Abstract: On the basis of introduction of map-matching problem and the uncertainty of the vehicle location information from the sensor, a real-time map-matching algorithm based on the curve-fitting and quasi-projection is put forward. By introducing isolated point tracking and GPS point of defection filtering, restriction caused by half-baked/incomplete GIS map database and errors in coordinate conversion can be eliminated. This algorithm may reduce operation complexity，and can run normally with incomplete GIS map data and partial traffic network data defections, which has better adaptability in real-time vehicle supervision and navigation system. The experimental results of the practical runs show that the new algorithm has good robustness and real-timeness.

Key words: map-matching, intelligent transport system(ITS), vehicle supervision, vehicle navigation

摘要： 在引入地图匹配问题的基础上，针对传感器给出的车辆位置信息的不确定性，提出了基于准投影和曲线拟合的实时地图匹配算法，以适应车辆监控导航实时性的要求。该算法以选取与车辆行进方向一致的道路为核心，利用孤立点轨迹提取和GPS坏点过滤，取消由电子地图数据不完整和坐标变换误差引起的限制，简化运算的复杂度，并能在电子数据不完整或局部路网数据错误的情况下正常运行，从而为车辆监控导航准确性研究提供了基础。实验表明，该算法能够适应实时监控导航系统的需求，而且具有良好的实时性和鲁棒性。

关键词: 地图匹配, 智能交通系统, 车辆监控, 车辆导航