Quality assessment for LiDAR point cloud registration using in-situ conjugate features

This study aims to develop a quantitative approach for evaluating the quality of a LiDAR point cloud registration solution. First, a highly-efficient technique for integrating multiple LiDAR point clouds was introduced using directly the in-situ conjugate features (including point, linear, planar features and groups of points). Then by applying the Non Iterative Solutions for Linear Transformations (NISLT) technique, the transformation between datasets can be directly solved without carrying out iterative computations or using any prior initial information. The weighted model for the observables was also developed to improve the reliability of obtained solutions. Finally, two quality indexes, namely the absolute consistency and relative geometric similarity, were proposed to give complete and realistic information on the quality of the integrated solutions. A case study for a real-field facility has also been performed to reveal the distinguishing capability and superiority of this newly-developed technique.