Advancing Rotation and Scale-Invariant Trajectory Representations for Maritime and Aviation MobilityData Analysis

Spatiotemporal data from connected vehicles, sensors, and GPS-enabled devices increasingly support applications ranging from traffic management to environmental analysis. Whereas earlier studies relied on aggregated flows or coarse-grained movement representations, high-frequency data now enable the analysis of fine-grained individual motion and the geometric properties of trajectories. In movement-constrained settings such as road traffic, geometric detail may be of limited value, but in unconstrained environments such as maritime and aviation mobility data, movement geometry can provide important cues about underlying activities. Despite this potential, geometry-based trajectory analysis remains underexplored and is often constrained by rigid decision rules that limit real-world applicability. This paper introduces RoSITa, a spatiotemporal analytics framework that leverages rotation- and scale-invariant shape signatures derived from a relative Hough transform to cluster and structure subtrajectory patterns for intuitive visual exploration. We demonstrate how RoSITa supports visual inspection to identify similar trajectory segments and show that it outperforms existing approaches on a classification benchmark when the task is driven by movement geometry. Finally, we present two real-world case studies illustrating its ability to capture 2D and 3D motion dynamics, enable interpretable human-in-the-loop analysis, and facilitate the generation of labeled data for future machine learning applications.