计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (34): 179-181.DOI: 10.3778/j.issn.1002-8331.2008.34.055

• 图形、图像、模式识别 • 上一篇    下一篇

基于Normalized Cut的图像分割改进算法

黄一岑,沈一帆   

  1. 复旦大学 计算机科学与工程系,上海 200433
  • 收稿日期:2007-12-19 修回日期:2008-02-25 出版日期:2008-12-01 发布日期:2008-12-01
  • 通讯作者: 黄一岑

Improved image segmentation algorithm based on Normalized Cut

HUANG Yi-cen,SHEN Yi-fan   

  1. Department of Computer Science and Engineering,Fudan University,Shanghai 200433,China
  • Received:2007-12-19 Revised:2008-02-25 Online:2008-12-01 Published:2008-12-01
  • Contact: HUANG Yi-cen

摘要: 用Normalized Cut(N-Cut)准则分割图片时,会出现诸如过分割或者欠分割的不理想情况。在N-Cut结果的基础上提出了一种改进算法:利用合并-分裂方法,将颜色和纹理相似的区域合并,并对某些区域做进一步N-Cut分割。实验证明,改进后的算法不仅保留了N-Cut方法的优点,而且从一定程度上解决了过分割和欠分割的缺点,能够处理一些自然界的彩色图片,得到了比较理想的分割结果。

Abstract: The over-segmentation or incomplete segmentation may happen when using Normalized Cut(N-Cut) to segment images.This paper proposes an improved algorithm based on the result of N-Cut using classic merge and split method:merge some segments with similar color and texture;split some segments further.Experimental results validate the effects on color natural images by this algorithm,which not only keep the advantages of N-Cut,but solve the problem of over-and incomplete segmentation as well.