低能见度前下视航拍图像中机场跑道的自动检测  

Automatic detection of airfield runways in forward-looking aerial images with low visibility

作　　者：梁超圣[1] 柳健[1] 王志成[2] 

机构地区：[1]华中科技大学电子与信息工程系,湖北武汉430074 [2]华中科技大学图像识别与人工智能研究所,湖北武汉430074

出　　处：《华中科技大学学报（自然科学版）》2005年第3期27-29,共3页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：高等学校博士学科点专项基金资助项目 (2 0 0 10 4 870 30 ) .

摘　　要：提出了一种针对低能见度、前下视航拍机场图像中自动检测机场跑道的方法 .首先 ,采用基于递归的Otsu分割方法 ,从复杂背景中提取机场的主要轮廓 ;然后 ,利用基于形态学的开运算和膨胀进行形态滤波 ,开运算是为了滤除一些孤立噪声点 ,而膨胀运算是为了将分割后的机场成分连成一个完整的整体 ;最后 ,对形态滤波处理后的图像进行其区域生长 ,从而对机场主跑道进行定位 .试验证明 :这种方法对低能见度、前下视的航拍机场图像能够快速准确地检测、定位机场主辅跑道 .To solve the problem of detecting the airfield runways automatically in forward looking aerial images with low visibility, a method was proposed. The algorithms can be cast into three steps. The recursive segmentation method based on Otsu was used to extract the main outline of airfield image from complex clutters. Mathematical morphology methods such as dilation and opening operations were used to deal with the airfield image segmented by recursive Otsu method. Opening operation can remove some isolated noise while dilation operation can connect the airfield components into a whole. Region growth method was used to track and locate the primary and secondary runways of airfield. The experiments show that the algorithms have good performance of detecting and locating airfield runways in forward looking aerial images with low visibility.

关 键 词：前下视航拍图像 低能见度 自动检测 机场跑道 OTSU分割 形态滤波 区域生长 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]