摘要： 提出一种与数据有关的基与固定基相结合的合成孔径雷达(SAR)图像检测算法，分别提取2种不同空间下表示的目标特征，将这些特征在不同空间中融合并进行调制，形成显著图，从而将目标特征凸显出来，利用目标的先验知识检测目标，使用ADTS高分辨率机载SAR目标数据进行仿真。实验结果表明，该算法能够有效突出待测图像中的目标区域，并抑制非目标区域，具有较高的检测精度。

关键词: 非负矩阵分解, Gabor滤波器, 目标检测, 合成孔径雷达图像

Abstract: A Synthetic Aperture Radar(SAR) image detection algorithm which unifies with the data related base and the fixed base is proposed. The target features which are extracted in two kind of different spaces are fused separately. They are modulated each other. The target features can be protruded by this algorithm. The targets are detected by using the priori knowledge in the saliency map. Experimental results with ADTS target high-resolution airborne SAR data show that the target area can pop out with the background being restrained and the algorithm has higher detection accuracy.

Key words: Non-negative Matrix Factorization(NMF), Gabor filter, target detection, Synthetic Aperture Radar(SAR) image

中图分类号: 

• TP301.6