北京航空航天大学学报 ›› 2004, Vol. 30 ›› Issue (08): 718-722.

• 论文 • 上一篇    下一篇

基于自相关函数的自然纹理图像分形维数的估计

王华, 王祁, 孙金玮   

  1. 哈尔滨工业大学 自动化测试与控制系, 哈尔滨 150001
  • 收稿日期:2003-05-26 出版日期:2004-08-31 发布日期:2010-09-21
  • 作者简介:王 华(1975-),男,黑龙江讷河人,讲师, hitwh@hit.edu.cn.

Auto-correlation function based estimation of the fractal dimension of natural texture images

Wang Hua, Wang Qi, Sun Jinwei   

  1. Dept. of Automatic Measurement and Control, Harbin Institute of Technology, Harbin 150001, China
  • Received:2003-05-26 Online:2004-08-31 Published:2010-09-21

摘要: 提出了一种估计分形维数的新方法,并利用该方法估计自然纹理图像的分形维数.分形维数是描述图像粗糙程度的主要参数.将分数布朗运动推广到离散情况,研究离散高斯噪声的自相关函数的性质.根据自相关函数的性质,获得估计分形维数的自相关函数方法.合成分形图像和自然纹理图像被用来检验该方法的准确性,并与计盒方法进行了比较,结果显示自相关法是准确的和有效的.

Abstract: A novel method for estimating the fractal dimension has been proposed and the method has been applied to estimate the fractal dimension in natural texture images. Fractal dimension is an important parameter to characterize roughness in an image. Extending the basic theory of fractional Brownian motion to the discrete case, the characteristics of auto-correlation function of fractional Gauss noise has been studied. Based on the characteristics, an auto-correlation function method is obtained to estimate fractal dimension. Both synthesis images of fractal and natural texture images are used to test the proposed method. The method is compared with the box-counting method and the results show that the auto-correlation function method is accurate and efficient.

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发