Capital Currents - Video quality assessment
The FCC regulating video picture quality is a bad idea.
The FCC has a proceeding underway to update its cable TV technical regulations. For the most part, there is agreement that the version of SCTE 40 adopted last year contains the appropriate technical specifications for digital cable systems. But the FCC also asked whether it should regulate the picture quality that is delivered to viewers. The FCC claimed that viewers may experience “pixelization” or “tearing” of a video image, or other picture quality impairments, and asked for information on the status of objective picture quality assessment. The bottom line is that there is no accepted, standard way to objectively measure HDTV picture quality, but there is a lot of study underway.
Subjective picture quality assessment has been done for years, and the procedure is well known. You convene a group of viewers under very carefully controlled viewing conditions, play numerous video clips for them (both unimpaired reference copies and the processed or compressed video that is being assessed), have them record their subjective assessment on a scale of 1 to 5, and compile their results.
But that’s not what the FCC wanted to do; they wanted a way to take the digital bit stream and measure its picture quality without subjective viewing.
There are several standards bodies with their fingers in the video quality pie. They include ITU-T, ITU-R, the telephone industry’s Alliance for Telecommunications Solutions (ATIS) and the Video Quality Experts Group (VQEG). Within ITU-T, there is work in both Study Group 9 and Study Group 12. Within ITU-R, the work is in Study Group 6. Within ATIS, there is the Network Performance, Reliability and Quality of Service Committee (PRQC), as well as the IPTV Interoperability Forum (IIF). The VQEG is the leading video quality research group; it does not publish standards, but it develops test plans and conducts tests and reports its results to the ITU standards bodies.
In recent years, VQEG worked on the assessment of objective video quality models that were created to predict the quality of HDTV. The test plans examined Full Reference (FR), Reduced Reference (RR) and No Reference (NR) objective video quality models. FR models compare the impaired or coded picture with an unimpaired reference copy of the source video. RR models use an extract of the source video for comparison, which can be delivered to the receive location over a low-bandwidth channel. NR models do not have access to the source video at all.
FR models would be used for lab testing – for example, to compare the effectiveness of competing compression algorithms.
RR models use a reduced bit rate copy of the source video that consists, for example, of image edge information that is extracted from the unimpaired reference copy. This is based on the claim that the human visual system is sensitive to degradation around the edges. In other words, when the edge pixels of a video are blurred, evaluators tend to give low scores to the video. But it does require a low data rate channel to transmit the image edge information if testing is to be done at a subscriber location.
But in general, the NR approach is the one that would be needed to test picture quality at a subscriber location.
Candidate models were submitted in 2009 and were tested by comparing the model’s assessment with the subjective result from a group of viewers. VQEG's final report was approved June 30, 2010. It essentially picks one “winner” from the candidate FR models and one “winner” from the candidate RR models. All of the candidate NR models were evidently unsuccessful in replicating subjective assessments because they were all withdrawn.
The winning FR model was adopted by ITU-T as Recommendation J.341: objective perceptual multimedia video quality measurement of HDTV for digital cable television in the presence of a full reference. It is based on a model developed by SwissQual, and any commercial use requires a license from SwissQual and a non-disclosure agreement.
The RR model was adopted by ITU-T as Recommendation J.342: objective multimedia video quality measurement of HDTV for digital cable television in the presence of a reduced reference signal. There do not seem to be any commercial implementations of J.342.
So there is no accepted way for objective assessment of the picture quality at a subscriber location. And, anyway, picture quality depends on the viewer’s visual and aural acuity, viewing distance from the display, and the quality of the display. There are no industry standards for HDTV display quality. Moreover, picture quality problems can be caused by the satellite link and are outside the control of the cable operator.
So the cable industry submitted comments to the FCC explaining why FCC regulation of video picture quality was a bad idea. But the National Association of Telecommunications Officers and Advisors (NATOA), representing franchising authorities, took the opposite position. They said, “The Commission should require the use of consumer perception measures to assess signal quality.” That’s just what we need: locally convened panels of consumers in each franchise area to decide whether picture quality is satisfactory.