A system for 3D CAD-based inspection using range images

摘要

An automated visual inspection system is presented for the detection of defects in range images of castings. The system uses computer-aided design (CAD) model information in several stages, including surface classification and inspection. Inspection techniques are presented for the detection of common casting defects and for the inspection of dimensional tolerances and manufacturing features. The inspection algorithms are applicable to the class of objects that contain planar and/or quadric surfaces, rather than being specialized to the visual inspection of a single object. The methods have been applied to a large number of real range images of iron castings that have real defects. The inspection system correctly classifies over 90% of the casting images.