Predicting Color Image Match


Document ID: 
Robert Chung, Elena Fedorovskaya, David Hunter, and Pierre Urbain


Digital, Member: 
Digital, Non-Member: 
Photo, Member: 
Photo, Non-Member: 


When assessing color printing conformity, ANSI/CGATS TR016 specifies multiple tolerances, in terms of the 95th percentile DE00, and the color characterization target. This research investigates if the 95th percentile DE00 metric applies to color image match. The experiment involves the selecting the CRPC6 dataset as the reference, altering the CRPC6 reference with known colorimetric differences, generating ICC profiles and preparing pictorial color images in the specified printing conditions, conducting psychometric experiments. The results show that (1) color image match is proportional to the magnitude of the device-based 95th percentile DE00, and (2) color image match is scene dependent when the device-based 95th percentile DE00 is between 4-5 or approaches color conformity of a printing device. Color differences between two pictorial color images of the same scene can also be quantified by image-based CRF and the 95th percentile DE00. Image-based CRF only takes colors in the scene into consideration. In this research, image-based prediction of color image match did not show significant difference than the device-based prediction. This was largely due to the fact that the entire CRPC6 gamut is uniformly shifted in the direction of -L* and -b* by specified amounts. By means of simulation, we showed how image-based 95th percentile DE00 differs under the same devicebased 95th percentile DE00. The image-based 95th percentile DE00 possesses better potential in predicting color image match than the device-based approach.

Return To Search Results

Search Again