A Constrained K-Nearest Neighbor Approach for Semantic Indoor Trajectory Extraction

摘要

One of the indoor location-based services use cases is giving real-time information to users while they are inside a building. For example, in a museum, they can check next-to-visit recommendations or emergency guidance in real-time. Thus, it is vital to extract their movement as indoor trajectories. However, most indoor positioning methods have some degrees of error due to noisy observations. Those errors may lead to distant consecutive positions, which is impossible in a real-world case. Thus, we propose the semantic space-based movement constraints to produce indoor semantic trajectory without distant consecutive positions. The proposed approach is designed to extract semantic trajectory effectively in a real-time manner.