HVI Color and UV/Visible Spectral Response of Cotton Fibers

Classification of cotton color is assessed by both human classers and HVI instrumental measurements. However, there are demonstrated inconsistencies between them, partly due to the human classer’s subjective perception and partly due to HVI’s 2-filter limitation. To improve the agreement, a number of attempts have been made, including the consideration of +a (redness) factor, the scanning of a larger cotton surface, and the adoption of neural network classifiers. As a different approach, this work introduces relative yellowness (%+b) or relative lightness (%Rd) that were derived from HVI color Rd and +b readings, and explores the feasibility of applying one variable for the classification and grading of cotton color. In order to examine the effectiveness of %+b or %Rd and also its equivalency to that of +b or Rd, cotton physical properties and UV/visible spectra were utilized. The pair of %+b and +b had a more pronounced relationship than that of %Rd and Rd, and the first PC (PC1) scores revealed more remarkable correlation with %Rd than with Rd. Furthermore, the PLS model performance was significantly improved for cotton lightness description with the use of %Rd.