Feature Extraction of Wear Debris Shape and Its Identification with ANN

A new shape descriptor signature was used to describe the irregular microscopic wear debris shapes. The method extended the wear debris boundary to 1D representation, and extracted the shape features with translation, rotation and scale invariance. Used as input vector of radius basis function(RBF) neutral network, the wear debris shapes were divided into four classes: regular, irregular, circular and elongated. Result of application shows that due to the full use of the boundary information, signature not only describes the shape feature of wear debris, but also represents the edge detail of wear debris. And the classification system based on neutral network is fast in convergence, and high in accuracy.