Registration Process of Laser Scan Data in the Field of Deformation Monitoring

Terrestrial laser scanning is a new method for monitoring the deformation of complicated objects. In contrast to the classical technology, no discrete identical points are available to compare. This problem can be solved if parts of the point cloud are grouped and the points in these groups are seen as representatives of a parameterized surface. An example for such a proceeding is the use of target spheres for the registration of scans. The radius of the spheres is known and the centre point coordinates can be calculated by a best-fit algorithm. These coordinates can then be introduced in a classical transformation calculation using identical points. However, such an approach has crucial disadvantages. First, it is necessary to use artificial targets and it can be of high effort to place these targets. Secondly, only a very small part of the redundancy contained in overlapping scans is used for the calculation. The resulting accuracy of the orientation parameters is therefore often not sufficient for detecting deformations. In this study, we compare scans of an object, where the possible deformation is in the order of the measurement noise. For that reason registration of point, cloud data is of great importance. The accuracy of different methods for registration and co-registration is presented.