This month, we are taking a look at the use of eCognition within the humanitarian sector. The January 2018 edition of GeoConnexion International Magazine published an article entitled “High-Level Humanitarian Support” about a project done by the University of Salzburg’s Department of Geoinformatics (Z_GIS) in association with Médecins Sans Frontières (MSF).
A key objective of the project is the automatic extraction of population information from satellite images “because crises can produce rapid changes in populations, it’s essential to be able to develop accurate data quickly and to repeat the measurements as needed” according to Dr. Stefan Lang (Z_GIS).
Z_GIS is acquiring VHR satellite imagery (< 1m resolution), from providers such as Digital Globe and Airbus, which are processed in Trimble’s eCognition software to estimate the number of dwelling within a refugee camp at a given time. Furthermore, the processing can be repeated over time due to the automated nature of the eCognition rule set giving team members an understanding of population change and movement.
Working with eCognition, the Z_GIS team has developed specialized rule sets “that effectively train eCognition to recognise and classify individual features within an image”, in this case the goal is to differentiate between dwellings from other camp structures. The initial work, developed based on QuickBird imagery from the Zam Zam camp in Darfur has been successfully transferred to other camps and integrated additional sensor types such as the GeoEye-1 satellite, demonstrating the “greater flexibility and efficiency” offered by eCognition’s Object-based Image Analysis (OBIA).
Since refugee camps are changing quickly, it is important that the rule set account for the rapid acquisition and processing of satellite data coming from various sensors. According to Dr. Dirk Tiede (Z_GIS), “there are a lot of situations where we need the software to adapt quickly”. In some cases, new maps and results are being delivered every few months. This requires the full integration of both eCognition Developer and Server – eCognition Server provides the user with the means to expand their processing capabilities through batch processing. In the past this was a bottleneck for time sensitive projects. “The distributed computing possibilities of eCognition Server” are able to process “an entire VHR satellite scene in just a few minutes”.
This on-going project is an excellent example of the value earth observation data can play in humanitarian aid applications. Unfortunately, natural disasters and other humanitarian crises will continue to occur throughout the world. Hopefully, we can improve the relief provided by NGOs like MSF and the Red Cross by getting advanced tools like eCognition into the hands of dedicated specialists.