计算机工程与应用 ›› 2010, Vol. 46 ›› Issue (12): 146-148.DOI: 10.3778/j.issn.1002-8331.2010.12.043

• 图形、图像、模式识别 • 上一篇    下一篇

用全局稳定FCHNN在CIELab色彩空间进行图像分割

刘春波1,王会进1,罗志平2,苏锦钿2   

  1. 1.暨南大学 信息科学技术学院 计算机系,广州 510632
    2.华南理工大学 计算机科学与工程学院 软件开发环境和信息系统集成实验室,广州 510006
  • 收稿日期:2008-10-10 修回日期:2009-01-13 出版日期:2010-04-21 发布日期:2010-04-21
  • 通讯作者: 刘春波

Using globle stable FCHNN for image segmentation based on CIELab color space

LIU Chun-bo1,WANG Hui-jin1,LUO Zhi-ping2,SU Jin-tian2   

  1. 1.Department of Computer,College of Information Science and Technology,Jinan University,Guangzhou 510632,China
    2.Lab of SDII,School of Computer Science and Engineering,South China of Technology,Guangzhou 510006,China
  • Received:2008-10-10 Revised:2009-01-13 Online:2010-04-21 Published:2010-04-21
  • Contact: LIU Chun-bo

摘要:

在张星明等人所做工作的基础上,进一步从李雅普诺夫稳定性原理出发,改进性地设计了具有全局渐进稳定性和全局指数稳定性的模糊竞争Hopfield神经网络(简称GS-FCHNN),用GS-FCHNN进行彩色图像分割,从彩色图像在CIELab空间的色彩分布图获得具有明显色差的色彩值,取该值作为模糊核,从而为神经元建立基于最大隶属度原则的状态函数,实现彩色图像的模糊聚类,达到图像目标分割的目的。实验结果表明:较FCHNN算法,运算时间有明显加快,分割效果不受色彩分布的复杂度的影响,对高斯噪声的自适应能力得到进一步加强。

Abstract: According to the research results of Zhang,an advanced Globle Stable Fuzzy Competitive Hopfield Neural Network is proposed.Using the GS-FCHNN for color image sementation,take the value with obvious color difference in CIELab as fuzzy nuclear,then construct the membership function based on the principle of maximum degree of membership for all neurons,realize image segmentation based on color image fuzzy clustering.The experiment results indicate that the algorithm is better than Zhang,the effect of segmentation will not be interfered for complicated color distribution image,get more speed,better adaptive ability to Gaussian noise.

中图分类号: