Research On Application Of Nonlinear Partial Least Squares Regression In Computer Color Matching

In the study of textile color matching problems, the traditional method by computer color matching hardly meets product’s demanding. The nonlinear partial least squares regression method is introduced in this paper, and the method is compared with the principal components analysis method. The comparison of calculated and experiment shows that the matching method in this paper is feasible and effective, which provide a new method in color matching field.First, the research purpose of computer color matching is introduced. The research development and main research direction of computer color matching are reviewed. The structure and the innovation of this dissertation are also addressed.Second, the basic knowledge of computer color matching is presented. The basic principle of colorimetry containing color space, color mixing and color analysis. The steps of computer color matching in engineering application are also briefly introduced.Third, the principal components analysis method is presented. The principle components through this method is obtained, and then the nonlinear equations among textile dye recipe and three stimulus values are modeled through curve and surface fitting method. In order to get the dye recipe, the nonlinear equations can be solved by the iterative method.Fourth, the nonlinear partial least squares regression method is presented. The relationship of and three stimulus values is obtained by using this method. Through iterative method we may reasonably get the suitable textile dye recipe in the given three stimulus values.Fifth, the comparison and analysis between principal components analysis method and nonlinear partial least squares regression method are reviewed. Experimental results indicate that the latter method is better than the former.Finally, the results are summarized and the future work under research is addressed.