摘要
本文提出了一种基于非下采样Contourlet变换与非线性各向异性扩散的方法进行含噪图像的去噪和增强。首先对含噪图像进行非下采样Contourlet分解,对每个分解层的各个子带进行非线性收缩和拉伸,以达到抑制噪声和增强图像特征的目的。然后,对去噪增强后图像的Contourlet小系数进行空间域的非线性各向异性扩散,以去除由于进行非下采样Contourlet去噪所造成的为伪Gibbs现象和side-band效应。实验结果表明,本文方法相比于无扩散的Wavelet和Contourlet方法相比,不仅对图像进行了去噪和增强,而且有效的抑制了伪Gibbs现象和side-band效应。
This paper proposes a hybrid method for image enhancement and noise reduction by using the NonSubsampled Contourlet Transform （NSCT） and Nonlinear Anisotropic diffusion. For the purpose of reducing pseudo-Gibbs and Contourlet-like artifacts, an improved gain function which integrates feature enhancement and noise reduction is introduced to nonlinearly shrink and stretch the NSCT coefficients firstly. Then, the enhanced results are further processed by the nonlinear diffusion where only the nonsignificant, i.e., thresholded, NSCT coefficients are changed by means of a diffusion process in order to reduce pseudo-Gibbs artifacts. Numerical experiments show its good performance to cancel oscillations near the discontinuities and eliminate Contourlet-like artifacts while preserving the strong edges and shapes of the features of surfaces, in comparison to existing methods.
出处
《电子设计工程》
2015年第12期126-129,共4页
Electronic Design Engineering
作者简介
贾雨（1988-），男，陕西府谷人，硕士研究生，助理工程师。研究方向：图像处理与模式识别。
