计算机工程与应用 ›› 2014, Vol. 50 ›› Issue (5): 132-136.

• 图形图像处理 • 上一篇    下一篇

改进的Contourlet变换的图像去噪算法

唐  飞1,杨恢先1,曾友伟1,李  利2,谭正华2   

  1. 1.湘潭大学 材料与光电物理学院,湖南 湘潭 411105
    2.湘潭大学 信息工程学院,湖南 湘潭 411105
  • 出版日期:2014-03-01 发布日期:2015-05-12

Image denosing algorithm based on improved Contourlet transform

TANG Fei1, YANG Huixian1, ZENG Youwei1, LI li2, TAN Zhenghua2   

  1. 1.College of Material and Photoelectronic Physics, Xiangtan University, Xiangtan, Hunan 411105, China
    2.College of Information Engineering, Xiangtan University, Xiangtan, Hunan 411105, China
  • Online:2014-03-01 Published:2015-05-12

摘要: 为了更有效地去除图像中的噪声,提出一种基于Contourlet变换的新阈值与改进阈值函数的图像去噪方法。新阈值考虑了图像经Contourlet变换后在低尺度子带和高尺度子带的去噪特性,较好地克服通用阈值的不足;改进阈值函数对阈值处理的Contourlet变换的高频细节系数分为上频、中频和下频,并进行相应的处理,较好地弥补各阈值函数的缺陷。实验结果表明,与通用阈值相比,新阈值能更有效地去除图像中的噪声,获得更高的峰值信噪比(PSNR),结合改进阈值函数后,图像的边缘细节得到更好的保护,视觉效果更好。

关键词: 图像去噪, ontourlet变换, 新阈值, 改进阈值函数, 低尺度子带, 高尺度子带, 高频细节系数

Abstract: In order to remove the image noise more effectively, a new threshold and an improved threshold function based on Contourlet transform are proposed. The image denosing characteristics of low scale subband and high scale subband are considered, the new threshold overcomes the shortcomings of general threshold better; high frequency detail coefficients which are used to threshold processing are devided into top frequency, middle frequency and bottom frequency after Contourlet transform, and are processed accordingly, the improved threshold function makes up for deficiencies of each threshold functions better. Experiment on image denoising shows that compared with general threshold, the new threshold can remove the image noise more effectively and achieve higher Peak Signal-to-Nosie Ratio(PSNR). After combining the improved threshold function, the image details can get better protection and the visual quality can be better.

Key words: image denoising, Contourlet transform, new threshold;improved threshold function, low scale subband, high scale subband, high frequency detail coefficients