Three-dimensional holistic approximation of measured points combined with an automatic separation algorithm

Area or line-oriented optical probing systems increasingly complement tactile Coordinate Measuring Machines (CMM). These optical techniques can measure many types of curved surfaces with a high point density. Due to the relatively large surface registered during one probing cycle, these measured point clouds can include measured points belonging to different geometric elements. A manual separation by the user is possible, but may lead to non-reproducible results. Therefore, an automatic separation is required, which refers to the nominal geometry. The presented algorithm is based on an approximation according to the L2-Norm. An automatic separation is included in each iteration step by three-dimensional segmentation functions. This paper presents the results and stability of an algorithm for the so-called micro-cup geometry of a micro deep-drawing punch.