DocumentCode :
691157
Title :
Urban Area Detection for Polarimetric Synthetic Aperture Radar Image
Author :
Chen Zhi-Peng ; Xue Hui-feng
Author_Institution :
Coll. of Autom., Northwestern Polytech. Univ., Xian, China
fYear :
2013
fDate :
21-23 Sept. 2013
Firstpage :
1029
Lastpage :
1034
Abstract :
Polarimetric scattering feature parameters are significant for target interpretation in polarimetric synthetic aperture radar (PolSAR) images. To improve the polarimetric scattering similarity parameters (PSS), an enhanced polarimetric similarity parameter (EPS) is proposed in this paper. Besides emphasizing some scattering mechanisms, this novel parameter improves the contrast between one certain scattering type and other scattering types. Furthermore, by employing the sparsity feature of the enhanced span image, nonnegative matrix factorization (NMF) is modified to realize the urban area detection for PolSAR images. The effectiveness of this proposal is verified by real data experiments.
Keywords :
matrix decomposition; object detection; radar imaging; radar polarimetry; synthetic aperture radar; EPS; NMF; PSS; PolSAR images; enhanced polarimetric similarity parameter; enhanced span image; nonnegative matrix factorization; polarimetric scattering feature parameters; polarimetric scattering similarity parameters; polarimetric synthetic aperture radar image; scattering mechanisms; sparsity feature; target interpretation; urban area detection; Coherence; Eigenvalues and eigenfunctions; Feature extraction; Matrix decomposition; Urban areas; Vectors; double-bounce scattering; enhanced polarimetric similarity (EPS); nonnegative matrix factorization (NMF); polarimetric scattering similarity (PSS); polarimetric synthetic aperture radar (PolSAR);
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Instrumentation, Measurement, Computer, Communication and Control (IMCCC), 2013 Third International Conference on
Conference_Location :
Shenyang
Type :
conf
DOI :
10.1109/IMCCC.2013.229
Filename :
6840619
Link To Document :
بازگشت