Automatic defect classification of printed wiring board solder joints

An automatic windowing algorithm is developed for use in automatic printed wiring board solder joint inspection. The method uses Hough curve detection to locate the solder joint within the image. Eleven good features selected by C.-C. Lee (1987) for use in the solder joint inspection task are studied in detail for purposes of optimizing the inspection process. This is done with the aid of a minimum distance classification algorithm that allows for classification of the solder joint based on user selected features. Automatic windowing, feature selection, and classification algorithms are compiled into a complete solder joint inspection system.<
>