Fast information acquisition on building features from digital aerial images

In the domain of emergency response to natural disasters and man-made conflict events, rapid situation assessment is crucial for initiating effective emergency response action. Estimation of damages caused by a disaster is a major task in the post disaster mitigation process. To enhance the relief and rescue operation in the affected area it is required to get a near real time damage model. For this purpose a fast method of data acquisition with suitable methods for extracting the man-made objects is required. In this work a methodological protocol is proposed to generate building description which can be used in a simulation system for training emergency forces. In this regard it is sufficient to extract 2D roof outlines of the buildings and attribute them with a constant height value, resulting in flat roof buildings. The process of building extraction from imagery can be separated in two tasks: first the detection of the object, second the recostruction. Another focus of this research is on the development of methodologies to extract automatic nDSM in order to remove non-ground objects such as trees, and cars from the dataset.