Towards Automating the Generation of Facade Textures of Virtual City Models

This paper is concerned with concepts and techniques that support the automated generation of facade textures. In contrast to terrain textures, which can be derived automatically, capturing and processing facade textures still represent costly and time intensive tasks. In this paper we introduce a number of basic tools that allow us to work with aerial imagery captured by the High Resolution Stereo Camera (HRSC). In addition to the nadir view of standard aerial photography an HRSC image set contains multiple tilted views, in which facades are depicted more prominently. We develop a camera model for OpenGL that approximates the unique imaging sensor used for HRSC. With our tools, we can visually explore original HRSC imagery and synthesize idealized HRSC imagery based on virtual 3D city models. The visual exploration of HRSC images allows us to select regions from real-world images that have special properties useful for algorithm development, whereas the synthesis is used to generate test data and to run simulations.