Visual Quality Control of Planar Working Pieces: A Curve Based Approach Using Prototype Fitting

In this paper we elucidate a method for visual quality control and vision metrology of planar objects in manufacturing. We present an efficient approach to comparing a masterpiece with an arbitrary working piece detected on a line of a factory. After extracting the contour of the object, we use prototype fitting to determine the best transformation which maps the masterpiece to the extracted contour points. In contrast to classic ICP like methods, we suggest an encoding of the masterpiece by a curve, which allows a fast computation of point-curve distances, in order to guarantee a more accurate, faster and less memory intensive performance.