Color Space Conversion via Gamut-Based Color Samples of Printer

Many types of electronic imaging devices are currently available, including color cathode ray tube (CRT) devices, ink jet printers, offset printing devices, thermal transfer printers, and all of these devices utilize devicedependent color spaces for color specification. However, device-dependent color spaces do not relate to an objective definition of color or human color perception. Therefore, CIE developed device-independent color spaces to give a quantitative measure for all colors that is not dependent on the imaging device. As a result, the production of color consistency between various devices, that is, the concept of device-independent color reproduction, has received widespread attention. Device-independent color reproduction requires a color conversion between device colors and device-independent colors, specified by the standard color space like the CIEL*a*b* color system to achieve color consistency. Color reproduction on a CRT monitor is based on an addictive mixture of three primaries, for which the color space conversion is usually performed with a 1-dimensional nonlinear mapping and matrix transformation. In contrast, color reproduction in a printer is based on a subtractive mixture of either three primaries, Cyan (C), Magenta (M), and Yellow (Y), or four with the inclusion of Black (K). However, the color stimulus generated on paper is usually quite difficult to predict when only based on amounts of ink for these primaries. In other words, because of the complicated nonlinear relationship between the device-dependent input and the device-independent output signals of a printer, it is difficult to control the CMYK color signals in an 1-dimensional nonlinear mapping followed by multiplication with a matrix. Several methods have been proposed for estimating the amounts of primary inks necessary to produce a desired color stimulus. These include an analytical method using the Neugebauer equations,1 the polynomial regression model,2–5 3-dimensional interpolation using a lookup table(LUT),6–10 and neural network methods.11–13 The analytical model involves a prediction that uses several device measurements, however, this method suffers from an inevitable discrepancy between printer outputs because the analytical methods are not accurate enough due to the many disturbance elements in real printer systems. In the polynomial regression method, a system is assumed to be a black box and the parameters are obtained from the input–output relationships. Three-dimensional interpolation creates a data table of measured color values and then interpolates this table to determine the input signal for generating a desired color output. Neural network methods model the mapping between the printer color signal and the output color stimulus values using pre-determined weighting factors. When compared to the analytical method using Neugebauer equations, the regression, neural network, and LUT conversion methods all produce a high accuracy in color conversion. In these methods, a device is regarded as an unknown static system, and its input– output relationship is modeled using input values and Color Space Conversion via Gamut-Based Color Samples of Printer