2006 · 15 citations · 2 references
ComputingEarth SciencesGeoinformaticsComputer VisionGeographic Information SystemsRemote SensingGeospatial Computer VisionRemote Sensing Image AnalysisRemote Sensing ImagerySatellite Image InterpretationAccuracyAutomated Image-based MonitoringCell CountingHigh-resolution Satellite Image AnalysisHigh-resolution Satellite Imagery AnalysisMathematical MorphologySatellite Image AnalysisSatellite Imagery Analysis
In the framework of the European project GMOSS (Global Monitoring for Security and Stability) the CRPSM (Centre di Ricerca Progetto San Marco) is developing automatic procedures for detecting and counting dwelling units in refugee camps. The possibility of monitoring refugee camps, using very high spatial resolution satellite (VHSR) images, has been already demonstrated, in the past, by several authors. This paper aims at reporting our results on the way of developing new algorithms to improve the performances of previous techniques in terms of detection accuracy and to extend their applicability to a broader set of background conditions. The algorithms developed by CRPSM to estimate, in an as automatic as possible way, the refugee camps population (tents) are based on the mathematical morphology (MM). The method has been applied to Ikonos and Quickbird VHSR images of several refugee camps (Goz Amer, Mille, Lukole, etc) located in Africa, to solve the tents counting problem. The error associated with these techniques, computed comparing the number of tents detected by the different algorithms with the one obtained by a visual counting, results satisfactorily low. The developed techniques have been compared with other already available methods (eCognition, etc.) and the results are also described.
2