中文摘要：

针对SAR（synthetic aperture radar）ATR（auto target recognition）算法中的ROI（region of interest）提取通常由一个CFAR（constant false alarm rate）检测器和聚类算法来完成,该方法在高波段SAR目标检测中具有优良的性能,而在UWB SAR（ultra-wide band synthetic aperture radar）叶簇遮蔽目标检测中效果不佳。提出了一种适于叶簇遮蔽目标检测的ROI提取方法,该方法由小滑窗中值滤波、低门限CFAR检测、形态学操作和聚类算法四部分组成,能够在叶簇遮蔽目标检测中很好地完成ROI提取,基于实际UWB SAR图像的ROI提取结果验证了该方法的有效性。

英文摘要：

ROI （region of interest） extract in SAR （synthetic aperture radar） ATR （auto target recognition） algorithm is usually accomplished by a CFAR （Constant False Alarm rate） detector followed by a clustering algorithm. It has an excellent performance on target detection in high frequency SAR images. But it isn＇t effective to extract ROI when it is used to detect foliage-concealed targets in UWB SAR images. So a ROI extract method is proposed for the foliage-concealed target detection. The proposed method is composed of a median filter of small window; a CFAR detector of low threshold; a morphologic operation and a clustering algorithm. It is effective to complete ROI extract in the foliage-concealed target detection. The ROI extract results of different extract methods based on a factual UWB SAR image testify the efficiency of the proposed method.