TRAINING-FREE COUNTER PROPAGATION NEURAL-NETWORK FOR PATTERN-RECOGNITION OF FABRIC DEFECTS

We present an application of a Training-free counter propagation netwo rk (TFCPN) to detect fabric defects. The TFCPN, which is a modificatio n of Hecht-Nielsen's counter propagation network (CPN), learns through a simple recording algorithm devoid of any training, while retaining the topology of the CPN model. The mathematical justification for the modification is also presented. Four kinds of fabric defects-neps, bro ken ends, broken picks, and oil stains-most likely to be found during weaving are considered for recognition by the network. Results show th at fabric defects such as these inspected by means of image recognitio n in accordance with the TFCPN agree approximately with initial expect ations. The CPN reported in this paper is training-free, and it can le arn complicated textile design problems.