For a sustainable adoption of IP-based OTTvideo services subscriber Quality-of-Experience (QoE) isking. This has resulted in inquiries like “How do youmeasure streaming video quality?” Not a trivial question,indeed! In fact existing standardized video qualitymeasurements barely suffice when considering OTTstreaming applications. Fundamentally, all traditionalobjective testing standards are based on short videosequences, of a few seconds in length. Addressing adaptivestreaming services, however, quality significantly dependson transmission artifacts resulting from network congestionand the client’s behavior in response to it, i.e. the streamingprotocol, re-buffering and bitrate adaption, in a long-termcontext.In an attempt to propose a paradigm shift for nextgenerationperceptual video testing standards towardsstreaming video services, OPTICOM devises PEVQ-S, anadvanced framework algorithm for full-reference picturequality analysis in streaming environments. This paper willinitially summarize existing state-of-the-art video qualitytesting standards, developed by VQEG and the ITU, andanalyze their limitations with regard to being deployed forOTT video services. Based on a fundamental requirementanalysis to understand adaptive streaming artifacts, thefoundation for the novel test method will then be outlined. Inparticular, it will be shown that the architecture of the novelapproach will be suited to overcome the limitations ofstandardized perceptual video metrics with regard toadaptive streaming of longer video sequences, whilemaintaining maximum backward compatibility (and thusaccuracy) with ITU-T J.247 for short term analysis.
展开▼
