Digital video compression, storage, and transmission systems are essential components of the National Information Infrastructure (NII). These systems can introduce fundamentally different kinds of impairments than are created by traditional waveform reproduction methods. Examples of compression related impairments in video communications are error blocks, localized image smearing, and jerky motion. For digital video systems, the information content of the source video signal plays a crucial role in determining the amount of compression that is possible, and the severity of compression artifacts. Thus, in general, the user-perceived quality of a digital video transmission system is a dynamic function of both the system and the input signal. The urgent need for performance measurements that correlate with the perception of digital video transmission quality has led ITS to seek new, fundamentally different, video quality measures that are objective and perception-based. Such measures, which were chosen on the basis of their correlation with the subjective video quality assessments of human users (e.g., viewer panels), are derived from the electrical and mathematical properties of the digitized input and output signals, and thus are implementable in automated test equipment. They achieve technology independence by recognizing important perceptual attributes and subsequently predicting human reactions to imperfections in received visual information.The Figure above presents a block diagram of a video quality measurement system that is being developed by ITS. The measurement system is composed of two sub-systems -- a source instrument and a destination instrument. The source instrument attaches non-intrusively to the source video and extracts a set of source features that can be used as a reference to quantify perceptual video quality changes. The destination instrument attaches non-intrusively to the destination video and extracts an identical set of destination features.
Objective quality parameters can then be obtained by comparing the source features with the corresponding destination features. Two types of features have proven to be most useful for measuring video quality. One type measures spatial distortions in the video (e.g., blurring) and the other type measures temporal distortions in the video (e.g., jerky motion). Designers of modern video transmission systems often trade off spatial and temporal performance aspects, and some have even made these trade-offs user selectable. For instance, in low bit-rate applications like videophone, users are sometimes given the option of seeing clear pictures at very low frame rates or blurred pictures at higher frame rates. The perception-based video quality features require very few bits to represent and can thus be easily and economically transferred between the source and destination video ends, which may be separated by many thousands of miles. If a transmission system is used for interactive (two-way) communications, then video delay is another important quality attribute of the transmission system. Video delay can be computed by the ITS measurement system via a time alignment correlation process applied to a motion energy feature.
The objective parameters can be combined to produce a composite estimate of what the viewer response would be to the perceived video quality. Experiments have demonstrated excellent correlations between composite objective estimates and subjective test data. These results are technology independent in that they are applicable to an extremely wide range of video systems, including very high quality 45 Mb/s studio NTSC systems. ITS has been actively transferring this promising video quality measurement technology into the private sector and the appropriate standards bodies, such as ANSI accredited Telecommunications Working Group T1A1.5 and ITU-T Study Group 12.
