Registration of LiDAR data through stable surface matching

The accuracy of final LiDAR points on ground depends on various factors including GPS/IMU position and orientation, laser range and scan angle measurements. As a result, the adjacent overlapping swaths or flight strips fail to match with each other. Surface matching is a common procedure used to relatively register the LiDAR data and reduce the mismatch between the overlapping strips. However, in case of smooth topography with fewer features, classical registration algorithms render poor results. This paper presents a modified registration technique for the optimal relative registration of LiDAR data over a flat terrain with fewer features. In this method, the feature samples used in the algorithm are selected depending on their geometric stability. The points from overlapping strips are matched through iteratively closest point (ICP) method based on minimizing point to plane normal distance.