Low Complexity Spatial Similarity Measure of GPS Trajectories

We attack the problem of trajectory similarity by approximating the trajectories using a geographical grid based on the MGRS 2D coordinate system. We propose a spatial similarity measure which is computationally feasible for big data collections. The proposed measure is based on cell matching with a similarity metric drawn from Jaccard index. We equip the proposed method with interpolation and dilation to overcome the problems missing data and different sampling frequencies when comparing two trajectories. The proposed measure is implemented online in the framework of Mopsia.