改进的无人机影像处理技术在新冠疫情复工建设中的应用

疫情期间复工复产需要准确的现场信息及高精度分辨率的正射影像作为决策依据，但管理人员无法通过密集的现场巡视获取信息。无人机外业数据采集速度快、采集精度高、数据量全，但传统的无人机正射影像处理速度无法满足项目刻不容缓的需求，因此本文对无人机外业快速采集-快速内业处理技术-目标成果识别作了全方位的研究，基于改进SURF无人机影像快速处理算法并针对大型工程设备在影像上呈现规则矩形、特征点突出、明暗对比关系明显等关键要素，快速处理得到正射影像，并在此基础上运用Fast R-CNN网络识别现场机械设备为复工决策提供决策依据，相应成果已在新冠疫情期间雄安高铁站工程建设复工复产中得到了应用。

Application of improved UAV image processing technology in the construction of COVID-19

During the epidemic period, it needs accurate field information and high-resolution orthophoto image as the basis for decision-making, but managers can not obtain information through intensive field inspection. The UAV field data acquisition speed is fast, the acquisition precision is high, the data quantity is all and so on superiority, but the traditional UAV orthophoto image processing speed cannot satisfy the project urgent need, therefore this article has made the omni-directional research from the UAV field field fast acquisition-the fast internal processing technology-the target achievement recognition, based on the improvement SURF UAV image fast processing algorithm, the method has the advantage of being fast. According to the advantages of fast and the key elements such as regular rectangle, prominent feature points and obvious contrast between light and shade in the image of the major engineering equipment, the orthophoto image can be quickly processed and fast can be used on this basis R-CNN network identifies on-site mechanical equipment to provide decision-making basis for the resumption of work. The corresponding results have been applied in the resumption of work of Xiong'an high-speed railway station during the period of COVID-19 epidemic.