基于航拍图像的自动桥梁检测

在空中打击中，河流上的桥梁作为主要的交通枢纽成为导弹重要的攻击目标。本文算
法使用高斯拉普拉斯算子和方向统计直方图的方法检测得到河流、公路和桥梁上的兴趣点和兴趣点主方向，兴趣点沿主方向生长得到河流、公路和桥梁等目标区域。然后根据兴趣点的主方向对河流、公路和桥梁区域进行边缘检测，进一步分割出河流、公路和桥梁的边缘像素与主干像素。考察主干像素判断是否有较为宽阔的河流存在，如果存在河流则对比边缘像素和主干像素从中分割出桥梁并确定桥梁位置。上述算法能够有效地从航拍图像中识别有一定宽度的河流，并能准确地检测出桥梁的存在和位置。
关键词：图像处理；图像分割；兴趣点；主方向；河流提取；桥梁检测
Abstract: At long range air assault with missiles, the bridges as important traffic hubs are principal military target. The algorithm to extract bridges firstly detects the interest points and their main directions of aerial image by direction scanning and LOG detector and the roads, rivers and bridges are roughly recognized. And then according to main directions of interest points the algorithm divides to the edges and trunks of the roads, rivers and bridges. The algorithm detect if the broad rivers and the bridges exist. If exist, the bridges are extracted and theirs place are exactly located by comparing edges image and trunks image. The algorithm has highly veracity and reliability. Interest Points
Key Words: image processing; image division; interest points; main direction; bridge
extraction; bridges detection