The optimization of color-prediction models for colored cotton fiber yarns

Color matching is necessary in the manufacture of colored fiber yarns, and its accuracy is one of the main goals in computer-aided color matching. A limited number of pre-colored fibers are blended to match the target color through predicted recipes. Of the color-prediction models applied in computer-aided color matching, two common ones, Stearns-Noechel and Friele, were selected to be modified to improve the color-prediction accuracy in this paper. The models were modified in three ways, in which the pending parameters in the models were determined through statistical analysis depending on median, wavelength, and components; thus, the Stearns-Noechel model was modified to Stearns-Noechel 1, Stearns-Noechel 2, and Stearns-Noechel 3, and the Friele model was modified to Friele 1, Friele 2, and Friele 3. The six modified models were affirmed through 261 colored fiber yarns prepared from five primary cotton fibers, including two-, three-, four-, and five-primary blends. The prediction results of 261 samples showed that Stearns-Noechel 3 had the highest accuracy among the modified Stearns-Noechel models, especially for four-primary blends with an average color difference of 0.50 Color Measurement Committee (CMC) (2:1) units, whereas Friele 1 had the highest accuracy among the modified Friele models, especially for five-primary blends with an average color difference of 0.46 CMC (2:1) units. These results indicate the modified model Stearns-Noechel 3 can be used in color prediction when there are fewer than five yarn components, and Friele 1 can be used for five components with lower color differences that better meet color-matching requirements in practical production.