Optimising the Choice of Colours of an Image Database for Dichromats

Colour appears to gradually play more and more significant role in the modern digital world. However, about eight percent of the population are protanopic and deuteranopic viewers who have difficulties in seeing red and green respectively. In this paper, we identify a correspondence between the 256 standard colours and their dichromatic versions so that the perceived difference between any pair of colours seen by people with normal vision and dichromats is minimised. Colour dissimilarity is measured using the Euclidean metric in the Lab colour space. The optimisation is performed using a randomised approach based on a greedy algorithm. A database comprising 12000 high quality images is employed for calculating frequencies of joint colour appearance used for weighting colour dissimilarity matrices.