AUTOMATIC OPTICAL INSPECTION OF MICRO DRILL BIT IN PRINTED CIRCUIT BOARD MANUFACTURING USING SUPPORT VECTOR MACHINES

Automatic optical inspection (AOI) Of Micro drill bits becomes move and more important with the rapid expanding of Printed Circuit Board (PCB) manufacturing industry. Traditional methods are mainly focus on geometric defects inspection Of micro drill bits. This paper proposes a phase identification method using Support vector machines (SVMs) for micro drill bits inspection. Three kinds of drill bit features are extracted as the input of SVMs. The Gaussian radial basis function (RBF), is adopted as kernel function, and one-against-one strategy is employed to extend dual category SVMs for multi-classification of micro drill bits. Classification performance is compared with artificial neural network, decision tree and k-nearest neighbor classifiers. The results indicate that the approach works well for AOI of micro drill bits.