Neural network painting defect classification using Karhunen-Loeve transformation

This paper deals with the problem of painting defect detection on reflectin g surface objects. The problem has been approached with an optical inspecti ng method. A laser beam hits the object surface. The light scattered from t he rough surface generates a digital speckle. The speckle is affected by th e painting defect. Using the Karhunen-Loeve transformation, the speckle pat tern is transformed into a feature vector. This information is used to trai n the neural-networks in recovering the defect. The reliability and effecti veness of a prototype is validated by experimental results. At the end, the proposed method is compared with another optical inspection method. (C) 20 00 Elsevier Science Ltd. All rights reserved.