Towards 3D City Modeling through Combining Ground Level Panoramic and Orthogonal Aerial Imagery

Aerial images and ground-level images are two major types of image sources used by many existing image-based urban 3D modeling approaches. Due to the rapid development on hardware and camera technologies, panoramic images have become widely accessible and they have many advantages in supporting the 3D reconstruction tasks. In this paper, we have proposed an automatic image registration approach to determine the spatial relationship among a large set of ground-level panoramic images and an orthogonal aerial image. The heights of the buildings can be estimated through our matching algorithm of buildings rooftop edges on the panoramic image and the buildings footprints on the aerial image. We have also developed a method to automatically extract and rectify the building textures from the panoramic images for texture mapping of 3D city models.