计算机工程与应用 ›› 2018, Vol. 54 ›› Issue (9): 183-188.DOI: 10.3778/j.issn.1002-8331.1612-0253

• 图形图像处理 • 上一篇    下一篇

广义模糊熵图像阈值分割参数选取的ADE方法

姜圣涛,穆学文   

  1. 西安电子科技大学 数学与统计学院,西安 710126
  • 出版日期:2018-05-01 发布日期:2018-05-15

ADE method of parameter selection in image thresholding based on generalized fuzzy entropy

JIANG Shengtao, MU Xuewen   

  1. School of Mathematics and Statistics, Xidian University, Xi’an 710126, China
  • Online:2018-05-01 Published:2018-05-15

摘要: 针对广义模糊熵图像阈值分割参数不能自动选取,提出自适应差分进化(Adaptive Differential Evolution,ADE)的广义模糊熵图像阈值分割方法。利用自适应差分进化算法作为优化工具来选取广义模糊熵阈值分割所需要的最佳参数,引入自适应变异算子和提出交叉概率自适应函数对优化过程进行控制,通过把参数带入广义模糊熵的补函数得到图像的阈值,进而得到图像最优分割。为验证其有效性与可行性,分别同基本图像质量评价准则的模糊熵图像阈值分割算法和粒子群优化广义模糊熵图像阈值分割算法相比较,实验表明,针对不同细节的图片,该算法所得分割结果多数情况下背景信息更少,目标信息更清晰,用时更短,分割更稳定且效果良好。

关键词: 广义模糊熵, 自适应差分进化, 阈值分割, 模糊集, 补函数

Abstract: Aiming at the problem that the parameters of image threshold segmentation for generalized fuzzy entropy can not be selected automatically, the image threshold segmentation method for generalized fuzzy entropy based on adaptive differential evolution is proposed. The optimal parameters for threshold segmentation of generalized fuzzy entropy are selected by using the adaptive differential evolution algorithm as the optimization tool, and the adaptive mutation operator and the crossover probability adaptive function are introduced to control the optimization process. The threshold of the image is obtained by adding the parameters to the complement function of the generalized fuzzy entropy, and then the optimal segmentation is obtained. In order to verify the effectiveness and feasibility of the proposed algorithm, this paper compares it with the image threshold segmentation algorithm for fuzzy entropy based on the basic image quality evaluation criteria and the image threshold segmentation algorithm for generalized fuzzy entropy based on the particle swarm optimization. Experiments show that, for different details of the picture, the segmentation results of the algorithm have less background information in most cases, its target information is more clearly, and it takes less time. In addition, its segmentation is more stable and effective.

Key words: generalized fuzzy entropy, Adaptive Differential Evolution(ADE), threshold segmentation, fuzzy set, complement function