电子学报 ›› 2017, Vol. 45 ›› Issue (11): 2728-2734.DOI: 10.3969/j.issn.0372-2112.2017.11.021

• 学术论文 • 上一篇    下一篇

一种自适应初始轮廓的水平集演化方法的研究

翁桂荣, 朱云龙, 钱森   

  1. 苏州大学机电工程学院, 江苏苏州 215021
  • 收稿日期:2016-08-16 修回日期:2016-12-27 出版日期:2017-11-25
    • 作者简介:
    • 翁桂荣,男,1963年出生,1985年毕业于南京理工大学,苏州大学机电工程学院教授.从事图像处理及生物信息处理等研究工作.E-mail:wgr@suda.edu.cn;朱云龙,男,1992年出生,苏州大学机电工程学院研究生.从事图像处理研究工作.E-mail:20154229024@stu.suda.edu.cn

Research on an Adaptive Level Set Evolution Method for Initial Contour

WENG Gui-rong, ZHU Yun-long, QIAN Sen   

  1. School of Mechanical and Electric Engineering, Soochow University, Suzhou, Jiangsu 215021, China
  • Received:2016-08-16 Revised:2016-12-27 Online:2017-11-25 Published:2017-11-25

摘要: 距离规则水平集存在对噪声、初始轮廓敏感、收敛速度慢以及容易从弱边缘处泄露等不稳定问题.结合待分割目标灰度统计信息和图像梯度信息,提出了一种自适应初始轮廓的水平集演化方法,利用图像信息构成的自带符号目标信息函数代替面积项中的边缘指示函数,解决水平集方法对初始轮廓敏感问题.另外,还设计一个自我调整的面积项系数解决水平集方法对收敛速度慢以及弱边缘处泄露问题.实验结果表明:本文方法不仅可以减少图像分割时间,提高了分割质量,同时能够解决对初始轮廓敏感问题.

关键词: 水平集, 距离规则, 主动轮廓模型, 图像分割

Abstract: The Distance Regularized Level Set Evolution (DRLSE) model is sensitive to initial contour and weak boundary images.Meanwhile,it is easy to leak from the weak edges and the curve sometimes converges slowly.To these issues,an improved level set evolution method is proposed in the paper,which combines target gray level statistical information with image gradient information.The boundary stopping function in the model of DRLSE is replaced by a target information function based on image information.And the constant coefficient associated with the weighted area was modified as an adaptive variable sign coefficient to deal with slow convergence and weak edge leakage.Experiments show that this method is free of initial contour.Besides,it reduces the time of image segmentation and improves the quality of segmentation.

Key words: level set, distance regularized, active contour model, image segmentation

中图分类号: