Adaptive Ship Detection for Single-Look Complex SAR Images Based on SVWIE-Noncircularity Decomposition

• Published in: Sensors (Basel, Switzerland) Switzerland), 2018-09, Vol.18 (10), p.3293
• Main Authors: Zhao, Yu-Huan, Liu, Peng
• Format: Article
• Language: English
• Subjects: ship detection; SVWIE-noncircularity decomposition; synthetic aperture radar
• ISSN: 1424-8220; 1424-8220
• DOI: 10.3390/s18103293

In this paper, we present an adaptive ship detection method for single-look complex synthetic aperture radar (SAR) images. First, noncircularity is analyzed and adopted in ship detection task; besides, similarity variance weighted information entropy (SVWIE) is proposed for clutter reduction and target enhancement. According to the analysis of scattering of SVWIE and noncircularity, SVWIE-noncircularity (SN) decomposition is developed. Based on the decomposition, two components, the high-noncircularity SVWIE amplitude ( ) and the low-noncircularity SVWIE amplitude ( ), are obtained. We demonstrate that ships and clutter in SAR images are different for detector and detector can be effectively used for ship detection. Finally, to extract ships from the background, the generalized Gamma distribution (G Γ D) is used to fit statistics of clutter and the constant false alarm rate (CFAR) is utilized to choose an adaptive threshold. The performance of the proposed method is demonstrated on HH polarization of Alos-2 images. Experimental results show that the proposed method can accurately detect ships in complex background, i.e., ships are close to small islands or with strong noise.