A CLASSIFICATION ALGORITHM AND OPTIMAL FEATURE-SELECTION METHODOLOGY FOR AUTOMATED SOLDER JOINT DEFECT INSPECTION

A classification algorithm and optimal feature selection methodology a re developed for implementation on an automated solder defect inspecti on system. Computer-generated three-dimensional geometric models of so lder joint defects are used to train the system and simulate defect cl assification. System performance is assessed through simulation and is shown to be successful at classification of dominant solder defects, including simultaneously occurring defects and overlapping defect clas ses.