Neural-Fuzzy Systems for Color Classifications in Textiles

Summary

This chapter introduces two applications of neural networks, fuzzy clustering and fuzzy logic to color classiifiications in textiles. The first application is the identiifiication of color patterns on a printed fabric. Color separation is necessary when the colorfastness of a printed fabric is evaluated because diiffferent colors in the fabric may change in different rates during the laundering or other treatments. A regular colorimeter cannot perform color separation and reports only the average color of an area that may contain multiple colors. The self-organizing-map and fuzzy clustering algorithm can be used to automatically separate colored patterns for independent evaluations. The second application in this chapter is the color classiifiication of cotton fibers using fuzzy logic. Cotton colors are grouped into a number of classes having blurring and overlapping boundaries. The partition of the classes in the current grading diagram does not reflect the grouping nature of cotton colors, yielding signiifiicant disagreements with the visual ratings. Fuzzy logic appears to be effective in dealing with ambiguity and uncertainty in cotton color grading.