A Map Matching Based Framework to Reconstruct Vehicular Trajectories from GPS Datasets
Roniel Soares de Sousa, Azzedine Boukerche, Antônio A. F. Loureiro · 2020
This paper proposes a new framework to reconstruct vehicle trajectories from GPS datasets. GPS-embedded vehicles generate massive vehicular trajectory data that are crucial for many applications, such as route recommendation, traffic analysis, urban planning, and Intelligent Transportation Systems (ITS). In order to achieve that, it is necessary to employ processing techniques such as noise filtering, segmentation, and map matching to prepare the data to be properly used by real applications. The proposed framework aims to reconstruct the trajectories completely, even from low sampled data, which naturally present gaps. Experimental results show the effectiveness and efficiency of the framework to reconstruct trajectories with different sampling rates and different characteristics.