NL FR EN
www.belgium.be

Automatic Extraction of Man Made Structures from VHR Optical and SAR data for Change Detection (EMSOR)

Research project OR/20/003 (Research action OR)

Persons :

  • M.  ACHEROY Marc - Royal Military Academy of Belgium ()
    Financed belgian partner
    Duration: 1/3/2006-28/2/2009

Description :

Objectives

The Pleiades/Cosmo-Skymed satellite constellation will provide an optic and SAR combination of very high resolution (VHR) sensors (resolution ~1 m) for which two main fields of application will be envisaged: security and cartography.
In the context of security, the relevance of remote sensing is well known since this technique represents sometimes the only way to get information in situation of difficult direct access (e.g., in the case of flooded areas). Furthermore, SAR sensors are particularly attractive due
to their all-weather operation capability.
For cartography applications, demand for up-to-date geographic data is increasing, due to fast changes occurring in many regions (urban areas expansion, forest clear-cuts, fires, etc.).

The difficulties related to the use of such data (mainly due to geometrical distortion and anomalies relevant at this resolution) make the performance evaluation of such systems a relevant topic of investigation.

The general objective of the EMSOR project is to foresee the performance of the future Pleiades/Cosmo-Skymed constellation with respect to the extraction of man-made structures.
To this end, the techniques developed during the project lifetime will be applied to images that offer similar resolution and bands.
More in detail, the following specific fields have been envisaged:

- man-made structure detection (from SAR and optical data and at different level of
refinement),
- security applications (for rapid mapping and damage assessment),
- routine cartography.

As a final result, tools (software) should be provided enabling the exploitation of the VHR optical and SAR data for man-made structures detection and monitoring in time. The improvement provided by the fusion of the detected features with respect to such applications
should be then assessed. Finally, expected practical outputs are maps with extracted cartographic features and/or risk-damage area maps.


Methodology

The proposed approach
- is based on a multi-resolution strategy involving a coarse and a fine structure detection on
the two types of images as well as on fused data;
- takes into account existing knowledge and collateral information such as vectorial data and DEMs.

These two aspects correspond to the point 2D-1 of the ORFEO programme. Furthermore, the involved SAR images will be X-band which addresses point 2D-2. The “existing knoweldge” mentioned above refer among other to in-house software for image co-registration, edge and
change detection developed and tested for previous projects. The expertise of the other network partners will be also taken into account for integrating this knowledge. This software will be adapted, further developed and tested on data with specifications similar to those of the
new satellite constellation.


Study area

Vaux


Data

IKONOS, SAR

Documentation :