Combining geometrical and semantical image information for the improvement of Digital Elevation Models

In response to the strong need for accurate and reliable three-dimensional terrain data, a couple of (semi-)automatical methods for the improvement of these data sets have been developed. However, in order to detect local blunders without loosing other important information like breaklines it is necessary not only to further develop single methods (as we will present a novel geometrical approach called constraint opening) but also to integrate all available sources - especially semantical information derived from optical imagery. We will describe and discuss corresponding examples.