Constraint Energies for the Adaptation of 2d River Borderlines to Airborne Laserscanning Data Using Snakes

The German Authoritative Topographic Cartographic Information System (ATKIS) stores the height and the 2D position of the objects in a dual system. The digital terrain model (DTM), often acquired by airborne laser scanning (ALS), supplies the height information in a regular grid, whereas 2D vector data are provided in the digital landscape model (DLM). However, an increasing number of applications, such as flood risk modelling, require the combined processing and visualization of these two data sets. Due to different kinds of acquisition, processing, and modelling discrepancies exist between the DTM and DLM and thus a simple integration may lead to semantically incorrect 3D objects. For example, rivers may flow uphill. In this paper we propose an algorithm for the adaptation of 2D river borderlines to ALS data by means of snakes. Besides the two basic energy terms of the snake, the internal and image energy, 3D object knowledge is introduced in the constraint energy in order to guarantee the semantic correctness of the rivers in a combined data set. The image energy is based on ALS intensity and height information and derived products. Additionally, features of rivers in the DTM, such as the flow direction or the river profile, are formulated as constraints in order to fulfil the semantic properties of rivers and stabilize the adaptation process. Furthermore, the known concept of twin snakes exploits the width of the river and also supports the procedure. Some results are given to show the applicability of the algorithm.