复杂路段的角度差和后续点地图匹配方法

针对仅依靠距离和轨迹与路径的相似性来判断正确道路的方法，在并行和交叉路段的复杂路网环境中，易匹配到相邻路段或部分不可达的路段，导致匹配错误的问题，提出采用线性回归模型的方法，其能更精确地描述道路形状在转弯处的变化，根据路段方向和移动对象移动的方向差判断并行或交叉路段，并通过参照多个后续点的匹配情况，实现复杂路段处的地图匹配，减少匹配错误。此外，还提出简化聚类的路网补全方法，可以解决部分GPS点周围缺失候选路段无法匹配的问题，并采用四叉树索引地图数据，提高效率。对比实验结果表明在轨迹点数较少时，与现有的基于隐马尔可夫模型的概率匹配方法相比，时间最多减少了62%,更适合时效性要求高的应用场景，与传统的几何匹配方法相比，匹配精度提高了3%,且更适合复杂路段匹配。

Map matching method for complex road sections via difference and subsequent points

Map matching is important in navigation, traffic analysis and other applications. In the complex road network environment of parallel and cross sections, it is easy to match the adjacent sections or some inaccessible sections, which leads to the wrong matching. The proposed linear regression model was more accurate to describe the change of road shape at the turning. According to the difference between the road section direction and the moving direction of the moving object, it judged the parallel or intersecting road sections. By referring to the matching of multiple subsequent points, the map matching at complex sections could be realized, which could reduce the matching error. In addition, the proposed road network completion method based on simplified clustering could solve the problem that the missing candidate road sections around some GPS points could not be matched. And quad-tree structure was used to index map data to improve the efficiency. The results of comparative experiment showed that when the number of trajectory points was small, compared with the existing probabilistic matching method based on hidden Markov model, the time is reduced by 62%, which is more suitable for the application scenarios with high timeliness requirements. Compared with the traditional geometric matching method, the matching accuracy is improved by 3%, and it is more suitable for complex road matching.