A Sequential Estimation Algorithm of Particle Filters by Combination of Multiple Independent Features in Evidence

We investigate a robust sequential estimation algorithm of particle filters, which combine multiple features of visual objects, in order to obtain reliable evidential information from independent sources of sensor data. Most of particle filter algorithms are based on conditional density propagation in Bayesian inference rules. In this paper, it is modified by the conjunctive rule of independent features. Therefore, the proposed algorithm is more reliable since it demonstrates the solution to both efficiency depletion and over-sampling in particle filters.