目前关于机场检测的研究已有很多，但大多以红外或可见光遥感图像为处理对象，而对合成孔径雷达（SAR）图像的处理主要集中在含有机场的小场景情况。针对大场景高分辨率SAR图像中机场快速检测问题，在现有检测方案的基础上提出了一种新的机场检测方法。首先对原始图像进行预处理，然后对图像进行自适应阈值分割并提取图像中的感兴趣区（ROI），最后提取ROI的尺寸、形状、对比度和拓扑特征并送入二叉决策树进行辨识。实验结果表明，该检测方法具有较高的准确率和较快的处理速度。

At present, there has been many researches on airport detection, but most of them are about infrared or optical images, and the research about synthetic aperture radar（SAR）images are mostly focusing on small scale instances. Aiming at large-scale SAR image with high resolution, a new method was proposed based on the existing detection scheme. Firstly, pre-process is implemented on the image. Then suspicious regions of interesting（ROI）are extracted from the pre-process result. Finally, all ROIs are identified by using four features and binary decision tree. Experiments show that the method can satisfy the request of speed with high detection accuracy.