Abstract
The aim of this research work is to establish what level of numeric colour difference between two digital images is visually acceptable by average human observers on a calibrated monitor under a fixed set of viewing conditions. A visual ordinal category method is introduced and a description of the experimental design is provided. Results based on over 4000 visual and numeric comparisons are reported. All results are analysed by statistical methods. The correlation between visual assessments and numerical assessments are found by means of the Pearson product moment coefficient. Overall colorimetry accuracy and metamerism are discussed.
A very high level of correlation is found to exist between the visual ordinal categories assigned by observers and the equivalent numeric (ΔE CMC 2:1) colour difference. Individual observers, within the group of 20 observers tested, were found to be consistent in their ordinal categories. However, some observers were consistently severe in their judgement and others significantly more lenient.