Opinion-rating model applied to television-transmission-system studies

Abstract

The means by which a recently described opinion-rating model may be applied to the performance of television transmission systems are outlined. The starting point is the statistical description of the objective magnitude of each distinct type of impairment arising in the system. For each type, the relationship between objective magnitude and picture quality must be known or estimated. Three distinct continua are invoked for the model, each having a functional relationship with the ‘psychological scale’ of opinion regarding the picture quality. The first, the t scale, may be regarded as the direct scale of opinion level; it is used to interpret the results of laboratory experiments and to express system performance meaningfully. The second scale, the T scale, is useful because the variability of observers' standards results in a near-normal distribution on it. The third scale, the J scale, is one on which the effects of different coexisting impairments may be arithmetically summed; the scale resembles the existing I, or ‘imp’ scale, but, whereas the I scale is based on the mean opinion score, the J scale may be used to express the probability distribution of opinions as well as the average value. By way of example, the model is applied to fading interference in broadcasting, and to the measured performance of a long point-to-point transmission chain in the United Kingdom, involving 5 main links, a rebroadcasting link, and several lesser components. An interesting conclusion of the latter study is that the variation in subjective picture quality resulting from the combined effects of all the varying transmission impairments was unimportant. Although viewers could have noted that the most prominent impairment was not always the same, variations in opinion level would have been almost entirely due to the variability of the viewers themselves.