摘要
针对传统道路养护依赖人工巡查、成本高且效率低的问题,提出以无人机航拍影像为数据源,实现公路标线的自动识别与精准提取。首先,依据路面颜色及梯度变化特征分割影像中的路面区域,并精确定位其连续部分;随后,融合FasterR-CNN网络与连通区域的颜色、面积双重特征,剔除干扰区域,保留完整标线。试验结果表明:该方法在复杂道路环境下仍保持高准确率,具有良好的适应性与鲁棒性。
In response to the problems of traditional road maintenance relying on manual inspections which are of high costs and low efficiency,in this paper,using drone aerial images as data sources to achieve automatic recognition and accurate extraction of highway markings is proposed.Firstly,segment the road surface area in the image based on its color and gradient changes and accurately locate its continuous parts.Subsequently,the Faster R-CNN network was integrated with the color and area features of the connected regions to eliminate interference areas and preserve the complete markings.The experimental results show that this method maintains high accuracy in complex road environments and has good adaptability and robustness.
作者
魏薪海
WEI Xinhai(Team 133 of Gansu Coalfield Geological Bureau,Baiyin 730910,China)
出处
《经纬天地》
2025年第4期37-40,共4页
Survey World
