Segmentation of unsorted cloud of points data from full field optical measurement for metrological validation

Modern industry requires the increase of quality of manufactured products with the simultaneous minimization of production time and cost. Therefore the development of faster, more precise measurement techniques is needed. There are many full field optical systems in use that offer multi directional measurement that meet these requirements. The raw output measurement data from these systems is in the form of unsorted clouds of points which may include millions of measurement points. This data has a different structure than data acquired by traditional contact methods. In addition phenomena connected with optical measurement such as reflection and occlusion result in various errors in the obtained cloud. Therefore a new method of analysis has to be developed to process the data and prepare it for metrological verification. This article presents an algorithm to manage measured data from full field optical systems. This includes segmentation of clouds of points so that each point is associated with its corresponding surface of the CAD model and then exported to certified metrological software for analysis.

► For accurate analysis of the object's shape it has to be described by thousands or millions of measurement points. ► Data analysis is required to split the measurement results into areas of uniform shape and assign their corresponding surfaces from the CAD model. ► This process is a recognition problem where we attempt to find higher-level patterns from given low-level information. ► The resultant data is exported to certified metrological software for measurement uncertainty analysis.