Outlier Detection of Crowdsourcing Trajectory Data Based on Spatial and Temporal Characterization

As an emerging type of spatio-temporal big data based on positioning technology and navigation devices, vehicle-based crowdsourcing has become a valuable trajectory resource. However, been collected by non-professionals with multiple measurement terminals, resulting in certain errors collection. In these cases, to minimize the impact outliers obtain relatively accurate data, it is crucial detect clean outliers. This paper proposes efficient outlier detection (CTOD) method that detects from sequence both spatial view temporal view. Specifically, we first use adaptive clustering algorithm Delaunay triangulation (ASCDT) remove location offset points sequence. After that, most basic attributes points, 6-dimensional movement feature vector constructed for each point as input. The feature-rich reconstructed using proposed convolutional network autoencoder (TCN-AE), Squeeze-and-Excitation (SE) channel attention mechanism introduced. Finally, effectiveness CTOD experimentally verified.