Approaches to Discriminate the Characteristic Generic Hand of Fabrics

Abstract

Discriminating between the hand of different fabrics is an important issue in the textile industry. This study focuses on the approaches to discriminate between fabrics with different characteristic generic hands. Four groups of woven fabrics from cotton, linen, wool, and silk are the fabric types in this study. Discriminant analysis and the neural network method are used to characterize and discriminate between the different fabric groups. The parameters for analysis are selected using a stepwise method from sixteen mechanical properties based on the KES-FB system. The results show that cotton, linen, wool, and silk groups of fabric can be characterized and discriminated by discriminant analysis and the neural network method with a 100% classification accuracy rate. There are fewer parameters with the neural network method than with the Fisher and canonical discriminant functions.