电子学报 ›› 2015, Vol. 43 ›› Issue (9): 1841-1849.DOI: 10.3969/j.issn.0372-2112.2015.09.024

• 学术论文 • 上一篇    下一篇

基于图像分解的稀疏正则化多区域图像分割方法

李亚峰   

  1. 宝鸡文理学院计算机学院, 陕西宝鸡 721016
  • 收稿日期:2014-01-02 修回日期:2015-01-05 出版日期:2015-09-25
    • 作者简介:
    • 李亚峰 男,1977年生于陕西省安康市,副教授,博士,研究方向为图像处理与模式识别.E-mail:liyafeng770729@126.com
    • 基金资助:
    • 国家自然科学基金 (No.61379030); 陕西省教育厅专项科研计划项目 (No.14JK1048); 陕西省自然科学基础研究计划基金 (No.2015JM6329); 宝鸡文理学院院级科研重点项目 (No.ZK15057)

A Sparsity Regularized Multiregion Image Segmentation Method Based on Image Decomposition

LI Ya-feng   

  1. Department of Computer Science, Baoji University of Arts and Science.Baoji, Shaanxi 721016, China
  • Received:2014-01-02 Revised:2015-01-05 Online:2015-09-25 Published:2015-09-25

摘要:

针对图像具有不同特征的成分,提出一种基于图像分解的多区域图像分割模型和算法.首先将图像分解项引入到图像分割模型中,递减了纹理和噪声对分割的影响;其次使用稀疏正则化方法保持分割区域的边缘几何结构;最后基于增广Lagrange乘子法,给出一种由扩散流引导的小波迭代阈值图像分割算法.一系列实验结果表明,提出的方法抗干扰能力强,对噪声具有更好的鲁棒性.提出的方法不仅能够分割结构图像,并且能够分割较复杂的纹理图像.

关键词: 图像分割, 图像分解, 稀疏表示, 小波, 变分模型

Abstract:

Taking into account different feature components of images this paper presents a multiregion image segmentation model and algorithm based on image decomposition.Firstly,we introduce image decomposition term into the proposed image segmentation model.Image decomposition term can reduce the influence of texture and noise on our segmentation tasks.Secondly,we use sparsity regularization method to protect the edges and shape of the segmented regions.Finally,based on the augmented Lagrange multiplier method,we present an iterative wavelet shrinkage image segmentation algorithm which is guided by a diffusion flow.A series of experimental results show that the proposed method has strong anti-interference ability and it is more robust to noise.The proposed method can segment not only images with simple construction but also complex texture images.

Key words: image segmentation, image decomposition, sparse representation, wavelet, variational model

中图分类号: