基于移动用户众包数据的室内平面图感知

Indoor Space Plan Perception Based on Mobile User Crowdsourcing Data

摘要: 针对大型室内环境空间结构和对象实体经常变化进而影响提供实时位置服务的问题，设计了一种利用移动用户众包位置数据感知室内平面图的方法。方法使用来自移动用户的众包数据进行室内平面图的重建，首先应用Voronoi图对室内空间进行空间区域划分，以寻找室内环境下空间的拓扑关系；然后利用拓扑关系约束有向无环图支持向量机(DAG-SVM)生成空间对象的几何特征；最后，联结拓扑关系和几何特征构建室内空间平面图。实验结果表明本文的方法能够感知室内空间平面图。

Abstract: Aiming at the problem that the spatial structure of large indoor environment and object entities change frequently to affect the provision of real-time location service, a method of using the mobile user crowdsourcing location data to sense the indoor floor plan is designed. The method uses crowdsourced data from mobile users to reconstruct the indoor floor plan. Firstly, the Voronoi map is used to divide the indoor space into space to find the topological relationship of the space in the indoor environment, then using the topological relationship to constrain the DAG-SVM algorithm to generate the geometric features of the spatial object; finally, the topological relationship and the geometric features are used to construct the indoor space plan. The experimental results show that the proposed method can sense the indoor space plan.