Cotton Color Distributions in the CIE L*a*b* System

Information about cotton color distributions in the CIE L*a*b* is critical to developing new chromatic technology for color grading, since the CIE L*a*b* is a more standardized system than the R_dab system currently used in the cotton industry. The objectives of this study are to examine the distributions of cotton colors in the CIE L*a*b* and to explore models of the probability density functions of individual color grades (classes). Based on the experimental data of the USDA physical standards for Upland cotton and samples from the 1999 crop, cotton colors are naturally separable in the CIE L*a*b* space, the probability distributions of color grades are normal, and the distributions of two adjacent grades overlap to some extent. This work demonstrates how to construct a multivariate normal distribution for a color grade based on the statistical data of samples in that grade, and how to determine a hard decision boundary between two adjacent grades that minimizes misclassifications. The color distribution models are helpful for establishing more effective classification methods for color grading systems.