Automated extraction of pair wise structure lines using airborne laserscanner data in coastal areas

The coastline of the North Sea is characterized by a large number of different morphological objects like dikes, dunes and tidal creeks. Due to the tidal effects and other natural forces the shape, size and position of such objects may change rapidly over time. In order to securely protect shorelines and coastal areas, a permanent monitoring has to be performed. In the past, mainly terrestrial surveys and aerial imagery have been used to obtain information about changes in time. In general these surveys include morphological features like structure lines. Important changes of the monitored objects can be detected by comparing identical morphological features of different time epochs. Unfortunately, the terrestrial surveys are very costly and timeconsuming. For this reason airborne laserscanning has been adopted to map changes of morphological objects. The first step of the monitoring task is the extraction of morphological features from the irregularly distributed 3D point cloud. One strategy of the extraction task is to fit suitable 2D functions to the 3D-points. Due to the fact that the choice of the used function represents known a priori information, structure lines can be derived from the estimated parameters of the function. However, a 2D approximation of the searched structure line is generally needed. This paper presents a new method for the extraction of structure lines from airborne laserscanner data in coastal areas using a hyperbolic tangent function. The method is based on a strategy that pre-defines the number of structure lines to be searched, the shape of the surface, the number of functions to be used and the approach how to calculate the structure lines from the surface function. Additionally, it is shown that the necessary 2D approximation can be estimated by digital edge detectors using a raster representation of the irregular laser points. Two meaningful examples are presented to demonstrate the capability of the algorithm.