Airport detection in synthetic aperture radar (SAR) images has attracted much concern in the field of remote sensing. Affected by other salient objects with geometrical features similar to those of airports, traditional methods often generate false detections. In order to produce the geometrical features of airports and suppress the influence of irrelevant objects, we propose a novel method for airport detection in SAR images. First, a salient line segment detector is constructed to extract salient line segments in the SAR images. Second, we obtain the airport support regions by grouping these line segments according to the commonality of these geometrical features. Finally, we design an edge-oriented region growing (EORG) algorithm, where growing seeds are selected from the airport support regions with the help of edge information in SAR images. Using EORG, the airport region can be mapped by performing region growing with these seeds. We implement experiments on real radar images to validate the effectiveness of our method. The experimental results demonstrate that our method can acquire more accurate locations and contours of airports than several state-of-the-art airport detection algorithms.

中文翻译：

---

通过显着线段检测器和面向边缘的区域生长在 SAR 图像中进行机场检测

合成孔径雷达（SAR）图像中的机场检测引起了遥感领域的广泛关注。受其他具有与机场相似几何特征的显着物体的影响，传统方法经常会产生错误检测。为了产生机场的几何特征并抑制无关物体的影响，我们提出了一种新的SAR图像机场检测方法。首先，构造一个显着线段检测器来提取SAR图像中的显着线段。其次，我们通过根据这些几何特征的共性对这些线段进行分组来获得机场支持区域。最后，我们设计了一种面向边缘的区域生长 (EORG) 算法，该算法借助 SAR 图像中的边缘信息从机场支持区域中选择生长种子。使用 EORG，可以通过使用这些种子执行区域生长来绘制机场区域。我们在真实雷达图像上进行实验以验证我们方法的有效性。实验结果表明，与几种最先进的机场检测算法相比，我们的方法可以获得更准确的机场位置和轮廓。