2009 | OriginalPaper | Chapter
Trajectory Modeling Using Mixtures of Vector Fields
Authors : Jacinto C. Nascimento, Mário A. T. Figueiredo, Jorge S. Marques
Published in: Pattern Recognition and Image Analysis
Publisher: Springer Berlin Heidelberg
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
Trajectory analysis plays a key role in human activity recognition and video surveillance. This paper proposes a new approach based on modeling trajectories using a bank of vector (velocity) fields. We assume that each trajectory is generated by one of a set of fields or by the concatenation of trajectories produced by different fields. The proposed approach constitutes a space-varying framework for trajectory modeling and is able to discriminate among different types of motion regimes. Furthermore, the vector fields can be efficiently learned from observed trajectories using an expectation-maximization algorithm. An experiment with real data illustrates the promising performance of the method.