SST: Synchronized Spatial-Temporal Trajectory Similarity Search

Zhao, Peng; Rao, Weixiong; Zhang, Chengxi; Su, Gong; Zhang, Qi

doi:10.1007/s10707-020-00405-y

Cited by 6 publications

(7 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The top‐K trajectory spatiotemporal similarity query searches for the K trajectories from a trajectory dataset D that are most similar to the target trajectory T in a spatiotemporal dimension. An efficient spatiotemporal index structure is usually needed to speed up queries of similar trajectories and to avoid calculating similarity with every trajectory in the dataset (Chen et al., 2005; Dan et al., 2019; Gong et al., 2020; Ma et al., 2012; Mustafa et al., 2018; Sousa et al., 2020; Zhao et al., 2020). The index structure for vector trajectory data has always been a research difficulty because of its three‐dimensional characteristics.…”

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…in the calculation of similarity (Su et al, 2020;Zhao et al, 2020). For example, as shown in Figure 1a, it is difficult to find the similarity difference between the target trajectory T O and other trajectories T A , T B , T C using the linearly weighted similarity measures.…”

Section: Introductionmentioning

confidence: 99%

“…Second, it uses one of the common spatial similarity measures (e.g., Discrete Fréchet, LCSS) to calculate the spatial similarity, and integrates the partial similarities of each window to obtain the final spatiotemporal similarity of trajectories. (2) spatiotemporally weighted trajectory similarity (Shang et al., 2018; Zhai et al., 2019; Zhao et al., 2020). The principle is to measure the temporal similarity and spatial similarity separately, and then calculate their linear weighted sum as the spatiotemporal similarity of trajectories.…”

Section: Introductionmentioning

confidence: 99%

“…The principle is to measure the temporal similarity and spatial similarity separately, and then calculate their linear weighted sum as the spatiotemporal similarity of trajectories. However, because the trajectory data has high‐dimensional, large‐scale and dynamic update characteristics, existing research still has the following challenges: (1) Existing spatiotemporal similarity measures are not spatiotemporally synchronized, and this causes inaccuracy in the calculation of similarity (Su et al., 2020; Zhao et al., 2020). For example, as shown in Figure 1a, it is difficult to find the similarity difference between the target trajectory T O and other trajectories T A , T B , T C using the linearly weighted similarity measures.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

gsstSIM: A high‐performance and synchronized similarity analysis method of spatiotemporal trajectory based on grid model representation

Liu

Qiao

et al. 2022

Transactions in GIS

View full text Add to dashboard Cite

Existing spatiotemporal similarity analysis methods for trajectories have the problems of spatiotemporal unsynchronization and low efficiency in processing large‐scale datasets, which cannot satisfy the increasingly urgent requirements of real‐time or quasi‐real‐time applications. To address these problems, this article proposes a grid‐based and synchronized spatiotemporal similarity analysis method based on the spatiotemporal grid model called gsstSIM. First, a low‐dimensional and multi‐scale trajectory coding representation is implemented based on the spatiotemporal grid model. Second, a synchronized spatiotemporal similarity measure is proposed based on trajectory codes. It transforms the similarity analysis from complex geometric calculations to simple algebraic operations of code sets, which reduces the computational complexity. In addition, the trajectory encoding representation with space‐time collinearity enables gsstSIM to measure the synchronized spatiotemporal similarity. Third, the efficient Multi‐scale grid index, called MSGrid, is established to realize fast query of top‐K similar trajectories for large‐scale datasets. Experimental results demonstrate that gsstSIM is more robust to noise positioning points and various sampling rates than the state‐of‐the‐art algorithms STLCSS, TWS and SWS. It can achieve a second‐level response of spatiotemporal similarity query in processing large‐scale datasets, which is much faster than existing algorithms. The proposed method has promising to support the applications with high time‐efficiency requirements such as epidemic tracking and traffic condition calculation.

show abstract