A REFINED SENSOR MODEL FOR PANORAMIC CAMERAS

Digital terrestrial panoramic cameras are providing a new device for taking high-resolution images. The produced images have been used widely for purely imaging purposes such as an indoor and landscape imaging, cultural heritage recording, tourism advertising and image-based rendering. Due to the high information content of those images they may also be used as an efficient means for 3Dobject reconstruction. Therefore the need arises to establish a sensor model and investigate the potential of the system in terms of accuracy. With a well-defined sensor model we have additional powerful sensors for image recording and efficient 3D object modeling. In this paper we present the result of physical measurements of tumbling for the SpheroCam. A method for tumbling error modeling of panoramic cameras is introduced and we will show how it can be used in bundle adjustment with additional parameters. Then we report the result of modeling and accuracy testing using a 3D testfield and compare it with the case when tumbling is not being modeled for EYESCAN and SpheroCam. The results indicate subpixel accuracy for both systems.