Research on Objective Tracking of Mean Shift Algorithm Based on Particle Swarm Optimization

In light of Mean Shift's inability to update model during objective tracking process, an updating solution for models of Means Shift algorithm is proposed by utilization of Particle Swarm Optimization. This solution improves each Eigen value probability, as a single particle, in model image characteristic space by using Particle Swarm Optimization algorithm, time variations according to probability can be calculated to acquire variation of all Eigen value in models, which in turn, results in updating of models. In the solution, the combinational advantage of Particle Swarm's global and regional search is fully utilized to acquire self-adaptable and optimal models. Experiment results indicate the solution can effectively solve models' un-matching problems resulted from spinning and masking of moving objective so as to realize accurate and fast objective tracking and improve self-adapting ability of tracking algorithm.