An automated process to segment 3D-pointclouds into geometric primitives

Along with new developments in sensor technology like industrial computed tomography, the number of measuring points in the generated point clouds gets increasingly larger up to several million points. In many metrological applications geometric primitives like planes, cylinders and cones have to be processed. Without automation, it is very time consuming and difficult to find parts representing single geometric primitives in the point cloud and to cut out manually the respective points. An algorithm was developed to segment point clouds automatically into separate point clouds, each representing a geometric primitive. One application is in the field of metrology as a pre-processing step before fitting geometric primitives. Besides a variety of other applications is possible e.g. in the field of automatic object recognition.