Abstract: Aiming at the problems that the image denoising effect is poor by the traditional wavelet domain Wiener filter, and there are many pseudo Gibbs, this paper proposes a new denoising algorithm of wavelet domain Wener filter and Perona-Malik fusion. Firstly, the thermal diffusion iteration of the partial differential equation is simulated by wavelet-domain Wiener filter to remove noise, then the result obtained through the above will be done the second denoising with the Perona-Malikalgorithm. During the iterative process, the effective information of the image is preserved by the adaptivity of the noise weight coefficient η as possible as it can. The results of simulation experiment and the analysis from comparison with other algorithms indicate that this algorithm has a good ability of denoising and suppressing the pseudo Gibbs, preserves the edge details of the image and improves the PSNR (peak signal to noise ratio) of the image.

Key words: wavelet domain, domain boundary, Wiener filter, Perona-Malik, pseudo Gibbs

摘要： 针对传统的小波域维纳滤波图像降噪效果不理想，并产生伪Gibbs效应的问题，提出了一种小波维纳滤波和Perona-Malik融合去噪的新算法。该算法首先采用模拟偏微分方程的热扩散迭代，在小波域上进行维纳滤波去噪，由此得到的中间结果再通过Perona-Malik算法进行二次去噪，并在迭代过程中通过噪声权系数η的自适应性，在去噪过程中最大程度地保留图像的有效信息。仿真实验结果及与其他算法的对比分析表明，该算法具有较好的去噪和抑制伪Gibbs效应的能力，有效保存了图像的边缘细节，同时也提高了峰值信噪比（peak signal to noise ratio，PSNR）。

关键词: 小波域, 领域边界, 维纳滤波, Perona-Malik, 伪Gibbs