A Novel Objective Quality Assessment Method for Perceptually-Coded Cloud Gaming Video

摘要

Cloud Gaming (CG) as a viable alternative to console gaming is gaining more acceptance and growing its market share in the gaming industry. In CG, the game events are processed in the cloud and the resulting scenes are streamed as a video sequence to players. In this new paradigm, one of the most important factors that has a significant impact on user quality of experience is video quality. To address the inherent high bandwidth requirement of CG, game videos should be compressed. This compression may have a negative impact on the user's quality of experience (QoE) and the assessment of this impact on user satisfaction is a challenging task. Over the years, many research works have investigated the objective and subjective quality of video, but none are directly suitable for the assessment of perceptual video quality in the context of CG. Other methods, such as eye-tracking weighted peak signal-to-noise ratio (EWPSNR) that may work in this context, require an eye-tracking device that is not always available. In this paper, we propose a new weighted PSNR objective quality method that does not require any eye-tracker or information from the game designer (such as the importance of objects in the game) to measure game video quality. Our evaluation based on 3 actual games show that our proposed method has 51% and 11% better correlation with the Mean Opinion Score (MOS) compared to PSNR and SSIM measures, respectively.