二维不可分滤波器级联小波的图像去噪算法  

Image denoising algorithm by cascading two-dimensional nonseparable filters and wavelets

作　　者：程源源[1] 周祚峰[1,2] 水鹏朗[1] 

机构地区：[1]西安电子科技大学雷达信号处理国家重点实验室,陕西西安710071 [2]中国科学院西安光学精密机械研究所,陕西西安710119

出　　处：《系统工程与电子技术》2008年第11期2075-2078,共4页Systems Engineering and Electronics

基　　金：国家自然科学基金资助课题(60472086)

摘　　要：针对目前数字图像处理中的图像多方向信息提取问题,提出了一种利用二维不可分滤波器提取多方向信息的小波域图像去噪算法。该算法首先设计两个具有方向性的二维不可分滤波器,然后通过小波变换简单有效地提取了图像六个方向上的方向信息,最后用带有椭圆型方向窗的小波域局部阈值维纳滤波算法对含噪图像进行去噪。通过仿真实验表明,图像多方向信息的提取简单有效,且去噪效果相比于已有的基于二维可分小波的图像去噪算法有了显著的提高。To study the problem of extracting the multi-directional information of images in digital image processing, an image denoising algorithm based on extracting multi-directional information via two-dimensional （2-D） nonseparable filter in wavelet domain is proposed. Firstly, two 2-D nonseparable directional filters are designed, and then six directional features of the image are extracted simply and effectively via the wavelet transform. Finally, the wavelet-based image denoising by thresholding and Wiener filtering with elliptic directional windows is used to denoise the noisy image. The experimental results show that the proposed algorithm can simply extract multi-directional information and achieves more significant improvement than the existing image denoising algorithms using the separable 2-D wavelets.

关 键 词：图像去噪 阈值维纳滤波 方向信息 二维不可分滤波器 方向窗 

分 类 号：TN973.3[电子电信—信号与信息处理]