| The objectives of this research are to analyze the LiDAR data errors and to develop alternative methodologies for estimating the biases in the LiDAR system parameters. The impact of the random/systematic errors on the derived point cloud is investigated in terms of LiDAR strip configuration such as flight altitude, direction, and scan angle. Two alternative methods have been developed to be used in cases where the point cloud coordinates of overlapping strips are available, but where raw measurements are not utilized. The simplified method consists of two steps: first, the 3D transformation parameters are estimated using the discrepancies between parallel overlapping LiDAR strips; second, the biases in the system parameters are derived from the estimated transformation parameters. The quasi rigorous method can deal with non-straight, non-parallel overlapping strips over rugged terrain with the help of time-tagged LiDAR point cloud and trajectory position data. In this method, laser firing points are estimated using the trajectory position data; then, the flight direction, beam direction, and encoder angle are calculated without system raw measurements. The proposed methods utilize a surface matching procedure, denoted as "ICPatch", which is beneficial in the absence of man-made objects in rural areas. The ICPatch procedure finds the closest point-patch pairs from overlapping strips, where one strip is represented by original points, and the other strip is represented by triangular patches. This research introduces two approaches for the similarity measure between the matched point-patch pairs. In the volume constraint, the volume of the tetrahedron which consists of the matched point and triangular patch are utilized as a constraint. For a point-based similarity measure, pseudo-conjugate points are derived from the matched point-patch pair, and the weight matrices for the pseudo-conjugate points are modified to handle the non-conjugate problem. The feasibility of the proposed methods is verified using simulated and real datasets. Using the simulated data, the proposed methods were investigated whether they are sensitive to the assumptions used in the derivation of them. The improvement of relative and absolute accuracy of point cloud after the calibration was evaluated using real LiDAR data. |
