利用浮动车GPS轨迹识别与提取城市道路交叉口

道路交叉口作为道路交汇的枢纽，是路网的重要组成部分，也是最重要的基础地理信息数据之一。浮动车GPS数据具有易获取、低成本和数据量大等优点，但工作同时伴随不少噪点。为了降低噪点对交叉口提取过程的影响，提高计算效率，本文运用KNN算法建立空间索引；计算向量夹角，判定道路出入口，粗筛取交叉口附近点；分别采用K-means算法、DBSCAN算法和层次算法进行聚类分析，进一步确定交叉口位置。最后以成都某区域浮动车GPS数据为例，提取道路交叉口并进行了对比分析，进一步表明本文方法可以服务于智能交通研究与应用。

Detection and extraction of urban road intersections using GPS trajectories of floating vehicles

Road intersection plays an important role in transport hub and road network, it's also the basic data of geographic information system. GPS trajectory data of floating vehicles is easy availability and low cost, but the data contains certain amounts of noises. To alleviate the impact of noise and improve compute efficiency of GPS data, this paper proposes a method to detect and extract urban road intersections. Construct data index based on the k-nearest neighbors (KNN) algorithm. Estimate the angles between vectors and coarsely detect road intersections. And then apply three clustering algorithms (K-means, DBSCAN and hierarchical clustering algorithms) to fine extracted road intersections. We test the proposed method on GPS dataset from Chengdu city to evaluate its performance. This research can be effectively used in practical intelligent transportation scenarios.