Alignment and fusion of digital terrain models : case study of planetary surfaces

Data alignment and fusion are crucial steps in exploiting complementary data from different satellites. In planetary science, and particularly in remote sensing, the majority of data are not correctly aligned due to inaccuracies in sensor location. This makes merging and joint analysis imprecise and time-consuming, as planetary scientist have to correct these alignments manually. Existing alignment algorithms are designed for feature-rich data and pairwise alignment. The present study considers the case of Digital Terrain Models (DTMs), which are 3D data formats where elevation information is derived from a pair of 2D photogrammetric observations. This 3D information provides additional information useful for alignment. However, existing 3D alignment algorithms still suffer from the above-mentioned limitations with this data. The aim of this study is to develop a method to automatically align rigidly an arbitrary number of DTMs while producing a topographic model that fuses their information. We introduce a self-adaptive envelope, coupled to DTMs, whose role is both to produce a high resolution topographic model and to direct the alignment of the DTMs to a common reference.