Spatiotriangulation with multisensor VIR/SAR images

The objectives of this research study was to evaluate the spatiotriangulation applied to multisensor satellite images, which enabled the simultaneous geometric processing of a large number of images and strips together to reduce the control point requirement. The spatiotriangulation is based on the three-dimensional physical models developed for multisensor images at the Canada Centre for Remote Sensing, Natural Resources Canada and on a least squares block bundle adjustment process with orbital constraints. The spatiotriangulation was applied to 49 images in six blocks (Landsat-7 ETM+, panchromatic SPOT-4 HRV, multiband ASTER, multimode radar RADARSAT-1, and ERS-1) acquired over the Rocky Mountains, Canada, from different viewing/look angles. The first results of least squares block bundle adjustments showed that the same error residuals (around 20 m) were obtained with the different image blocks whether independently or simultaneously processed. In addition to ground control points (GCPs), elevation tie points (ETPs), with a known elevation value, instead of normal tie points were used in the overlaps because the viewing/look-angle differences of overlapping images were generally small (8°). The second and most important results were related to simultaneous bundle adjustments of the largest "master" Landsat-7 block (600 km × 500 km) using 25 GCPs in the two outer strips and the smallest "slave" block(s) using no GCP but only ETPs. The errors, verified by a large number of independent check points (ICPs) in the "slave" blocks, were between 15-35 m (1.5-2 resolutions), depending on the "slave" block. However, the combined image pointing and map errors of ICPs (25-30 m) are included in these 15-35-m error results, and the internal accuracy of the blocks should, thus, be better (around one resolution). The research study demonstrated, thus, the possibility to use the largest block with a reduced number of GCPs to simultaneously adjust single image(s)/strip(s) or smallest block(s) with only ETPs, and with no degradation in the accuracy.