Motion Trajectory Clustering for Video Retrieval Using Spatiotemporal Approximation

Abstract

A new technique is proposed for clustering and similarity retrieval of video motion clips based on spatio-temporal object trajectories. The trajectories are treated as motion time series and represented either by least squares or Chebyshev polynomial approximations. Trajectory clustering is then carried out to discover patterns of similar object motion behaviour. The coefficients of the basis functions are used as input feature vectors to a Self-Organising Map which can learn similarities between object trajectories in an unsupervised manner. Encoding trajectories in this manner leads to efficiency gains over existing approaches that use point-based flow vectors to represent the whole trajectory as input vector. Experiments on two different motion datasets – vehicle tracking and pedestrian surveillance - demonstrate the effectiveness of our approach. Applications to motion data mining in video surveillance databases are envisaged.