结合双树复小波变换的四阶偏微分影像去噪方法

针对低阶偏微分方程去噪模型通常会使图像的灰度趋于分段常量而产生阶梯效应的问题,结合小波多尺度分解在图像处理中的应用,提出一种结合双树复小波变换(DT-CWT)的四阶偏微分方程(PDE)影像去噪模型。首先采用DT-CWT对噪声影像进行多尺度分解,保留分解后的低频分量不变,其他层复高频分量采用四阶PDE去噪模型去噪,然后重构相应层的高、低频分量,得到最终去噪后影像。对不同噪声强度下的ZY-3卫星影像去噪实验的结果表明,采用本文方法去除遥感影像噪声相比现有方法得到的去噪结果信噪比平均提高了1~2 d B,提高了去噪影像的结构相似度,在有效去除影像高斯噪声的同时能够较好地保留图像细节信息。

An Image Denoising Method Combining DT-CWT and Fourth-Order PDE

It is proved that the image gray level tends to be piecewise constant in the low order partial differential equation denosing model.Considering the applications of wavelet multi-scale decomposition in image processing,a kind of remote sensing image denoising model combining dual tree complex wavelet transform and fouth order partial differential equation is put forward.Firstly,the noise images are multiscalely decomposted by DT-CWT.Then the low frequency components are reserved and the noise in high frequency components of other layer are removed by the fourth order PDE model.Finally,the high and low frequency components of the corresponding layer are reconstructed in order to get the final image.According to the denoising experiments results of the ZY-3 satellite images which have different noise intensities,it is proved that the average PSNR of results by the proposed model increases 1 ～ 2 dB and the structural similarity increases as well.The denoising model proposed can effectively preserve the image details while removing noise.