基于非下采样轮廓小波变换的图像去噪

Image denoising via nonsubsampled wavelet-based contourlet transform

针对采用下采样滤波器结构的轮廓波、轮廓小波在图像去噪过程中会引入伪吉布斯现象,利用小波变换(WT)和非下采样方向滤波器组(NDFB)构造了一种新的多尺度、多分辨率图像的非下采样轮廓小波变换(NWCT)。WT去除了拉普拉斯金字塔滤波器(LPF)的计算冗余,NDFB保证了该变换具有平移不变性。为了验证该变换的有效性,对其进行了图像去噪实验。实验结果表明,所提出方法能获得比WT、轮廓波变换(CT)、轮廓小波变换(WCT)更高的峰值信噪比(PSNR),并且能够很好地抑制伪吉布斯现象。

In order to restrain the pseudo-Gibbs phenomena in image denoising applications caused by the contourlet transform or the wavelet-based contourlet transform,a geometrical image transform is constructed by combining 2D wavelet transform and nonsubsampled directional filter banks.The wavelet transform removes the computational redundancy of the Laplacian Pyramid filters(LDF),while the nonsubsampled filter banks enable the proposed method to be shift-invariant.To assess the applicability of the proposed method,numerical examples of image denoising are performed.The results show that the proposed method can obtain a higher PSNR and restrain the pseudo-Gibbs phenomena compared with wavelet,contourlet and wavelet-based contourlet transforms.