关键词: 图像分割, 雁群, 粒子群优化, 二维最大熵, 阈值

Abstract: According to large amount of calculation，Particle Swarm Optimization（PSO） might easily fall into local optimal values and long runtime in two dimension entropy image segmentation，this paper proposes a two dimension entropy segmentation of improved PSO.This method is based on PSO enlightened by geese flock.Using the ideas in literature，the velocity formula is improved，and the random interfere strategy is introduced to accelerate the converge speed and overcome the part optimal values problem from two aspects.The simulations show that the searching time is shorten，the convergence speed is improved and the real-time effect of image segmentation in threshold optimization is improved too.

Key words: image segmentation, goose flock, Particle Swarm Optimization（PSO）, two dimension maximum entropy, threshold value