Surface segmentation by variable illumination

Surface segmentation is a method to divide a surface into areas of homogeneous properties. Meaningful surface properties, such as the reflection characteristics or the local surface orientation, are derived from series of images by estimating the parameters of a reflection model. The images of the series show the surface illuminated from variable directions. A priori knowledge about the surface geometry can be used to improve the illumination strategy. Segmentation results often correlate with surface defects and thus can be applied as preprocessing step for a subsequent detection of defects. The performance of the approach is demonstrated with test specimens and cutting tools.