Preliminary application research of regression analysis in camouflage color matching

Abstract

It is an important factor for affecting camouflage effects, but copying background color with empirical color matching has poor accuracy. This paper chose cadmium red, cadmium yellow and other colors basing on the performance requirements of same-color same-spectrum; prepared pastes and swatches with different grinding times particle sizes, were to measure their spectral reflection characteristics; then selected cadmium red, cadmium yellow and cobalt blue as the base pigments, which are roughly the same as the natural background-green in the infrared band spectral reflectance curve. Carried out a large number of experiments to prepare several green color samples from pale green, light green, medium green to blackish green and recorded the basic color ratio to construct a basic green pigment formula library according to the color reduction principle of color mixing. Respectively establish a mathematical model of the chromaticity values and the mass ratio of the yellow and blue base pigments, verify the model by using multiple nonlinear regression methods in SPSS, and verify the regression significant effect of the model built. Using the color-difference calculation formula, a nonlinear function model is established, and the minimum point of the function is found by MATLAB neural network genetic algorithm, and the mass ratio of the base pigment of the simulated background color is obtained. The results show that the spectral curve satisfies the requirement of the same-color same-spectrum compared to the background, which can meet the needs of camouflage color matching and provide a scheme for camouflage color matching automation and intelligent.