一种从航空图片中自动检测桥梁的方法

A New Method for Detecting Bridges Automatically

如何自动、正确且快速地从遥感或航拍图片中检测战略目标 ,是导弹能否准确击中目标的关键 ,在军事领域具有重要意义。提出了一种从航片中对桥梁自动检测的算法。该方法从河流的识别入手 ,利用小波变换的阈值分割法和高斯 -拉普拉斯变换的分割法 ,将河流部分提取出来并二值化。然后对其进行多步噪声滤除从而使河流的提取更加完整。最后沿着河流搜索有无桥梁存在。实验结果表明 ,对于航空图片中存在的桥梁 。

In order to ensure that a missile can hit its target accurately, it is of key importance to be able to detect bridges automatically from aerial images. Past methods rely mainly on the features of the particular bridges. We employ a different approach based on the sharp contrast that exists naturally between a river and a bridge in aerial image. Sections 2 and 3 discuss in much detail the rough and fine segmentation respectively of a river; Fig.1 shows the steps of segmentation, with the river's skeleton in Fig.1(d) as the final result. Section 4 discusses briefly that, with the river's skeleton known, the bridges can be easily and automatically detected. Section 5 gives the results of three experiments as shown in Figs.2, 3 and 4 respectively; these results show clearly that our new method for detecting bridge automatically from aerial image appears to be feasible.