Georeferencing Using Automated Reference Point Search for Engineering Geodesy Purposes

The primary objective of this paper is the automation of the point cloud georeferencing process through the algorithmic localization of various control point types and the subsequent calculation of transformation parameters. The proposed approach utilizes iterative methods to identify matching control points between the reference and local networks, with a strong emphasis on robustness against gross measurement errors. The adjustment pipeline is executed in several core steps: first, reference targets are identified based on their geometric shape, and their local coordinates are determined. Subsequently, the coordinates derived from the point cloud are compared with the actual coordinates provided as the transformation baseline. Utilizing the spatial configuration of both networks, the system pairs points that satisfy the geometry of the scanned point cloud, assigns real-world coordinates to the control points, and computes the initial transformation key. Then, through reverse transformation in the spatial regions of undetected targets, the algorithm systematically verifies the presence of control points, calculates their coordinates, and includes them in the iteration process to refine the final transformation key. A seven-parameter Helmert transformation, implemented via the Umeyama algorithm, was utilized for the coordinate system conversion. The final output consists of a georeferenced point cloud accompanied by the root-mean-square errors of individual reference points. The experimental results demonstrated transformation residual root-mean-square error (RMSE) values of 13.6 mm and 15.6 mm, which are fully comparable to the manual point-picking approach, exhibiting a discrepancy of only 3.0 to 4.0 mm. Crucially, the presented automation achieved a multi-fold reduction in processing time compared to the manual workflow.