合成孔径雷达成像的最终目的是对成像场景中感兴趣的目标进行有效的检测、识别和监视。该文针对实际复杂地面环境场景中的人造目标检测问题,提出一种有效的SAR图像目标检测方法。该方法提出一种基于信息熵的改进Lee滤波算法,在图像平滑的同时,保持图像的细节和邻域特征信息;然后,采用门限化分割技术对高分辨SAR图像中的目标进行粗略检测;最后采用方向梯度能量分形特征提取方法,对图像中目标的尺寸和形状进行准确的估计,实现对目标的精确检测。实测数据的处理结果验证了该方法的有效性。
The aim of Sythetic Aperture Radar (SAR) imaging is to detect, identify and track the target in the complicated scene. In the paper, a new method is proposed to detect the man-made target in the complicated background. First, the improved Lee filtering algorithm using information entropy is proposed to eliminate the effect of noise and keep the detailed characteristic; then, a threshold-based segmentation technique is employed to do a cursory target detection in the high-resolution SAR image; finally, the characteristic (the size and the shape) of target is picked up based on the Directed Gradient Power (DGP). After above operation, the target is accurately detected. Meanwhile, the measured data is used to validate the effectiveness of the method.