Calibration of an Ir Camera System for Automatic Texturing of 3d Building Models by Direct Geo-referenced Images

In the last decade the capability of thermal IR-cameras has stepwise increased while the prices have gone down. This enables to use IR technology in wide field of applications. Infrared (IR) cameras can be used to capture the IR radiation of buildings. By analyzing data of thermal sensors it is possible to detect the areas with the highest loss of heat. Nowadays in most cases thermal images are analyzed only qualitatively. However, without analyzing the geometry of the data acquisition or not dealing with the geometry of the captured scene, it is difficult to combine thermal data with other imagery or semantic information stored in the spatial data base. Aim of our work is the automated texture mapping of existing 3D building models with images recorded by IR cameras. In this way, 2D images are connected with a 3D GIS database and can be further analyzed together with the information about the geometry of objects. For texture mapping the external orientation of the image is required. In general a direct geo-referencing is possible using data from inertial navigation system (INS) and GPS. For this purpose the camera system including GPS/INS has to be calibrated. In order to utilize the GPS/INS data correctly, the parameters of the boresight and the leverarm have to be estimated. An accurate estimation of internal orientation parameters have to be taken into account. A determination of distortion was also performed. As the acquisition of the images was conducted from a helicopter platform which was vibrating with a high frequency and amplitude, some additional problems occurred. Finally the texture mapping of selected parts of the Technische Universitaet Muenchen 3D building model was performed.