一种小波域改进非局部均值滤波算法

提出了一种噪声图像高效滤波算法。该算法对经典非局部均值滤波算法从边缘保持效果和计算复杂度两个方面加以改进。提出一种基于图像结构相似度(SSIM)相似性检测算子,并将其与传统的高斯加权欧氏距离进行加权融合,从而实现对经典非局部均值滤波的改进,可实现对图像边缘和平坦区域滤波的有效兼顾。将其引入到小波变换域,对于高频子图像,首先采用Canny算子实现自适应边缘检测,获得边缘和非边缘图像,采用改进非局部均值滤波和经典非局部均值滤波分别加以处理,然后实现图像的融合;最后实现小波系数重构。通过对实物图像和标准测试图像的仿真实验结果表明,该滤波算法的去噪效果较优,能基本实现对高强度随机噪声情形下的图像复原,从而印证了该滤波思路的可行性。

A improved non local means algorithms based on wavelet transform domain

A new efficient noise image algorithm is proposed.The edge effect and computational complexity of classical non local means filtering algorithm are improved.A new image similarity detection operator based on SSIM is put forward,and it is integrated with classical gaussian weighted Euclidean distance,so as to improve the classical non local mean filtering algorithm,to balance the edge filtering and flat area filtering.The improved non local mean filtering algorithm is introduced to the wavelet transform domain,firstly,the adaptive Canny operator is used to detect the edge of high dimension wavelet coefficient,so the edge image and the non edge image are obtained.The improved non local means algorithm and classical non local mean algorithm are used to deal with them respectively.Then,the operation of image fusion is conducted.Finally,the wavelet transform coefficients are reconstructed.The experimental results proved that ,the performance of the algorithm in this paper is superior than the others.