Control Point Networks, DTMs and Rectified Image Mosaics from Clementine Imagery

Abstract

We describe techniques which we will use to determine regional control point networks, DTMs (Digital Terrain Models) and rectified color image mosaics from large collections of spacecraft stereo images in particular Clementine data. First, overlapping images suitable for stereo analysis are identified, and tiepoints in overlapping images are selected manually. Next, latitudes, longitudes and heights of these tiepoints are determined in combination with the camera pointing data of all images and the spacecraft trajectory in a "bundle block adjustment". Thus, the resulting planetary coordinates are used as control points subsequent analysis. Next, stereo disparity data between image pairs are determined using the "Gruen-Otto-Chau stereo matching algorithm". These data are converted to planetary coordinates based on the spacecraft trajectory and altitude data determined during the bundle adjustment and are interpolated to form the DTM grid. Finally, all images are rectified on the basis of the elevation model, reprojected onto a map, and combined to mosaics. The procedure allows us to archive precise registrations of the color filter images. Furthermore, precise surface slopes and illumination angles can be computed from the DTM to perform a physically more meaningful spectrophotometric analysis of the terrain. The performance of the bundle block adjustment and of the matching software will be demonstrated. Examples will be given of DTMs near Mare Orientale and the polar regions of the Moon. Software contributions from TUM, TUB, and UCL (Technical University Munich, Technical University Berlin, and University College London) are acknowlegded.