基于无人机影像的施工道路信息提取方法

Construction Road Information Extraction Method based on UAV Images

针对传统施工进度监测主要依靠监测员进行实地踏勘、施工时效性低且项目经济成本随之增加等问题,提出基于无人机影像的高时效性、高可靠性的施工道路提取方法,用以辅助进度监测。首先,基于影像多尺度分割思想对道路DOM数据进行多尺度分割;其次,利用基于面向对象的多尺度分割方法和基于像元的监督分类法提取施工道路信息,并对提取道路的结果进行精度对比和分析;最后,利用其他施工路段的道路影像图对面向对象的多尺度分割方法的通用性和可行性进行验证。结果表明,基于面向对象的多尺度分割方法提取结果的总体精度为87.05%,高于基于像元的监督分类法,具有更高的提取精度,而且在不同施工路段均具有可行性。

To address the problems that traditional construction progress monitoring mainly relies on field survey by monitors, low construction timeliness and higher project economic cost, this paper proposes a highly time-efficient and reliable construction road extraction method based on UAV images to assist progress monitoring. Firstly, the road DOM is segmented based on the multi-scale image segmentation;Secondly, the construction road information is extracted using the object-oriented multi-scale segmentation method and the supervised classification method based on image elements, and the accuracy of the extracted road results is compared and analyzed;Finally, the generality and feasibility of the object-oriented multi-scale segmentation method are verified using the road images of other construction sections. The results show that the overall accuracy of the object-oriented multi-scale segmentation method is 87.05%, which is higher than that of the supervised classification method based on images, and has higher extraction accuracy and is feasible in different construction road sections.