Photogrammetric matching of point clouds for 3D-measurement of complex objects

SD-measurement systems based on active triangulation using projection of structured light and recording in CCD-cameras are gaining more and more importance in industrial applications. Their advantages compared to conventional coordinate measurement machines are the non-contact measurement and fast acquisition of dense point clouds. Measurement of complex objects requires the registration from different views, resulting in point clouds with different coordinate systems. The main problem is the transformation of the resulting point clouds into a common world coordinate system. The precise transformation of each sensor coordinate system into the world coordinate system is neccessary to obtain a good accuracy of the whole measurement result. After a short discussion of some methods which are currently used for sensor orientation, a concept for 3D-measurement of complex objects based on photogrammetric matching of point clouds will be presented. The usage of circular features as matching elements will be discussed and a solution for the eccentricity problem will be presented. The world coordinates of the circular features are determined by photogrammetric methods. Afterwards, the point clouds of the topometric measurements can be transformed from the sensor coordinate system into the world coordinate system. Common problems of other transformation techniques like e.g. error summation are avoided. Furthermore, the accuracy of each transformation can be estimated by an error propagation model. The concept of the photogrammetric matching of point clouds will be demonstrated by 3D-measurement of some complex objects.