利用改进的Hausdorff距离匹配多尺度线要素

Multi-scale Linear Feature Matching Using Improved Hausdorff Distance

传统Hausdorff距离对噪声较敏感,在进行多尺度线要素匹配时,易导致漏匹配和误匹配。为提高匹配正确率,提出了一种改进的Hausdorff距离算法。针对部分匹配问题,采用曲线分割算法,以短曲线分割长曲线;针对点位分布差异问题,采用曲线加密算法,以匹配曲线的中间节点进行双向加密;针对曲线点集自身的噪声问题,以距离集合的中位数作为相似性指标,判断匹配要素是否为同名要素。选取不同尺度的行政区划界线和道路网数据进行匹配,以验证该算法。结果表明,该算法具有较好的匹配效果。

In the matching process of multi-scale linear feature,traditional Hausdorff distance is prone to be affected by noisy data.In order to improve the robustness of algorithm,we proposed an improved Hausdorff distance algorithm.Considering the problem of partial matching,we used the curve segmentation algorithm to segment the long curve with short curve.Aiming at the difference problem of point distribution,we used the curve encryption algorithm to gather the curve with the center node.With respect to the noise magnitude of curve points set,taking the median of distance set as the similarity index,we judged whether the matching features were the same.We selected different scales of administrative boundaries and road network data to validate the performance of proposed algorithm.The result indicates that the algorithm can effectively solve the matching problem of multi-scale linear feature.