Computer Engineering and Applications ›› 2009, Vol. 45 ›› Issue (32): 174-177.DOI: 10.3778/j.issn.1002-8331.2009.32.055

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

Research of improved algorithm for multilevel thresholding image segmentation based on fuzzy maximum entropy

YANG Kai,JIANG Hua-wei   

  1. College of Information Science and Engineering,Henan University of Technology,Zhengzhou 450001,China
  • Received:2009-03-05 Revised:2009-05-07 Online:2009-11-11 Published:2009-11-11
  • Contact: YANG Kai

模糊最大熵多阈值分割的改进算法研究

杨 凯,蒋华伟   

  1. 河南工业大学 信息科学与工程学院,郑州 450001
  • 通讯作者: 杨 凯

Abstract: Based on fuzzy maximum entropy for multilevel thresholding image segmentation,a novel improved algorithm combining genetic algorithm and ICM algorithm is proposed.First of all,the fuzzy maximum entropy function is used as the adaptation of genetic algorithm.Then improvements are appropriately made on parameter,lastly step of the new algorithm is proposed.The new algorithm overcomes some weaknesses of the traditional entropy methods,reduces times of computation and enhances efficiency of computation.Experimental results demonstrate that the new algorithm has some advantages,such as speediness,validity and practicability.

Key words: image segmentation, fuzzy maximum entropy, multilevel threshold, Iterated Conditional Modes(ICM), genetic algorithm

摘要: 基于模糊最大熵原则的多阈值分割,提出了遗传算法和ICM相结合的改进算法。该方法首先确定选取模糊熵函数作为适应度函数,然后对遗传算法中的编码方式、交叉算子、变异算子等参数进行了一些适当改进,进而给出了该算法的理论推导和算法的具体实现步骤。与通常的基于模糊最大熵原理进行阈值分割方法相比较,减少了计算量并且提高了运行效率,克服了常用方法在阈值求取时的一些不足,能够快速获得稳定的阈值。对比实验得出的结果,也说明了该方法的快速性、有效性、稳定性。

关键词: 图像分割, 模糊最大熵, 多阈值, 条件迭代法, 遗传算法

CLC Number: