Automatic co-registration of photogrammetric point clouds with digital building models

Abstract

Point clouds serve as a valuable resource for the modeling and updating of digital building models. For exploiting their full potential, they must be correctly registered within the building coordinate system. In this paper, we propose a highly automated co-registration algorithm for photogrammetric point clouds that works without the usage of control points. This drastically decreases the required interactive effort for the registration process. Instead, the building model serves as reference for the co-registration. The procedure extracts 3D line segments from the image data and associates them with bounding surfaces from the building model in order to calculate the necessary transformation parameters during an adjustment process. The validation shows that registration accuracies of ±3–5 cm can be achieved with our method. Since the stochastic properties of the input measurements are considered during the adjustment process, statements about the reached reliability of the transformation parameters can be made.