Dimensional control of metallic objects by artificial vision

This paper presents an algorithm which allows control of metallic objects by measurements obtained through artificial vision. The system consists of a single camera and a single light source. The algorithm is based on the relation between the surface orientations of an object and the gray level of the corresponding pixels. This relation allows the object profile to be reconstituted and the object dimensions to be controlled. The paper also deals with the evaluation of measurement accuracy, the origin of the errors and the limitations of our algorithm. Finally, an application of this algorithm is presented in the case of machine tool swells on a metallic object. The measurements obtained by this approach are compared to those obtained from a classical method based on a mechanical sensor. We conclude this paper with the improvements which can be brought to our method in order to reduce measurement inaccuracy