Automatic Construction of Indoor 3D Navigation Graph from Crowdsourcing Trajectories

Lacking indoor navigation graph has become a bottleneck in applications and services. This paper presents novel automated reconstruction approach from large-scale low-frequency trajectories without any other data sources. The proposed includes three steps: trajectory simplification, 2D floor plan extraction 3D construction. First, we propose ST-Join-Clustering algorithm to identify simplify redundant stay points embedded the trajectories. Second, an bitmap construction based on self-adaptive Gaussian filter is developed, then new improved thinning extract plans. Finally, present CFSFDP with time constraints topological connection between two different floors. To illustrate applicability of approach, conducted real-world case study using dataset over 4000 5 million location points. results showed that improves network accuracy by 1.83% 13.7% compared classical kernel density estimation approach.