Subjective Impairment of Television Pictures—Considerations regarding the Summability of Impairment Units

The Impairment Unit (imp) proposed by the British Post Office is examined in the light of the statistical models, either normal or binomial, used to represent observer opinion of television picture quality. For a single impairment, the interesting result is obtained that, given a binomial distribution of the mean opinion score, the logarithm of the imp is very nearly normally distributed. The imp concept is extended to multivariate distribution models representing opinion in the presence of more than one impairment. For two coexisting impairment, the imp sum, as a measure of the overall quality, is compared to the bivariate imp as above and is found to be a useful, if rough, estimate of the latter.

Print ISSN

0361-4573

Published

1972-12

Content type

Original Research

DOI

10.5594/J08181