Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Generalized interpretation scheme for arbitrary HR InSAR image pairs

: Boldt, Markus; Thiele, Antje; Schulz, Karsten

Volltext urn:nbn:de:0011-n-2649535 (2.1 MByte PDF)
MD5 Fingerprint: 86ff2aeac4f3f2f233d1255d81d2ce91
Copyright 2013 Society of Photo-Optical Instrumentation Engineers. One print or electronic copy may be made for personal use only. Systematic reproduction and distribution, duplication of any material in this paper for a fee or for commercial purposes, or modification of the content of the paper are prohibited.
Erstellt am: 31.10.2013

Michel, U. (Ed.) ; Society of Photo-Optical Instrumentation Engineers -SPIE-, Bellingham/Wash.:
Earth Resources and Environmental Remote Sensing/GIS Applications IV : 23. September 2013, Dresden, Germany
Bellingham, WA: SPIE, 2013 (Proceedings of SPIE 8893)
ISBN: 978-0-8194-9762-8
Paper 889305
Conference "Earth Resources and Environmental Remote Sensing/GIS Applications" <4, 2013, Dresden>
Konferenzbeitrag, Elektronische Publikation
Fraunhofer IOSB ()
SAR; high resolution; InSAR; interpretation scheme; classification; CovAmCoh; k-means; clustering

Land cover classification of remote sensing imagery is an important topic of research. For example, different applications require precise and fast information about the land cover of the imaged scenery (e.g., disaster management and change detection). Focusing on high resolution (HR) spaceborne remote sensing imagery, the user has the choice between passive and active sensor systems. Passive systems, such as multispectral sensors, have the disadvantage of being dependent from weather influences (fog, dust, clouds, etc.) and time of day, since they work in the visible part of the electromagnetic spectrum. Here, active systems like Synthetic Aperture Radar (SAR) provide improved capabilities. As an interactive method analyzing HR InSAR image pairs, the CovAmCohTM method was introduced in former studies. CovAmCoh represents the joint analysis of locality (coefficient of variation – Cov), backscatter (amplitude – Am) and temporal stability (coherence – Coh). It delivers information on physical backscatter characteristics of imaged scene objects or structures and provides the opportunity to detect different classes of land cover (e.g., urban, rural, infrastructure and activity areas). As example, railway tracks are easily distinguishable from other infrastructure due to their characteristic bluish coloring caused by the gravel between the sleepers. In consequence, imaged objects or structures have a characteristic appearance in CovAmCoh images which allows the development of classification rules. In this paper, a generalized interpretation scheme for arbitrary InSAR image pairs using the CovAmCoh method is proposed. This scheme bases on analyzing the information content of typical CovAmCoh imagery using the semisupervised k-means clustering. It is shown that eight classes model the main local information content of CovAmCoh images sufficiently and can be used as basis for a classification scheme.