A Printer Point Set Gamut Model Based on BP Neural Network

Abstract

In the color management system, device gamut is an important prerequisite for effective color management. Currently, device gamut is described by convex hull model. In this method, electronic standard color card is output on certain media, the chromatic values of the color blocks of the color card are measured as vertices to build the three-dimensional space of convex hull. In spite of the simplicity of the description method in manipulation, the factors that can't be ignored are as the following, the relatively representative of the color blocks on the color card and the device gamut has something to do with the factors such as device feature, as well as the surrounding light and device parameters. Near the edge of the convex hull, the device limit color points perhaps locate inside and outside the edge of the convex hull, so the convex hull model can't describe the device gamut exactly. In the research, based on mapping relation of digital image pixel value and its corresponding printed color, printer characterization model is established based on BP neural network with exactitude to a perfect certainty. Then, digital image pixel values with kinds of intervals are taken to calculate the color point sets by the printer characterization model, and the device gamut could be described by the color points sets. By visual method, both the convex hull model and the point set model are compared. The experiment shows that the point set models describe the printer gamut more precise than the convex hull model.