Prof. Dr.-Ing. Manfred Ehlers

Change detection is a very important challenge. Various change detection methods are documented in the literature but most of them focus on two-dimensional data which limit their applicability. Therefore, in this project we focus on the development of three-dimensional change detection methods considering three different scenarios. First, we assume that there is no three-dimensional information for both old and new scenes to detect changes. We propose using two-dimensional change detection algorithms followed by approximate three-dimension extraction of the changed regions. Here, we will develop novel depth estimation methods based on shadow and shape information. Second, we assume that we have the three-dimensional information (in terms of stereo image pairs) to be available either for the old or the new scene only. To detect changes in three-dimensions, we will construct the digital elevation model (DEM) based on the stereo image matching principle. For the scene having mono image, we will generate the DEM data by shadow and shape information. Using novel three-dimensional change detection methods, we plan to detect changes on these data sets. While detecting changes, we will also benefit from the two-dimensional information as in the first scenario. Third, we assume that for both old and new scenes we have the three-dimension information (in terms of stereo image pairs). Using these, we plan to develop novel three-dimensional change detection methods. Again, for this scenario we will benefit from the information in the first and second scenarios. Results obtained from three different scenarios will be compared and analyzed regarding the additional information integrating three-dimensional data. The outcome of this project will be a prototype of a fully developed change detection system to be used on satellite images. It will have a potential for detailed urban monitoring, damage assessment and disaster monitoring applications.