计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (22): 186-188.DOI: 10.3778/j.issn.1002-8331.2008.22.055

• 图形、图像、模式识别 • 上一篇    下一篇

基于各向异性扩散方程的并行图像去噪研究

程 群,田有先   

  1. 重庆邮电大学 计算机科学与技术学院,重庆 400065
  • 收稿日期:2007-10-10 修回日期:2008-01-02 出版日期:2008-07-11 发布日期:2008-07-11
  • 通讯作者: 程 群

Parallel image denoising research based on anisotropic diffusion equation

CHENG Qun,TIAN You-xian   

  1. College of Computer Science and Technology,Chongqing University of Posts and Telecommunications,Chongqing 400065,China
  • Received:2007-10-10 Revised:2008-01-02 Online:2008-07-11 Published:2008-07-11
  • Contact: CHENG Qun

摘要: 各向异性扩散方程是一种非线性PDE模型,在图像去噪中,通过非线性扩散因子来滤除噪声,同时能保留原有的边缘和纹理。但是当图像很大时,求解PDE的差分运算量将很大,满足不了实时系统的要求。针对该模型,在MPI并行编程环境下,利用图像像素的独立性和PDE求解的并发性,采用并行方式对图像去噪,在保证去噪性能的同时,极大地降低计算时间。

关键词: 各向异性扩散方程, 并行计算, 图像去噪, MPI

Abstract: Anisotropic diffusion equation is a nonlinear image processing methods by using Partial Differential Equation(PDE),with the nonlinear diffusion coefficient,it can restrain noise fairly good,reserve original edge and texture,especially for image denoise.However,the computing size of the PDE is huge when the image pixel is big and the time cost of computing is important for the real time system.For the anisotropic diffusion equation in the MPI based environment,parallel computing can reduce the computing time greatly by using independence of image pixel and subsequent of PDE while keeping the good performance of image deniosing.

Key words: anisotropic diffusion equation, parallel computing, image denoise, MPI