A Dynamic Features Selection Based Algorithm for 3D Objects Motion Estimation

In this paper, an approach of kinetic parameter estimation and real-time pose tracking for 3D moving objects is investigated. The main work includes two folds: Firstly, an extended kalman filter (EKF) is designed to estimate the kinetic parameter with a hybrid eye to hand/eye in hand multi-camera vision system. Secondly, a scheme of dynamic feature selection is proposed. One of the main innovations in this paper is that the maximum inscribed circle of the feature set involved in estimation is proposed to be the criterion of feature selection. Simulation results demonstrate that the accuracy of estimation can be obviously improved by using this strategy.