FOSS4G Europe 2024 workshops

Multimodal (2D/3D) change detection for natural disaster response
07-02, 16:00–18:00 (Europe/Tallinn), Room 335

This tutorial will highlight the capabilities of multimodal satellite imagery in addressing some of nowadays most impacting societal challenges using open source tools from CNES (French space agency).

In anticipation of the future CO3D mission, the French space agency (CNES) is currently developing open source tools for large-scale 3D data processing. With the arrival of 3D data on a more frequent basis, new use-cases for such data sources are emerging and enabling its applicability to pressing challenges such as improved crisis management in the aftermath of natural disasters. Change detection methodologies and their application to natural disaster response constitute a long-standing field within the remote sensing community, with initiatives like The International Charter Space and Major Disasters, open data programs of commercial satellite providers or annotated datasets like xview2 enabling a growing number of researchers and practitioners to continuously improve existing solutions. However, until now, the availability of post-event imagery, and thus related algorithms, have mainly been limited to monoscopic acquisitions, whereas with this type of constellations, a growing number of 3D data should also cover areas impacted by climate-related hazards.

In this context, CNES is currently working on the hybridization of 2D and 3D data for multimodal change detection. The proposed tutorial will present and teach some of this work being carried out by CNES through a pipeline that can be used for rapid mapping and longer-term risk and recovery management in order to help improving and enriching common existing approaches.

During this tutorial, you will discover how to:
- generate Digital Surface Models from sets of stereo satellite images using the CARS 3D reconstruction library :
- extract the Digital Terrain Models from Digital Surface Models through the Bulldozer library and derive a final Digital Height Model based on the obtained results :
- extract semantic information from the 2D imagery by using classification models and spectral indices
- combine 2D and 3D change indicators in order to quantify and localize the potentially changed areas on a map
- visualize, filter and extract information on these changes using the uncertainties provided by the tools

David Youssefi received the Electrical Engineering degree from the École Nationale Supérieure d’Electronique, Informatique, Télécommunications, Mathématique et Mécanique, Bordeaux, France. From 2013 to 2018, he was a Study Engineer at Communications et Systèmes, Toulouse, where he worked on remote sensing image processing especially in 3D processing. He is currently with the Centre National d’Etudes Spatiales (French Space Agency), Toulouse, as a Ground Segment Software Development Engineer, where he is responsible for software development for Pleiades and design for the future Earth Observation mission CO3D from CNES.

Dimitri Lallement joined the Centre National d’Etudes Spatiales (French Space Agency) in 2017 and integrated the Earth Observation Lab (EOLAB) in 2020. His main research topics are DTM extraction and 3D change detection from high resolution satellite imagery.

After receiving his M.Sc. degree in Computer Science from Humboldt-Universität zu Berlin (Germany), Christian Hümmer has been working as a research associate in the university’s “Computer Vision” research group in various interdisciplinary research projects in the fields of image processing and machine learning, as well as satellite image quality assessment in collaboration with the German Aerospace center DLR. He joined the French Space Agency CNES in 2021, where he is currently working as an image processing engineer in the Earth Observation Lab, focusing on uncertainty-aware semantic segmentation for EO applications and multimodal change detection.