Automatic Texturing of 3D Models of Urban Areas Using Image Sequences from Airborne TIR Cameras

Thermal infrared (TIR) images of the building hull are often used for detecting damaged and weak spots in the insulation of the building hull. However, imaging and manually analysing each face is very time consuming. This process can be carried out more efficiently using a mobile mapping system (MMS) equipped with a TIR video camera. Frontal façades of the buildings which are visible from street level are captured by a terrestrial TIR camera mounted on a moving vehicle. To capture the roofs, an airborne platform has to be used. Façades in inner courtyard which are not visible to the street-view camera can be photographed using an oblique looking camera. An assignment of the TIR images to the 3dimensional (3D) building models should be carried out to obtain the spatial reference and combine the images with the spatial information. This can be achieved via texture mapping. Building textures can be extracted automatically if the exterior orientation of the sensor is known. An approximated exterior orientation is given by the navigation data. As the accuracy of the navigation data is not sufficient for a precise texture extraction, a co-registration method is necessary to achieve well fitting textures. Furthermore, in image sequences every face of the 3D model appears multiple times, thus a strategy for the best texture selection needs to be developed. Finally, objects and thermal structures should be detected in textures and embedded in 3D building models.