欢迎访问中国科学院大学学报,今天是

中国科学院大学学报 ›› 2005, Vol. 22 ›› Issue (1): 24-29.DOI: 10.7523/j.issn.2095-6134.2005.1.004

• 论文 • 上一篇    下一篇

一种规整化的各向异性扩散相干斑抑制算法

乔明, 王新楼, 邹谋炎   

  1. 中国科学院电子学研究所 北京10008
  • 收稿日期:2004-04-28 修回日期:2004-05-25 发布日期:2005-01-15

A Regularized Anisotropic Diffusion for Speckle Reducing

Qiao Ming, Wang Xinlou, Zou Mouyan   

  1. Institute of Electronics, Chinese Academy of Sciences, Beijing 100080, China
  • Received:2004-04-28 Revised:2004-05-25 Published:2005-01-15

摘要:

针对基于各向异性扩散的相干斑抑制算法 (SRAD)中存在的不合理性,即算法在均匀区域内降噪特性一般,并且扩散系数的非凸可能导致解不收敛到正确的图像,从理论上进行了系统的分析和论证,并依据保持图像细节的规整化准则,构造出新的扩散控制系数。在此基础上,进一步提出了一种更合理的规整化各向异性扩散模型 (RSRAD),该模型可以有效地克服SRAD中存在的不合理性。实验结果表明,RSRAD算法优于SRAD。

关键词: 各向异性扩散, 相干斑, 偏微分方程, 降噪

Abstract:

Speckle reducing anisotropic diffusion (SRAD) algorithm still has disadvantages. Firstly , the diffusion coefficient is not a convex function , which leads to an unstable solution. At the second point , SRAD is anisotropy even in the homogeneous areas , this will depress the ability of reducing speckle. In this paper we systematically an- alyze the problem mentioned above and propose a new diffusion coefficient based on the edge-preserving regulariza- tion method. We also construct a new anisotropic diffusion module , R-SRAD , which is more reasonable. The ex- perimental results show that our R-SRAD is more effective in comparison with the SRAD.

Key words: anisotropic diffusion, speckle, PDE , denoising

中图分类号: