An Assessment of WorldView-1 Positional Accuracy based on Fifty Contiguous Stereo Pairs of Imagery

This paper presents an assessment of WorldView-1 3D positional extraction accuracy based on 50 overlapping and contiguous stereo pairs of WorldView-1 imagery covering approximately 50,000 square kilometers of the earth's surface. Absolute accuracy is both predicted using error propagation and measured using reference points. Shear, a form of relative accuracy, is also assessed between each overlapping stereo pair of images. Accuracy is assessed for two different processing approaches: extraction from individual stereo pairs, and extraction from stereo pairs following the fusion of information across all overlapping stereo pairs. The latter approach was implemented with Metric Information Network (MIN) processing, an efficient process with results equivalent to a simultaneous block adjustment. Both approaches used no ground control, although in a separate sub-experiment, MIN processing was supplemented with a sparse set of ground control (GPS surveyed) points.