Models for the prediction of color in graphic reproduction technology

Abstract

A general method which permits the prediction of color in graphic reproduction is disclosed. This method involves generalizing models used to predict density in black and white printing. The models used in black and white are applied on a point-by-point basis throughout the spectrum to provide the prediction under the new method. The precision of the new method is compared to the precision of the existing method. To evaluate the new method, two models used in black and white printing are generalized to permit the prediction of color. These models are for predicting color in monochrome printing for both continuous tone and halftone processes. A third model, which is specifically for the prediction of color, was recast using the new method, so that its accuracy could be improved. The two black and white models, when generalized, were found to offer adequate precision. Typically, a color error on the order of a just noticeable difference (for human observers) was noted for these two models. The third model did not offer sufficient precision. The color errors for the predictions of the multicolor model tended to be approximately five times this amount. The question of halftone dot overlap is also answered experimentally. The model used in the prior literature was found to offer a high degree of precision. In addition to being evaluated for absolute precision, the models were evaluated for their precision relative to their analogs contained in the prior art. A relative reduction of 25 percent color error was realized when the new approach was used.