Training Convolutional Neural Networks for Translational Invariance on SAR ATR

David Malmgren-Hansen, Rasmus Engholm, Morten Østergaard Pedersen

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

719 Downloads (Pure)

Abstract

In this paper we present a comparison of the robustness of Convolutional Neural Networks (CNN) to other classifiers in the presence of uncertainty of the objects localization in SAR image. We present a framework for simulating simple SAR images, translating the object of interest systematically and testing the classification performance. Our results show that where other classification methods are very sensitive to even small translations, CNN is quite robust to translational variance, making it much more useful in relation to Automatic Target Recognition (ATR) in a real life context.
Original languageEnglish
Title of host publicationProceedings of EUSAR 2016: 11th European Conference on Synthetic Aperture Radar
PublisherIEEE
Publication date2016
Pages459-462
ISBN (Print)978-3-8007-4228-8
Publication statusPublished - 2016
Event11th European Conference on Synthetic Aperture Radar (EUSAR 2016) - Hamburg, Germany
Duration: 6 Jun 20169 Jun 2016
Conference number: 11
http://conference.vde.com/eusar/2016/Pages/default.aspx

Conference

Conference11th European Conference on Synthetic Aperture Radar (EUSAR 2016)
Number11
CountryGermany
CityHamburg
Period06/06/201609/06/2016
Internet address

Fingerprint Dive into the research topics of 'Training Convolutional Neural Networks for Translational Invariance on SAR ATR'. Together they form a unique fingerprint.

Cite this