The research of spectrophotometric color matching based on multi-peaks Gaussian fit

Abstract

Spectrophotometric color matching is an important method for computer color matching, which is more accurate but difficult than tri-stimulus values color matching, because which will result in metamerism. The fundamental theory of computer color matching is the linear relationship between Kubelka-Munk function and concentration of dye. In fact, the spectral reflectivity of every pixel in hyperspectral image composed of subpixel mixing in instantaneous field of view. According to the Glassman laws of color mixing, the mixed pixel’s spectral reflectivity equals to the algebra sum of each reflectivity of subpixel multiply its area percentage. In this case, spectrophotometric color matching match the spectral reflectivity curve by adjusting the combined form of subpixel which constitute the pixel. According to numerical methods for Multi-peaks Guassian fitting, the spectral reflectivity curve can be fit as the sum of several characteristic peak, which accord with Normal Distribution. Then the spectrophotometric color matching can simplify the solution with infinite wavelength into solving the linear equations with finite known peak intensity. By using Imaging Spectrometer measure the color samples in standard color cards from different distance, the spectral reflectivity curve of each single color sample and the mixed color samples can be gotten, and the experiments results show that the spectrophotometric color matching based on Multi-peaks Gaussian fitting is superior to the tri-stimulus values color matching, and which is easy to operate.