摘要
相似度指标的选取及稳健与否对匹配结果的正确性至关重要。针对来自同一地区不同来源的同名实体匹配问题,从道路网同名线目标匹配实际需要出发,研究基于豪斯多夫(Hausdorff)距离的线目标几何相似距离计算方法;并对其进行改进,提出一种基于短边的中位数Hausdorff相似距离计算指标;与传统Hausdorff距离和离散Frécet距离线目标相似性度量指标做比较,实验结果得出:基于短边中位数Hausdorff距离的相似性度量指标能更客观合理地描述同名线实体之间的几何相对距离,提高同名实体匹配的正确率。
According to the problem that one or several similarity measures are necessarily needed when determining whether two objects from different sources are corresponding objects that represents the same real world entities,the matching results matters a lot on whether the similarity measure is robust or not.In this paper,Hausdorff distance proximity between two objects was proposed,and the calculation of distance from line to line based on Hausdorff distance was improved.Then,an alternative distance similarity measure（the short-line median Hausdorff distance）,referred to as the SM_HD was put forward and compared to other several similarity measures.Experimental results showed that SM_HD distance could describe the relative geometric distance in a more objective and reasonable way,which was conducive to improve the matching accuracy.
出处
《测绘科学》
CSCD
北大核心
2016年第8期14-18,共5页
Science of Surveying and Mapping
基金
国家自然科学基金项目(41401528)
浙江省森林生态系统碳循环与固碳减排重点实验室开放基金资助项目(KFJJ2012004)
浙江农林大学大学生科技创新训练计划资助项目(201203015)
