Trajectory Similarity Search in Spatial Networks

Abstract

Abstract

A New Similar Trajectory Search Algorithm Based on Spatio-Temporal Similarity Measure for Moving Objects in Road Networks

“…In general, we can represent a spatial network in terms of a directed graph G(V, E), where V is the set of vertices (junctions or intersections of the road network) and E is the set of edges (road segments) [5]. Hence, we can define a trajectory T of a moving object in a spatial network as t 2 ), .…”

“…Compared with the work of Hwang et al [3] and Chang et al [4], our algorithm has no requirement that a data trajectory should share at least a common node with a query trajectory to be a similar one. Compared with the work of Tiakas et al [5], which calculates nearly the same similarity for most of trajectories in a bi-directional spatial network, our algorithm provides a precise similarity measure by using a real network distance and thus is suitable for both a uni-directional and a bi-directional spatial network. In addition, our similar trajectory search algorithm can achieve good retrieval performance because it can prune out those trajectories which never become candidates for similar ones by using grid cell distances.…”

“…There exists a small volume of work on similar trajectory search for moving objects by considering spatial network distance [3]- [5]. Hwang et al [3] proposed a similar trajectory search algorithm in a road network, based on identical POIs (Points of interest) and TOIs (Times of interest) of the query trajectory.…”

“…However, the algorithm only considers similar Manuscript trajectories as those which share at least one common node with a query trajectory. In addition, Tiakas et al [5] proposed another type of similarity measure and a similarity search algorithm using an M-tree to search for similar trajectories of moving objects in a spatial network. They defined a spatial similarity measure as an average ratio of the minimum distance and the maximum distance between a query's node and a data trajectory's node.…”

A New Signature-Based Indexing Scheme for Efficient Trajectory Retrieval in Spatial Networks

“…Since moving objects move on predefined spatial networks instead of on Euclidean spaces, they exhibit similarity with respect to their trajectories, i.e., traversed paths [5]. Therefore, two moving objects can be correlated based on their trajectories.…”