遥感影像数据与地理信息系统(geographic information system,GIS)矢量数据的配准是遥感与GIS集成的基础。目前遥感影像与矢量数据的配准关键在于遥感影像特征的提取,而现有遥感影像特征提取方法存在特征提取不完整、配准失败和精度不...遥感影像数据与地理信息系统(geographic information system,GIS)矢量数据的配准是遥感与GIS集成的基础。目前遥感影像与矢量数据的配准关键在于遥感影像特征的提取,而现有遥感影像特征提取方法存在特征提取不完整、配准失败和精度不高等问题。由此提出了一种基于Mask R-CNN(region-based convolutional neural network)的遥感影像与矢量数据配准方法,首先,利用Mask R-CNN模型提取影像的道路交叉口作为影像控制点;然后,依据几何拓扑关系筛选矢量数据道路交叉口作为矢量控制点,再根据遥感影像与矢量数据控制点的欧氏距离确定同名控制点;最后,以同名控制点为基础实现遥感影像与矢量数据的配准。选取上海市矢量数据和高分二号影像数据进行配准实验,实验结果表明,所提方法鲁棒性强、精度高。展开更多
As the original Global Position System (GPS) data in Floating Car Data have the accuracy problem,this paper proposes a heuristic path-estimating algorithm for large-scale real-time traffic information calculating. It ...As the original Global Position System (GPS) data in Floating Car Data have the accuracy problem,this paper proposes a heuristic path-estimating algorithm for large-scale real-time traffic information calculating. It uses the heuristic search method,imports the restriction with geometric operation,and makes comparison between the vectors composed of the vehicular GPS points and the special road network model to search the set of vehicular traveling route candidates. Finally,it chooses the most optimal one according to weight. Experimental results indicate that the algorithm has considerable efficiency in accuracy (over 92.7%) and com-putational speed (max 8000 GPS records per second) when handling the GPS tracking data whose sampling rate is larger than 1 min even under complex road network conditions.展开更多
基金the National Basic Research Program of China ("973") (Grant No. 2005CB321900)the National Hi-Tech Research and Devel- opment Program of China (Grant No.2006AA12Z315)
文摘As the original Global Position System (GPS) data in Floating Car Data have the accuracy problem,this paper proposes a heuristic path-estimating algorithm for large-scale real-time traffic information calculating. It uses the heuristic search method,imports the restriction with geometric operation,and makes comparison between the vectors composed of the vehicular GPS points and the special road network model to search the set of vehicular traveling route candidates. Finally,it chooses the most optimal one according to weight. Experimental results indicate that the algorithm has considerable efficiency in accuracy (over 92.7%) and com-putational speed (max 8000 GPS records per second) when handling the GPS tracking data whose sampling rate is larger than 1 min even under complex road network conditions.