【摘要】 运用众源车辆轨迹数据提取道路信息需要解决轨迹点采样稀疏、高噪音、密度差异大等问题。为此,本文提出一种运用约束Delaunay三角网从车辆轨迹线集中提取道路边界的方法。首先,通过三角形边长度和Voronoi面积等几何特征表达轨迹点分布的聚集性差异,并将这两种不同几何维数的控制条件集成建立道路边界识别模型,运用"种子点"区域扩展方法实现道路边界的精确提取。最后,运用北京市出租车GPS轨迹进行试验,结果表明该方法适于车辆分布频率悬殊、时间跨度不同、道路网结构复杂的轨迹线数据处理。更多还原

【Abstract】 Extraction of road boundary accurately from crowdsourcing trajectory lines is still a hard work.Therefore,this study presented a new approach to use vehicle trajectory lines to extract road boundary.Firstly,constructing constrained Delaunay triangulation within interpolated track lines to calculate road boundary descriptors using triangle edge length and Voronoi cell.Road boundary recognition model was established by integrating the two boundary descriptors.Then,based on seed polygons,a regional growing method was proposed to extract road boundary.Finally,taxi GPS traces in Beijing were used to verify the validity of the novel method,and the results also showed that our method was suitable for GPS traces with disparity density,complex road structure and different time interval.更多还原