基于非局部自适应字典的乘性噪声去除

摘要

For better protecting the important information of the image,based on the BF model and the nonlocal sparse model,a nonlocal adaptive dictionary algorithm was put forward to remove multiplicative noise.The multiplicative noise was translated into additive noise by logarithmic transformation.PCA sparse dictionary was combined with iterative shrinkage algorithm to update the sparse code,and the denoising image in a logarithmic domain was obtained using Newton iterative method.Through an exponential function and error correction,the denoising image in the real domain was obtained.Experimental results show that,the proposed algorithm can remove multiplicative noise effectively while better maintain edges,details and texture structures of the image.%为更好保护图像的重要信息,在非局部模型及BF模型的基础上,提出一种乘性噪声去除算法.利用对数变换将乘性噪声转换为加性噪声,结合PCA稀疏字典和迭代收缩算法更新稀疏编码,用牛顿迭代法得到对数域中的去噪图像,通过指数函数以及误差校正得到实数域中的去噪图像.实验结果表明,该算法能够有效去除噪声,较好保留图像的边缘、细节和纹理信息.