Lidar Strip Adjustment with Automatically Reconstructed Roof Shapes

In recent years, the performance of lidar systems has steadily improved due to an increase in the achievable point density. The main error sources affecting the quality of lidar-derived secondary products such as digital terrain models (DTMs) and digital surface models (DSMs) are the result of systematic errors coming from insufficient boresight calibration. The effects of boresight misalignments are visually recognised as discrepancies of the laser point clouds at selected objects (such as building roofs) in overlapping areas of neighbouring lidar strips. In this work, a strip adjustment method is presented that corrects the systematic errors of the laser point cloud using five strip parameters (three shifts plus two angular corrections for heading and roll). The strip adjustment procedure uses 2D and 3D tie elements and takes advantage of a new 3D measurement technique applied to building roofs. The strip adjustment method is applied to three different datasets, consisting of conventional first/last pulse and full-waveform data. The results show that systematic discrepancies are typical for datasets that have already been processed by the flight companies. These errors can be reduced by more than 70% by applying the new strip adjustment approach. Preliminary investigations indicate that a higher laser point density and the inclusion of intensity and pulse width within full-waveform lidar data lead to better results.