You must have a javascript-enabled browser and javacript and stylesheets must be enabled to use some of the functions on this site.
 
   workshops     

 

A new approach for ship detection in SAR imagery based on correlations between different channels

Haiyan Li(1) and Yijun He(1)

(1) Institute of Oceanology, 7 Nanhai Road, Qingdao 266071, China

Abstract

Ship detection is a key requirement in the military, coastguard, fisheries, and commercial transportation sectors. Ship detection by synthetic aperture radar (SAR) has been studied extensively. The typical methods of Ship detection are K distribution clutter model with a constant false-alarm rate (CFAR) detector and ship wake detection with Radon transform by using single-polarization and single-band data. However´╝îradiometric information provided by traditional single-channel SAR is not generally sufficient in ship detection in that it could not fully characterize the scattering mechanisms of targets. More information is embedded in multi-channel, which has greater utilization in ship determination. Many researchers have explored the polarimetric information for ship detection. Two main types of polarimetric detection algorithms include coherent target decomposition and polarimetric entropy algorithms. Although these methods are effective for ship detection, they are limited only for full polarimetric data not for double-polarization (Envisat SAR) data. Aim to use dual-polarimetric SAR data sufficiently provided by Envisat especially the dual-polarimetric amplitude-only data, the paper proposes a new approach to detect ships. Different cross-correlation of ships and clutters between different polarimetric channels is vital to the method. The general expression for the polarimetric correlation coefficient is Ru, estimation of this parameter requires both amplitude and phase information. Owing to the data receiving process, almost Envisat data is amplitude-only data. To measure the degree of the correlation with amplitude data only, a two-dimensional cross correlation function (2D-CCF) is defined, which represents the degree of similarity between two images. The feasibility is verified by the simulated SAR images then the method is employed on a real SAR image. The proposed method takes advantages of the different statistical behavior among the ships and the surrounding sea, interpreting the information through the correlation in order to provide a more reliable detection. The method can be used under high sea state and have no relative to the movement of the ships. Moreover, the method is general which is not only suitable for dual polarization amplitude data but also for full polarization data and can be extended to multi-bands data. Although the validity of the method needs more datasets to characterize the performance under more general conditions, the analysis of the detection performance over both simulated and real images confirms the robustness of the proposed method.

 

 

  Higher level                 Last modified: 07.10.03