Tracking and shape estimation of deformable object using particle filter and adaptive vector quantizer
Takeshi Nishida, Norikazu Ikoma, Shuichi Kurogi · World Automation Congress · 2010
Recently, a rapid and robust information extraction method for tracking and shape estimation of a non-Gaussian probability density by combination of the particle filter and the competitive re-initialization learning (an adaptive vector quantization algorithm) had been proposed. Effectiveness of this method not only for robust state estimation of dynamical system but also for object shape estimation in dynamic scene had been suggested. Hence, a method for tracking and shape estimation of deformable object in dynamic scene is proposed based on this methodology. Further, effectiveness of the proposed method is shown by a numerical simulation and a real image experiment.