Computer Engineering and Applications ›› 2017, Vol. 53 ›› Issue (22): 40-44.DOI: 10.3778/j.issn.1002-8331.1607-0015

Previous Articles     Next Articles

Multi-objective evolutionary fuzzy clustering algorithm based on semi-supervision

WANG Jun, ZHAO Feng   

  1. School of Telecommunications and Information Engineering, Xi’an University of Posts and Telecommunications, Xi’an 710061, China
  • Online:2017-11-15 Published:2017-11-29


王  俊,赵  凤   

  1. 西安邮电大学 通信与信息工程学院,西安 710061

Abstract: In order to solve the traditional clustering image segmentation results not well because of the lack of effective guidance, it introduces a semi-supervised approach as a multi-objective evolutionary fuzzy clustering algorithm, and proposes a multi-objective  evolutionary fuzzy clustering algorithm for image segmentation based on semi-supervision. The proposed technique simultaneously optimizes the semi-supervised fuzzy compactness and fuzzy separation among the clusters and makes use of monitoring information to guide the clustering process. In the final generation, it produces a set of non-dominated solutions, from which the best solution in terms of a proposed validity index BI based on similarity measure is chosen to be the best clustering solution. Experimental results show that compared with other unsupervised fuzzy algorithms, the proposed clustering technique can effectively improve the clustering accuracy and the segmentation result in vision.

Key words: multi-objective evolutionary algorithm, image segmentation, semi-supervision, fuzzy clustering, similarity measure

摘要: 为了解决传统聚类由于缺少有效指导而导致图像分割结果不理想的问题,将半监督方法引入到多目标进化模糊聚类算法中,提出了一种基于半监督的多目标进化模糊聚类。图像分割算法通过构造基于半监督的类内紧致性函数和类间分离度函数,利用监督信息指导聚类过程获得非支配解集。为了从非支配解集中选择一个最优解,利用监督信息构造了基于相似性度量的有效性指标。实验结果表明,提出的方法在分割准确率和视觉效果上明显优于无监督的聚类方法。

关键词: 多目标进化算法, 图像分割, 半监督, 模糊聚类, 相似性度量