顾及几何与场景结构相似度的管线数据多步骤匹配算法

A multi-step pipeline data matching algorithm considering the similarity of geometry and scene structure

针对城市管线数据重复探测,综合管线数据生产成本高的问题,该文提出一种几何相似度与空间场景结构相似度相结合的管线数据多步骤匹配算法。以路划(stroke)作为匹配单元,使用长度、角度和空间距离指标通过分配不同的权重和阈值来计算几何相似度,将几何特征与邻域空间的空间拓扑关系相结合计算空间场景结构相似度。设计基于弧段分解的stroke部分匹配算法最大限度地保留了原有的两组地下管线数据。设计基于顶点分解的stroke部分匹配算法可以解决单线间局部匹配中的问题(如生长、缩短、拓扑变化等)。实验结果表明,该文所提算法具有较高的匹配精度、召回率和运行效率,研究结果可以有效降低综合管线数据的生产成本,提高专业管线数据质量。

In order to solve the problem of duplicate detection of urban pipeline data and high production cost of comprehensive pipeline data,a multi-step pipeline data matching algorithm was proposed in this paper,which combined geometric similarity with spatial scene structure similarity.With stroke as the matching unit,computational geometry similarity was calculated by giving different weights and thresholds to length,angle and spatial distance indicators.Spatial scene structure similarity was confirmed by combining geometric features with the spatial topological relationship of neighborhood space.The data of the original two sets of underground pipelines was retained to the maximum extent by the design of the stroke partial matching algorithm based on arc decomposition.The problems in local matching between individual lines were solved through the stroke partial matching algorithm based on vertex decomposition(such as growth,shortening,topology changes,etc.).The experimental results showed that the algorithm proposed in this article had high matching accuracy,recall rate,and operational efficiency.At the same time,the production cost of comprehensive pipeline data Was effectively reduced,and the quality of professional pipeline data was improved.