Defect Detection on an Inhomogeneous High-Contrast Surface

Abstract

difficult monitoring task required in the manufacture of different parts. This can be best demonstrated on the detection of different defects on a metallic surface, since the monitoring on metallic surfaces plays an important role in the manufacturing industry and in science. The variety of materials used (massive metals, tin etc.) is so vast that it appears impossible to detect the creation resemblances of any possible surface defects. The defect recognition can be considerably impeded by the highly variable gloss level of metallic surfaces or by the shadow areas created by surface folds. Beyond that, process-related impurities or surface discolorations from lubricants (oil or emulsion) can lead to various artefacts that can be erroneously detected as defects. It has to deal with an inhomogeneous highcontrast surface featuring different additional effects. Nevertheless, a flexible but explicit defect recognition needs to be performed on these surfaces. The discussed formalistic methods (see Chap. 1) cannot be used on an inhomogeneous high-contrast surface with global and local surface brightness variations. Such surfaces require the development of new algorithms that follow the physical principles of defect creation. The basic difference between the different real defects and pseudo-defects is in their edges, as described above. Thus, all surface defects show common characteristics that can be used to recognize and separate them from pseudo-defects.