Context and objectives
We propose to develop, from the tools available in the various laboratories of the network, a computer aided system for the planning of humanitarian mine clearance campaigns. The developed product will contain the two complementary levels (global and local) corresponding to information needs at two different spatial scales. The system will integrate different modules : a database in a geographical information system (GIS), a toolbox of classifiers, a toolbox of object detectors and a graphical user interface (GUI).
At the global level corresponding to a country or a large region, the system will allow to manage, centralise, and enrich the existing geographical information in order to plan the mine clearance operations according to the established priorities. At the local level, it will provide the information the field-workers require. This information, available from a GIS and a GUI, will allow to respectively define the areas to be cleared from mines first and to analyse the areas of interest in order to intervene.
Expected scientific results
Tables and documents synthesise the elaboration of the planning method and of the GIS software prototype. The table of user needs lists the required information, its spatial nature, possible sources, appropriate scale, and provides some comments. Another table conveys the system design, describing the expected functions and working scales. A graph describes, in operational mode, the data flow between concerned actors. A humanitarian demining legend is proposed in accord with IMSMA.
A software package working at four scales embedded in a GIS, compatible with IMSMA and the Belgian UXO Lao database has been produced. The Country Scale (1:1 000 000) is fed by the Digital Chart of the World, topographic maps, meteorological data and maps of refugees. The Region Scale (1:250 000 to 1:50 000) contains satellite images (SPOT, Landsat TM, RADARSAT), topographic maps and information from field survey. The Field Scale (1:10 000 to 1:1 500) contains aerial photos, very high-resolution satellite images (IKONOS), statistics and sketches. The Advancement Scale (1:500) is added in order to produce a detailed description of each minefield.
The SIC Graphical User Interface has been extended with interactive tools for supervised classification, for launching programs and visualising results. A pool of programs divided in several classes (pre-processing, processing and post-processing) has been set up. Several Image Processing methods have been designed, implemented and published (two conferences papers, one workshop).
The PARADIS method has been presented at a workshop, a conference and to the IMSMA management team. The main conclusion is that RS Data, especially high and very high resolution data, are very useful in this context.