Neural Network Technique for Fiber Image Recognition

Abstract

A neural network method of analyzing cross-sectional images of a wool/silk blended yarn is studied. The research has two major components: the process of original yarn cross-sectional images including image enhancement and shape filtering; and the determination of characteristic parameters for distinguishing wool and silk fibers in the enhanced yarn cross-sectional images. A neural network computing approach, single-layer perceptrons, is used for learning the target parameters. The established neural network model features a good capability of tolerance and learning, in contrast to traditional methods of image pattern recognition. The study indicates that preparation of the yarn sample slices is critically important to obtain undistorted fiber images and to ensure the accuracy of fiber recognition by the neural network model. The research concludes that the overall error estimate for recognizing wool or silk fiber is 5%.