Including virtual target points from laser scanning into the point-wise rigorous deformation analysis at geo-monitoring applications

Abstract

We present a method to include virtual target points from laser scanning into the point-based rigorous deformation analysis to derive precise 3D deformation vectors. This method overcomes the challenge of missing point identities in laser scans and is developed especially for geo-monitoring applications that demand an early identification of deformations at previously unknown positions. Our approach is based on virtual targets represented by local scan patches. Each patch is matched between overlapping stations and across different measurement epochs using the Iterative Closest Point Algorithm (ICP). Thus, similar to feature points, a number of homologous points is created and polar pseudo-observations are derived. This allows to integrate the observations into a free network adjustment and into a rigorous deformation analysis. We apply this method to the geo-monitoring of rock surfaces on Mt. Hochvogel where we use a scanning total station combining point-wise measurements to signalized targets and pseudo-observations derived from area-wise laser scans. In our application, numerous virtual target points could be created throughout the deformation object. The results show that the new method improves the accuracy and reliability of the subsequent rigorous deformation analysis and it, thus, allows for an early identification of deformations at geo-monitoring applications. Still there is some improvement in the selection of suitable patches needed, as the matching accuracy of the ICP strongly depends on the point distribution within the patches.