Affiliations: [a] Department of Geography, Faculty of Literature and Human Science, University of Hormozgan, Bandar Abbas, Iran
| [b] Department of Geospatial Information Systems, Faculty of Geodesy and Geomatics Engineering, K. N. Toosi University of Technology, Tehran, Iran
| [c] Department of Surveying Engineering, Faculty of Civil Engineering and Transportation, University of Isfahan, Isfahan, Iran
Correspondence:
[*]
Corresponding author. Mohammad Sharif, Department of Geography, Faculty of Literature and Human Science, University of Hormozgan, Minab Road, 3995, Bandar Abbas, Hormozgan, Iran. E-mail: mvsharif@gmail.com.
Abstract: Movements of objects take place in different contexts and their trajectories are highly influenced by the contexts. Several studies have been conducted in the last decade on similarity measuring of raw trajectories, but very few have used context information in this process. Because the context information is collected from multifarious sources, it is qualitatively and quantitatively heterogeneous and uncertain. Therefore, the current distance functions are unable to measure the similarities between trajectories by considering the heterogeneous context information. This article presents a new context-aware hybrid fuzzy model, named CaFIRST, to measure the similarity of trajectories by considering not only the spatial footprints of moving objects but also various types of internal and external context information. CaFIRST is able to handle multi-size trajectories that are contextually enriched by both quantitative (numeric) and qualitative (descriptive) values. The performance of CaFIRST was examined using two real data sets, obtained from pedestrians and cyclists in New York City, USA. The results showed the robustness of CaFIRST for quantifying the commonalities in multivariate trajectories and its sensitivity to small alterations in context information. Furthermore, the effects of internal and external context information on similarity values are shown to be remarkable.
