A study of extended PSO algorithm for video image processing and its application to a mobile robot

In this paper, we propose an object tracking method on video sequence using an extension of Particle Swarm Optimization (PSO) algorithm, which is inspired by the behavior of bird flocking and fish schooling. PSO is a useful algorithm because of its easy-to-understand algorithm, rapid exploration and easy implementation. In general, PSO is applied to optimization of static problem, and thus it is difficult to use PSO for the object tracking problems where the target moves in real time. To overcome this difficulty, we introduce two extensions. To preserve the feature of PSO, it is important to extend algorithm with the simple way. The first extension is to introduce parameters so that the individuals do not converge completely. The second one is a re-initialization technique to track object between each frame-image smoothly. With these extensions, we achieve a high efficiency real time object tracking algorithm on video sequence. We demonstrate the effectiveness and practical utility of our method through some simulations and experiments.