中文摘要：

针对全变分去噪模型会模糊图像边缘和纹理部分的问题,提出一种改进的去除乘性噪声的非局部正则化模型,并利用Split-Bregman算法进行求解。对观测图像取对数变换,将乘性噪声转化为加性噪声,利用全变分思想,以非局部TV范数作为正则项,通过图像区域与区域的灰度相似性来确定权重系数,进而更好地保持图像的纹理结构;在模型中加入紧凑项来保证去噪图像的紧凑性。对模型求解并进行数值仿真实验,结果表明：改进的去除乘性噪声的非局部正则化模型能够去除图像噪声且较好地保持其纹理部分。

英文摘要：

The denoising model based on the total variation method makes edge and texture of images blur. In order to avoid these problems, a modified nonlocal regularization model that can remove multiplicative noises is proposed and solved using the Split--Bregman algorithm. In this model, a multiplicative noise is con- verted to an additive noise by adopting a logarithm transformation for images observed. What＇s more, combi- ning with the total variation method, a nonlocal TV-norm regularization term is applied. In addition, weight coefficients are determined by the gray similarity among regions of images. Finally, to assure compactness of images, the compact term is added to the model. In this study, a series of experiments are conducted on some datasets. The experimental results demonstrate that the proposed model can remove noise and also keep the texture information of images.