Computer Engineering and Applications ›› 2010, Vol. 46 ›› Issue (15): 150-152.DOI: 10.3778/j.issn.1002-8331.2010.15.044

• 图形、图像、模式识别 • Previous Articles     Next Articles

Threshold selection method of image segmentation based on improved partical swarm optimization algorithm

LIU Ding-feng,TONG Xiao-nian   

  1. South-Central University for Nationalities,Wuhan 430074
  • Received:2009-01-04 Revised:2009-03-10 Online:2010-05-21 Published:2010-05-21
  • Contact: LIU Ding-feng

一种改进微粒群算法的图像阈值分割方法

刘丁峰,童小念   

  1. 中南民族大学 计算机科学学院 武汉 430074
  • 通讯作者: 刘丁峰

Abstract: Using Kittler and Illingworth criteration function as evaluation function,an image segmentation algorithm based on improved partical swarm optimization(IPSO) and minimum error method is provided in this paper.To overcome some limitations of PSO,especially the convergence rate,IPSO amends the selection strategy from genetic algorithms(GA),after that,the IPSO oriented to the high-speed overall search capability.The experiment result shows that the IPSO algorithm can not only obviously save the running time,but also improve the quality of image segmentation.

Key words: minimum error method, image segmentation, partical swarm optimization, genetic algorithm

摘要: 以Kittler和Illingworth准则函数作为评价函数,提出了一种利用最小误差法和改进微粒群算法对图像进行阈值分割的方法IPSO(Improved Partical Swarm Optimization)。为了改善PSO算法,特别是在收敛速度方面的局限,IPSO算法引入遗传算法的择优思想对基本微粒群算法进行改进,使得改进后的IPSO算法具有快速的全局搜索能力。实验结果表明,对于灰度分布较复杂的图像,改进的IPSO算法不仅降低了运算开销,而且获得了满意的图像分割效果。

关键词: 最小误差法, 图像分割, 微粒群算法, 遗传算法

CLC Number: