From Pléiades to Pléiades Neo: Analysis of potential improvements for photogrammetric 3D reconstruction

Abstract

Based on its expertise in geometry and 3D, the Centre National d’Études Spatiales (CNES) is developing new open-source
tools capitalizing on its knowledge. They are intended to be part of the CO3D mission’s ground segment, and to serve the
entire applicative ecosystem. Processing this data requires robust softwares able to scale up. As a result, CNES 3D tools
are massively parallelizable, using multiprocessor and even multi-node technologies, while remaining hardware agnostic.
These software offer simple interfaces so that they can also be used outside operational processing chains. The tools
presented in this article can be used to produce a Digital Surface Model (DSM) from stereoscopic image pairs (using
CARS). The main step for the DSM computation is the stereo matching and it’s also performed using a CNES 3D tool
(Pandora). They also allow to extract a Digital Terrain Model (DTM) from the produced DSM (using Bulldozer) and to derive
a Digital Height Model (DHM). Following this, comparisons between all these digital elevation models can also be made
(using Demcompare). In anticipation of this data, the CNES is already testing these tools on other sensors, including
Pléiades and more recently Pléiades Neo. This article describes the improvement in 3D reconstruction thanks to this new
generation of satellites. CNES’s 3D tools are licensed under a free and non-contaminating license (Apache v2 License),
enabling the exploitation and enhancement of Pleiades and now Pleiades Neo data. They also offer the possibility of
reproducing the results presented in the article.