Clustering Based Place Recognition (CluPlaR)

Algorithm

The CluPlaR algorithm combines individual geopositions (PositionVisit) for place visits (PlaceVisit) and their connections (Track). Fundamental for this aggregation is the clustering of the positions based on the time spent at a location and its radius. The PlaceVisit graph is created by the aggregation, which shows the chronological sequence of the place visits. Based on this graph, the place visits can be further aggregated to locations (Place) that are connected by Connections. The resulting PlaceGraph is time-independent.

The following animation shows how the CluPlaR algorithm works.

Application

The CluPlaR algorithm is part of an application for clustering spatial-temporal data and visualize them. The application processes geo positions saved in KML files. By adding values for residence time and radius of a place, the geo positions are clustered to PlaceVisits and Places.