J4 ›› 2012, Vol. 39 ›› Issue (2): 87-94+212.doi: 10.3969/j.issn.1001-2400.2012.02.015

• 研究论文 • 上一篇    下一篇



  1. (空军工程大学 工程学院,陕西 西安  710038)
  • 收稿日期:2011-01-13 出版日期:2012-04-20 发布日期:2012-05-21
  • 通讯作者: 汪云飞
  • 作者简介:汪云飞(1985-),男,空军工程大学博士研究生,E-mail: fxjuan_2008@yahoo.com.cn.
  • 基金资助:


Improved isoperimetric algorithm for image segmentation

WANG Yunfei;BI Duyan;HUANG Fei   

  1. (School of Eng., Air Force Eng. Univ., Xi'an  710038, China)
  • Received:2011-01-13 Online:2012-04-20 Published:2012-05-21
  • Contact: WANG Yunfei


为了解决等周分割算法用于图像处理时抗噪性较弱,迭代效率低的不足,提出了一种新的改进思路,将图中任意结点的密度值类比为矩形的长和宽,在矩形周长一定的条件下,用矩形当前面积与最大面积的比值设计出一种新的权重函数,直接避免了归一化处理的问题; 同时充分利用次优等周率参与迭代分割,提高了算法的迭代效率.分析和实验结果表明,在保持原算法分割质量的前提下,所提算法在有效提高原算法抗噪性的同时,进一步减少了迭代运算量的10%~30%,提升了算法的性能.

关键词: 图论, 图像分割, 等周算法, 权重函数, 次优等周率


An in-depth analysis is conducted on the graph-theory-based-isoperimetric-partition-algorithm (GTBIPA), which has such drawbacks as low noise immunity and low iterative efficiency when applied to image processing, and a new approach is presented. In this new approach, the intensity of arbitrary nodes in graph are analogized as the length and width of a rectangle. On the condition of a constant perimeter of a rectangle, a new weighting function is designed by using the ratio between the present and the maximum area of the rectangle, so that the normalized problem is directly avoided; and by the participation of suboptimal isoperimetric ratio in iterative partitioning, the iteration efficiency of the algorithm is improved. Experimental results show that this new algorithm can efficiently enhance the performance of the GTBIPA by improving its noise immunity and reducing the number of iterations by 10%~30%.

Key words: graph theory, image segmentation, isoperimetric algorithm, weighting function, suboptimal isoperimetric ratio


  • TP391