet error rates and demonstrates that video quality can be detected by using spatial and temporal perceptual activity features with high accuracy which allows maintaining quality perceived by the end-users.