HIERARCHICAL DATA FUSION WITH PHOTOGRAMMETRIC APPLICATIONS

Abstract

When two datasets are fused using least-squares adjustment, usually all the results will be affected by some change, even the reference data that are meant to provide information of such high quality that they should remain stable. In order to avoid this effect, the sequential adjustment is to be replaced by a strictly hierarchical method in which the esti-mation procedure is designed to reproduce everything that belongs to a "higher category" and to perform an adjustment in the least-squares sense for everything else. After presenting such a suboptimal estimator, but with the "reproducing property," this technique is applied to the integration of photogrammetric networks of substantially different scales.