一种基于Contourlets变换和全变差的图像去噪方法

为了在克服振荡现象的同时更多地保留图像细节部分，利用一种新的图像细节表征方法Contourlet变换，在Contourlet框架下嵌入全变差，提出了一种新的非线性无约束最优化图像去噪方法。由于Contourlet变换具有多分辨率分析和多方向的特性，使得在Contourlets域中建立模型比在小波域中建立模型更有优势。实验证明该方法在抑制噪声的同时不仅能够很好地减少阈值产生的振荡现象，同时也能很好地保持图像的边缘，是一种有效的去噪方法。

Image denoising based on Contourlet transform and total variation

In order to overcome the oscillation phenomenon and preserve more details in image denoising, a recently proposed Contourlet transform was used, which was an efficient directional multi-resolution image representation and behaved much better than the wavelet transform in capturing fine details, e.g., the edge and the texture. Unlike most conventional Total Variation (TV) minimization techniques in image processing, the TV regularity in this paper was directly imposed on the Contourlet domain. The proposed method performs efficient image denoising with fewer artifacts, and achieves a better compromise between noise removal and edge preservation.