Edge Detector for Polarimetric SAR ImagesUsing SIRV Model and Gauss-Shaped Filter
(English)In: IEEE Geoscience and Remote Sensing Letters, ISSN 1545-598X, E-ISSN 1558-0571Article in journal (Refereed) Submitted
The classic constant false alarm rate (CFAR) edgedetector with rectangle-shaped filter has been proven to beeffective and widely used in polarimetric SAR (PolSAR) images.However, in practical use, the assumption of complex Wishartdistribution is often not respected, especially in heterogeneousurban areas. In addition, as a simple smoothing filter, therectangle-shaped window is often shown to be easy to incur falseedge pixels near true edges. Therefore, its performance islimited. To overcome this restriction, we propose a new edgedetector for PolSAR images, which utilizes the sphericallyinvariant random vector (SIRV) product model to estimate thenormalized covariance matrix for each pixel and then replacethe rectangle-shaped filter with Gauss-shaped filter. Theperformance of our proposed methodology is presented andanalyzed on two real PolSAR data sets, and the results show thatthe new edge detector attains better performance than theclassic one, particularly for urban areas.
Place, publisher, year, edition, pages
Spherically invariant random vector (SIRV), urban areas, edge detection, Gauss-shaped filter, polarimetric SAR (PolSAR).
IdentifiersURN: urn:nbn:se:kth:diva-187943OAI: oai:DiVA.org:kth-187943DiVA: diva2:932589
QC 201606072016-06-012016-06-012016-06-07Bibliographically approved