Science Service System
You are here : Home : Proposals_Summary

Summary of Proposal MTH1028

TitlePartially supervised change detection in high resolution optical images and radar data for damage assessment after earthquake events
Investigator Buchholz, Tim - European Space Agency (ESA) ESRIN, EO Science, Applications and Future Technologies Department
Team Member
Dr. Marconcini, Mattia - European Space Agency (ESA) ESRIN, EO Science, Applications and Future Technologies Department
Dr. Fernández Prieto, Diego - European Space Agency (ESA), EO Science, Applications and Future Technologies Department
SummaryIn recent years the vast amount of classification tasks of satellite imagery were conducted with the help of fully supervised classifiers. Depending on the topic this approach was useful, but several research areas require just the knowledge of one (or a few) class(es) of interest. Especially for change detection purposes the partially supervised classification approach is useful for assessing the changes in one class or changes in one class to a second class. In this project a partially-supervised approach for one class classification and the further change detection will be used to detect changes in buildings and other man made structures after an earthquake event. The benefit of this approach should be assessed for damage mapping before and after a mayor earthquake event. The application of damage assessment requires often just the knowledge about one or a few classes. Therefore avoiding considering other classes than the class of interest could be an asset, because the necessary ground truth for all classes in an image is often hard to gain or not available. The method should be sensor independent, due to that several different sensor combinations (optical and radar) should be assessed.

Back to list of proposals

© DLR 2004-2016