Clustering methods for 3D vision data and its application in a probabilistic estimator for tracking multiple objects

Probabilistic algorithms have been fully tested as the best solution in multiples areas, and thus in tracking tasks. Different solutions with them have been proposed for multiple objects tracking. The proposal of the authors is based on a particle filter whose robustness and adaptability is increased by the use of a clustering algorithm. Two different proposals for the segmentation process are presented in this paper, and interesting conclusions are extracted from their functional comparison. Tracking results are also presented in the paper, showing the reliability of the proposals.