Reduced-dimension and wavelet processing of SMD images for real-time inspection

This paper presents a technique that uses a linear projection of images and other processing steps to arrive at a one-dimensional multiplierless correlation. This operation is used to detect the presence or absence of surface mounted devices (SMDs) in the inspection of printed circuit boards. Images with two types of illuminations are processed to produce two different decision schemes: one to detect component presence and the other to detect component absence. Additionally, the use of wavelet decomposition is examined as a pre-processing step in feature extraction, from which classification can be made.