Share Email Print
cover

Proceedings Paper

Surface Inspection Of Automotive Bodies By Reflective Computer Vision
Author(s): Y. Y. Hung; G. C. Jin; S. H. Tang
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

A simple but practical optical technique for automated surface inspection of car bodies is presented. The method which is based on light reflection is applicable to inspecting specularly reflective surfaces such as painted car bodies. A structured light signal consists of linear grating is imaged by a video camera via the surface to be inspected. With this arrangement, the surface being inspected acts as a mirror. Presence of surface flaws causes the grating to be locally perturbed. The grating-image is digitized and analyzed by a computer. Several algorithms are developed which automatically identifies the surface flaws by analyzing the perturbation in the grating-image. The technique allows surface flaws to be quantified in terms of slope deviation or depth variation. The sensitivity of the technique is very high permitting minute flaws to be detected. In the paper the theory of the technique will be presented together with experimental validation. The technique possesses numerous practical features such as requiring no special surface preparation, allowing evaluation in place, requiring minimum environmental safeguards, allowing rapid testing and evaluation, providing reliable and quantitative results, and it can be automated. Therefore the technique has clearly exhibited a great potential for being developed into a production-line inspection tool.

Paper Details

Date Published: 14 December 1988
PDF: 4 pages
Proc. SPIE 0955, Industrial Laser Interferometry II, (14 December 1988); doi: 10.1117/12.947663
Show Author Affiliations
Y. Y. Hung, Oakland University (United States)
G. C. Jin, Oakland University (United States)
S. H. Tang, Oakland University (United States)


Published in SPIE Proceedings Vol. 0955:
Industrial Laser Interferometry II
Y.Y. Hung; Ryszard J. Pryputniewicz, Editor(s)

© SPIE. Terms of Use
Back to Top