Computer Engineering and Applications ›› 2019, Vol. 55 ›› Issue (11): 179-186.DOI: 10.3778/j.issn.1002-8331.1803-0110

Previous Articles     Next Articles

Regularized Picture Fuzzy Clustering and Its Robust Segmentation Algorithm

SUN Jiamei, WU Chengmao   

  1. School of Electronic Engineering, Xi’an University of Posts and Telecommunications, Xi’an 710121, China
  • Online:2019-06-01 Published:2019-05-30



  1. 西安邮电大学 电子工程学院,西安 710121

Abstract: Considering the existing picture fuzzy clustering algorithm with poor rationality and weak ability of suppressing noise, a picture fuzzy clustering robust algorithm embedded with symmetric regularization term is proposed. Firstly, a symmetric regularized term is constructed by combining the neutral degree with the refusal degree of picture fuzzy clustering and let into the objective function of existing picture fuzzy clustering. Secondly, the spatial information constraint regulari-zation item is constructed by using the mean information corresponding to the pixel neighborhood. Finally, the regularized picture fuzzy clustering robust segmentation algorithm is obtained by means of Lagrange multiplier method. The segmentation results of different images interfered with noise show that the proposed segmentation algorithm is effective, and has stronger ability of suppressing noise than existing robust fuzzy clustering segmentation algorithms.

Key words: picture fuzzy clustering, regularization, robust segmentation, spatial neighbor information

摘要: 针对现有图形模糊聚类算法合理性差和抗噪能力弱的问题,提出嵌入对称正则项的图形模糊聚类鲁棒算法。将样本聚类所对应的中立度与拒分度相结合构造对称正则项,嵌入现有图形模糊聚类所对应的目标函数;同时,利用像素邻域所对应的均值信息辅助当前像素聚类并构造了空间信息约束正则项,采用拉格朗日乘子法获得正则化图形模糊聚类鲁棒分割算法。不同噪声干扰图像分割结果表明,所建议的分割算法是有效的,相比现有的鲁棒模糊聚类分割算法具有更强的抑制噪声能力。

关键词: 图形模糊聚类, 正则化, 鲁棒分割, 空间邻域信息