Premium
Color samples for colorimetric fidelity testing in television
Author(s) -
Hisdal B.
Publication year - 1985
Publication title -
color research and application
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.393
H-Index - 62
eISSN - 1520-6378
pISSN - 0361-2317
DOI - 10.1002/col.5080100412
Subject(s) - lightness , primary color , color difference , mathematics , statistics , set (abstract data type) , species evenness , fidelity , colorimetry , computer science , estimator , artificial intelligence , color space , pattern recognition (psychology) , computer vision , enhanced data rates for gsm evolution , paleontology , telecommunications , species richness , biology , programming language , image (mathematics)
Abstract The case for testing television (TV) colorimetric reproduction performance for groups of colors is discussed. The groups consist of colors which have some characteristic in common, like strong or moderate saturation. Such groups actually contain a large number of discernible colors but these are for testing purposes represented by a set with a limited number of reference samples. In recommended methods the evaluation of the performance for the group is based on the average of the individual color differences for the samples in the set. This average is here based on the CIE 1976 color difference ΔE uv * and designated ΔE a *. The question investigated is what properties the reference samples in the set should have to make D̊E a * the best estimator of performance for the whole group. It is shown that the samples should be chosen for evenness in distribution and uniformity in lightness, preferably representative of average scene lightness. These points seem to have been overlooked for some presently recommended sets as illustrated by an example. Experimental data for TV cameras indicate that a collection of 10 color samples might be satisfactory for colorimetric acceptance testing.