A Spectral Model for Halftone Color Prediction

In this paper, we take a spectral model based approach to the problem of color prediction of halftones. Our approach assumes no scattering through the colorant layers and models the surface and subsurface scattering in the paper substrate by a parametric closed form point spread function. Given a halftone bitmap, we synthesize a high-resolution halftone microstructure image assuming dot profiles with Gaussian roll-offs, from which we compute a high-resolution transmission image. We then convolve this transmission image with the point spread function of the paper substrate to predict the average spectral distribution of the halftone. We use the spectral distribution measurements made by a spectrophotometer to train our model i.e. to iterate over a set of values for the two parameters of our model: the parameter for the point spread function of the paper substrate and the proportionality constant to obtain the transmittance of the colorant layer from the colorant dot count and the absorption spectrum of the colorant. Preliminary experimental results show the efficacy of our method.