Method and apparatus for feature detection in a workpiece

Abstract

A feature detection apparatus includes a reflective imaging system, a transmissive imaging system, and memory for generating an image of a workpiece. The reflective imaging system generates a first bit image of a surface of the workpiece. The transmissive imaging system generates a second, different bit image of the density of the workpiece. The two images are then combined in memory into a more complete image that contains data describing the surface and interior of the workpiece. The apparatus also includes a filter for enhancing the combined bit image so that physical features of the workpiece are accurately detected. The filter is constructed to perform the following steps. For detecting a feature of interest, the filter determines a mean, standard deviation, and normal distribution of the image's pixel intensities. The filter then uses the normal distribution to map the pixel intensities above the mean to one set of enhanced pixel intensities and to map pixel intensities below the mean to another set of enhanced pixel intensities. Pixels of the enhanced image are sampled to determine if a pixel meets a threshold intensity for the feature being sought. For a pixel that meets the threshold intensity, a feature detection operation is applied to the pixel and its neighboring pixels to determine if the feature is present in the image. If so, the full extent of the detected feature is then determined.