计算机工程与应用 ›› 2007, Vol. 43 ›› Issue (11): 51-53.

• 学术探讨 • 上一篇    下一篇

一种快速保边的图像分割方法

钱芸 张英杰   

  1. 西安交通大学
  • 收稿日期:2006-05-15 修回日期:1900-01-01 出版日期:2007-04-11 发布日期:2007-04-11
  • 通讯作者: 钱芸

An Image Segmentation Method for Speeding up Convergence and Edge preserving

Yun Qian   

  • Received:2006-05-15 Revised:1900-01-01 Online:2007-04-11 Published:2007-04-11
  • Contact: Yun Qian

摘要: 本文在分段光滑Mumford-Shah(MS)模型的基础上提出一种快速的图像分割方法。该方法在分割过程中利用分段常数MS模型产生一种驱动力,在每个时间间隔加入这种驱动力来推动水平集函数的进化,提高曲线演化速度;另外,通过在MS模型中引入一个非线性函数Ψ(s)代替u+和u-实现保边。最后通过实验验证,本方法不仅克服了原始方法的一些缺点,而且提高了收敛速度。

关键词: 图像分割, Mumford-Shah模型, 曲线演化, 水平集

Abstract: An improved image segmentation method based on piecewise-smooth Mumford-Shah(MS) model is proposed in this paper. An extra force is produced using the piecewise-constant MS model during the segmenting. In each time step, the additional force is added to reinitialize the level set function, advance the speed of the curve evolution. Otherwise, a nonlinear function Ψ(s) is introduced to instead of u+ and u-for edge preserving in MS model. Finally, it’s proved that this new method not only overcomes some special drawbacks of the original method, but also advances the convergence speed.

Key words: image segmentation, Mumford-Shah model, curve evolution, level set