Context and objectives
In the near future, the National Geographic Institute (NGI) will cover Belgium with numerical topo geographical data at the 1:10.000 scale. These data, in constant evolution, should be up to date. In that context, the NGI wants to schedule data updating and provide its users with an information about the actualisation status of the data. Therefore, it is crucial to find the zones that changed the most compared to the current state of the database.
The present project will produce this information, focusing on the most important objects, i.e. the communication network and the built areas. The Signal and Image Centre (SIC) of the Royal Military Academy (RMA) will provide the NGI with a prototype software embedded in a Geographical Information System (GIS), computing a global change value in a rectangular grid of given cell size, using very high resolution satellite data
Expected scientific results
The project demonstrated that it is possible to detect changes on road network and build-up
area comparing between SPOT5 imagery and the NGI database.
However, it was shown in this report the difficulties met for the validation. It is thus necessary to
further validate the results in the NGI production framework before the integration in the global
workflow. This evaluation should take into account the quality of the images and retain only the
most adapted one. The method should use human photo interpretation to validate the results.
Moreover, the results show also that there remain a significant number of false alarms.
Therefore it would be worthwhile in the future to look for methods aiming at refining the analysis
and reducing these errors.
It could be also of interest to find a way to summarize the potential change information in a
more statistical form.
Finally, the project was developed basically using SPOT5 imagery and its performance for
detecting changes larger than 2 000 m² (equivalent to 8x10 pixels) seems the best at 5m
resolution. It could be interesting to prospect new methods and tools to explore which additional
information may be supplied by higher resolution imagery, like NGI orthophotographs, Ikonos,
Quickbird or the future Pleiades satellites, to detect smaller changes, or to have more details on
On an organisational point of view, the project highlighted the complementarities of both
partners and what mutual benefit such collaborations may provide. Not only about change
detection, but maybe tomorrow in the field of feature extraction or recognition.
As far as the visibility of the project is concerned, three workshops have been organized during
the lifetime of the project, and six papers published: two conference posters, one of which got a
best poster award, two conference papers and two articles in a peer-reviewed journal. All of
them are provided in annexe.
The participation to the ETATS project enabled RMA to be an active partner of the GMOSS
network of excellence (cf. 5.3), and to get the responsibility of the “Change Detection WP” and
of the “Workshop WP”. After one year and a half of participation in this international group, the
network has asked RMA to increase its participation in terms of researchers (one researcher in
the new political WP) and to be responsible of the “Integration WP”. Being a member of this
network, the possibility for RMA to participate to further actions in security and stability is thus
The German international expert Helmut Mayer participating to the steering committee was very
helpful in technical discussions and stimulated the team to participate to the EuroSDR project
for which road detection on IKONOS images was a challenge. According to the organizer, the
results provided by RMA looked quite promising, given the difficult test area.
NGI will use the application resulting from the project to evaluate degradation of its topogeographical
data. The ETATS system is already identified clearly as a process in the ongoing
Business Process Reengineering to help collecting updates. Temporal degradation of data will
also be available in the metadata for internal and external use. Images will be purchased in
2006 to feed the system and get the final evaluation and tuning. At the same time, the
contribution of the results to the other collaborating information channels will be investigated.
|Project leader(s):||IGN/NGI (Institut Géographique National/Nationaal Geografisch Instituut)|