Facade Texture Quality Estimation for Aerial Photography in 3D City Models

This contribution presents concepts and methods to estimate the quality of facade textures taken from a set of overlapping aerial images. It aims at providing a key component for a workflow to derive appearance elements such as facade textures for large-scale virtual 3D city models in an automated way. We describe a technique to project aerial imagery generated by the High Resolution Stereo Camera (HRSC) onto the 3D geometry of a virtual city model. This technique can be applied in real-time using 3D graphics hardware. It is based on rectified images that are extracted for each facade at a fixed spatial resolution along with a quality map encoding the actual effective spatial resolution of the projected aerial image. We demonstrate the potential of this combination of facade images and quality maps for applications such as facade coverage determination.