Computer Engineering and Applications ›› 2009, Vol. 45 ›› Issue (14): 191-193.DOI: 10.3778/j.issn.1002-8331.2009.14.059

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

Texture image segmentation based on FCM clustering with weighted position features

JIANG Yu-zhen1,ZHU Ying-hui1,OUYANG Chun-juan2   

  1. 1.College of Mathematics and Information Technology,Hanshan Normal University,Chaozhou,Guangdong 521041,China
    2.College of Information Science and Medium,Jinggangshan University,Ji’an,Jiangxi 343009,China
  • Received:2009-01-22 Revised:2009-04-01 Online:2009-05-11 Published:2009-05-11
  • Contact: JIANG Yu-zhen

加权方位特征的纹理图像FCM聚类分割

江玉珍1,朱映辉1,欧阳春娟2   

  1. 1.韩山师范学院 数学与信息技术系,广东 潮州 521041
    2.井冈山大学 信息科学与传媒学院,江西 吉安 343009
  • 通讯作者: 江玉珍

Abstract: Aiming at the weak regularity of natural texture,a weighted FCM texture segmentation method combined with statistical analysis of images grey distribution and texture energy is proposed in this paper.In the method,three position features such as horizontal position,vertical position and radial distance are attempted to adopt.The experiments show that weighted FCM clustering algorithm is simple,effective,controllable and good segmentation result.The weighted position features of samples can effectively improve the convergent validity of region segmentation,thus improve the segmentation precision of texture especially the uneven and rough texture.

Key words: texture segmentation, position feature, Weighted Fuzzy C-Means(WFCM), texture energy

摘要: 针对自然纹理的弱规则性特点,提出一种结合图像灰度分布、纹理能量统计的加权FCM纹理分割方法,并在该方法中尝试引入水平位置、垂直位置、中心径向距离等三种方位特征参数。实验证明,该加权FCM聚类算法简单高效、可控性强,纹理分割效果明显,其中样本的三种加权方位特征参数能增强区域分割的聚敛度,有效地提高纹理尤其是非均质粗糙纹理的分割精度。

关键词: 纹理分割, 方位特征, 加权模糊C均值聚类算法, 纹理能量