• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

计算机工程与科学

• 图形与图像 • 上一篇    下一篇

多目标粒子群和人工蜂群混合优化的阈值图像分割算法

赵凤1,2,孔令润1,2,马改妮1,2   

  1. (1.西安邮电大学通信与信息工程学院,陕西 西安 710121;
    2.西安邮电大学电子信息现场勘验应用技术公安部重点实验室,陕西 西安 710121)
     
  • 收稿日期:2019-06-17 修回日期:2019-09-24 出版日期:2020-02-25 发布日期:2020-02-25
  • 基金资助:

    国家自然科学基金(61571361,61102095,61671377);西安邮电大学西邮新星团队资助(xyt2016-01)

A thresholding image segmentation algorithm
based on multi-objective particle swarm and
artificial bee colony hybrid optimization

ZHAO Feng1,2,KONG Ling-run1,2,MA Gai-ni1,2   

  1. (1.School of Telecommunications and Information Engineering,Xi’an University of Posts & Telecommunications,Xi’an 710121;
    2.Ministry of Public Security,Key Laboratory of Electronic Information Application Technology for Scene Investigation,
    Xi’an University of Posts & Telecommunications,Xi’an 710121,China)
  • Received:2019-06-17 Revised:2019-09-24 Online:2020-02-25 Published:2020-02-25

摘要:

在图像分割中,为了准确地把目标和背景分离出来,提出了一种基于多目标粒子群和人工蜂群混合优化的阈值图像分割算法。在多目标优化的框架下,将改进的类间方差准则和最大熵准则作为适应度函数,通过粒子群和蜂群混合优化这2个适应度函数来获得1组非支配解。同时,为了提高全局和局部搜索能力,在蜂群进化时,将粒子群的全局最优解引入到人工蜂群算法的雇佣蜂阶段蜜源的更新中,并对搜索方程进行改进。最后通过类间差异和改进的类内差异的加权比值,从一组非支配解中选取最优阈值。实验结果表明,该算法能够取得理想的分割结果。
 

关键词: 阈值分割, 粒子群优化算法, 人工蜂群算法, 混合优化, 多目标优化

Abstract:

 In order to accurately separate the objects from the background in images, a thresholding image segmentation algorithm based on multi-objective particle swarm and artificial bee colony hybrid optimization is proposed. Under the framework of multi-objective optimization, the improved inter-class variance criterion and maximum entropy criterion are used as the fitness functions, and then the two fitness functions are optimized by particle swarm and bee colony hybrid optimization to obtain a set of non-dominated solutions. At the same time, the global optimal solution of particle swarm is introduced into the employed bee phase to update the honey source and the search equation is modified, so as to improve the global and local search abilities in the evolution of the bee colony. Finally, the weighted ratio of between-cluster variation and modified intra-cluster variation is adopted to select an optimal solution from a set of non-dominated solutions. Experimental results show that this algorithm can obtains ideal thresholding segmentation results.
 

Key words: thresholding segmentation, particle swarm optimization algorithm, artificial bee colony algorithm, hybrid optimization, multi-objective optimization