恒虚警率（constant false alarm ratio，CFAR）目标检测法用于合成孔径雷达（synthetic aperture radar，SAR）图像时，通常要求图像有强的对比度，而实际上此条件很难满足。为提高SAR图像目标的检测率，特别是低信噪比图像，从SAR成像机理入手，结合非下采样Contourlet变换（nonsubsampled Contourlet transform，NSCT）理论的多尺度、多方向和平移不变性等特点，提出了一种新的SAR图像目标提取算法，即TD-NSCT（target detection based on NSCT）算法。该方法融合了CFAR检测器和NSCT的优点，通过对分解系数特征的选取和组合，达到改善SAR图像信噪比、提高SAR目标检测率的目的。用不同实际SAR图像数据和不同方法进行了比较实验，实验结果表明TD-NSCT算法能有效提高SAR目标的检测率，特别是对于那些隐藏地物目标的低信噪比SAR图像，证明TD-NSCT算法是一种可行和有效的SAR图像目标检测算法。%When the CFAR target detection method is used to SAR images,the CFAR method usually requires the SAR images have strong contrast.In fact,this condition is very difficult to be satisfied in practical applications.In order to improve the de-tection rate of SAR image target,especially the low SNR SAR images,this paper proposed a new SAR image target detection al-gorithm from the SAR imaging mechanism and combining the characteristics of the NSCT theory such as multi-scale,multi-di-rection and shift invariance.The new method was called the TD-NSCT method.This method integrated the advantages of CFAR detector and NSCT,and achieved the purpose of improving the SNR of SAR images and increasing the detection rate of SAR targets by selecting and fusing the decomposition coefficients.Some different SAR image data and different methods performed the comparative test.The experimental results show that the TD-NSCT algorithm can effectively improve the detection rate of SAR targets,especially those hidden targets in the low SNR SAR images.So the TD-NSCT algorithm is a feasible and effective method for SAR image target detection.
展开▼
