AN INTEGRATED SEMANTICALLY CORRECT 2,5 DIMENSIONAL OBJECT ORIENTED TIN
Andreas Koch · 2005
This paper presents an approach for the homogenisation and integration of two-dimensional GIS vector data and 2,5D Digital Terrain Models (DTM). Due to inconsistencies between the data which are caused by different surveying and production methods, different modellings and large time shifts between acquiring the data, the data sets must be homogenised. The homogenisation considers the accuracy of the planimetric coordinates of the 2D GIS data as well as the accuracies of the heights of the DTM points. The aim of the homogenisation is to produce 2,5D objects represented in a semantically correct way, i.e. the objects have to correspond to our view. For instance, a locally restricted lake geometrically can be represented by a horizontal plane, i.e. the heights of all points inside the lake must be identical. To give another example, a road should not exceed maximum slope and curvature values in driving direction and across. These characterisations are formulated by equalities and inequalities and are introduced into an inequality constrained least squares adjustment which is solved using the linear complementary problem (LCP). In a further step an integration of the data is applied, which is based on a constrained Delaunay triangulation. The aim of the approach is the creation of an integrated data set containing objects with respect to the third dimension.