Efficient Online Algorithms for the Polygonal Approximation of Trajectory Data

The paper deals with the problem of how most efficiently to represent trajectory data with a piecewise -- constant velocity model of the object movement. Firstly, we experimented with fast and efficient offline algorithms based on heuristic and optimal algorithms for the approximation of planar and spatial curves for a given constraint on the maximum of the approximation error. Secondly, we developed efficient online algorithm for the approximation of trajectories adaptive to the size of buffer. The online algorithm can be successfully used to reduce spatiotemporal data and multidimensional signals.