Use of transformed reflectance functions for neural network color match prediction systems

Abstract

Attempts have been made to use different transformed reflectance functions as input for a fixed genetically optimized neural network match prediction system. Two different sets of data depicting dyed samples of known recipes but metameric to each other were used to train and test the network. All the transformed and untransformed reflectance functions gave good

recipe predictions when trained and tested by the same data sets (PF/4 being less than 4). However, the transformation based on matrix R of the decomposition theory showed promising results, since it gave very good colorant concentration predictions when trained by the first set of data dyed with one set of colorants while being tested by a completely different

second set of data dyed with a different set of colorants (PF/4 always being less than 10).