Automated inspection of solder bumps using visual signatures of specular image-highlights

The author describes a technique for automated inspection of solder bumps by analyzing spatial features and intensity characteristics of image highlights. It is based on high-contrast imaging of specular soldered surfaces against a reflective background using dark-field illumination for selective enhancement of surface topography. Bright-field images of these surfaces are also analyzed to resolve ambiguities in the defect-identification process. The reflectance of solder is studied, for the estimation of corresponding signatures of image specularities, which provide information about surface curvature, metallurgical constituents and topographic discontinuities. During inspection, a specular highlight in the run-time image is mapped into a feature space of reduced dimensionalities for comparison against the features of typical signatures. Based on this principle, visual inspection systems have been designed and implemented for high-speed detection of defective solder bumps and leadless ceramic chip carriers and metallized advanced VLSI wafers