6533b837fe1ef96bd12a3119

RESEARCH PRODUCT

Dimensional control of metallic objects by artificial vision: contribution to lighting condition studies

Bernard LamalleSophie Kohler-hemmerlinClaudine CoulotPatrick Gorria

subject

Machine visionComputer sciencebusiness.industryOptical engineeringGeneral EngineeringFunction (mathematics)Object (computer science)Atomic and Molecular Physics and OpticsSurface roughnessReflection (physics)Computer visionArtificial intelligenceControl (linguistics)business

description

We present an original method that enables real time automatic dimensional control of defects on metallic objects. The method is based on the theoretical and experimental study of light reflection mechanisms that can be seen as a function of the surface roughness and of the light source size. We show that it is possible to reconstruct the profile of an object from a single gray level image. An industrial application enables us to validate our results. Finally, an error computation fixes the limits of our method and enables us to better understand illumination problems.

https://doi.org/10.1117/1.602172