CONDENSATION - conditional density propagation for visual tracking-英文文献.pdf

Condensation conditional density propagation for visual tracking Michael Isard and Andrew Blake Department of Engineering Science University of Oxford Oxford OX PJ UK Int J Computer Vision in press ABSTRACT The problem of tracking curves in dense visual clutter is challenging Kalman ltering is in adequate b ecause it is based on Gaussian densities which b eing unimo dal cannot represent simultaneous alternative hyp otheses The Condensation algorithm uses factored sampling previously applied to the interpretation of static images in which the probability distribution of p ossible interpretations is represented by a randomly generated set Condensation uses learned dynamical mo dels together with visual observations to propagate the random set over time The result is highly robust tracking of agile motion Notwithstanding the use of sto chastic metho ds the algorithm runs in near realtime Contents Tracking curves in clutter Discretetime propagation of state density Factored sampling The Condensation algorithm Sto chastic dynamical mo dels for curve motion Observation mo del Applying the Condensation algorithm to videostreams Conclusions A Nonlinear ltering B Derivation of the sampling rule