The Research Of Moving Object Tracking Based On Swarm Intelligence Algorithm

Tracking objects is one of the most important projects in the field of computer vision. It is applied widely in many ways, such as visual navigation, military guidance, traffic monitoring, medical diagnostic etc. The algorithm of particle filter is the core technology for tracking objects. It is formed by non-intelligent particles with swarm characteristic. In order to improve precision and robustness of tracking, making intelligent behavior into particle swarm is researched in the paper.Searching best state by swarm intelligence algorithms, including of genetic algorithms, particle swarm optimization algorithms and shuffled frog leaping algorithm, are studied in the paper. The mechanisms of collaboration, interactive and evolutionary among swarms and the dynamic tracking mechanism of particle filter are researched. These algorithms are interacted and integrated in order to make the technology of particle filter more intelligent.The tracking algorithm of shuffled frog leaping is proposed. The non-intelligent particles are given intelligent mechanism, such as grouping, selecting, exchange information, coordination and evolution. The tracking process is become a constantly correcting themselves by optimization searching way. The experiments showed that the searching ability by shuffled frog leaping algorithm was improved effectively and the tracking performance was superior to that of tracking algorithms by particle filter, genetic and particle swarm optimization.The speed relaxation iteration strategy and hybrid crossover are proposed to improve the tracking algorithm of shuffled frog leaping. The advantage of other intelligence algorithms is applied into the proposed model. When an object is accelerating and turning, the unable tracking problem can be solved by the speed relaxation iteration strategy. When an object is encountered, losing problem and degeneration phenomenon can be solved by crossover operator. The experiments showed that the diversity of individuals was gained and the tracking precision was improved by the improved model.