Description
Every year thousands of people are displaced by conflicts or natural disasters. Knowing how many people are gathered in camps is crucial for an efficient relief operation. However, it is often difficult to collect exact information on the total number of the population. Very high resolution satellite data can provide accurate estimates of the number of dwellings in such camps as a proxy for the total population. The goal of this project is to improve the dwellings detection methodology and to develop an operational stand-alone application in order to provide reliable and consistent results, independent of the imaging sensors type and different environmental and exposition conditions. The main topic of the dedicated technology enhancement is the improvement and automation of pre-processing approach, feature extraction algorithm and results verification. The final outcome of this project is an application for decision support in emergency response, ready to use by the European Commission entities, humanitarian aid agencies, UN organizations.
Summary of project results
Every year thousands of people are displaced by conflicts or natural disasters. Knowing how many people are gathered in camps is crucial for an efficient relief operation. However, it is often difficult to collect exact information on the total number of the population. Very high resolution satellite data can provide accurate estimates of the number of dwellings in such camps as a proxy for the total population. The goal of this project was to improve the dwellings detection methodology and to develop an operational stand-alone application in order to provide reliable and consistent results, independent of the imaging sensors type and different environmental and exposition conditions. The goal of this project has been reached. The dwellings detection methodology has been improved and an operational stand-alone application has been developed. The main topic of the dedicated technology enhancement is the improvement and automation of pre-processing approach, feature extraction algorithm and results verification. The analysis provides reliable and consistent results, independent of the imaging sensors type and different environmental and exposition conditions. The final outcome of this project is an application, ready to use by a European Commission entities, humanitarian aid agencies, nongovernmental relief organizations and United Nations organizations for decision support in emergency response. The main features of the stand-alone application can be presented as follows: • Automatic assignment of a image class in a data hierarchy tree; • Manually change of threshold for “high-texture” class in camp border delineation procedure; • Manually change of threshold for “vegetation” class in vegetation mask calculation procedure; • Automatic calculation of dwellings size by applying the granulometry, i.e. structures size distribution; • Automatic calculation of intensity threshold by verifying the cross section of structures and their surrounding on satellite imagery; • Manually selection of Structural Element for containment criterion; • Manually selection of area size (amount of randomly selected samples) to be verified in results quality assessment; • Interface for visual interpretation; • Calculation of correlation coefficient and main statistics.
Summary of bilateral results