基于相邻小波系数的图像去噪算法  

Neighboring Wavelet Coefficients Based Image Denoising Algorithm

作　　者：王蕊[1] 李钊[2] 李俊山[1] 罗蓉[1] 

机构地区：[1]第二炮兵工程学院403室,陕西西安710025 [2]二炮驻石家庄地区军事代表室,河北石家庄050081

出　　处：《无线电通信技术》2007年第6期40-41,45,共3页Radio Communications Technology

摘　　要：针对传统的小波去噪方法在滤除噪声上的明显不足,提出了一种基于相邻小波系数的图像去噪算法。充分考虑了滤噪系数与相邻系数间的相关性,首先采用3×3像素的系数滑动窗口对噪声系数进行均值采样,然后结合双阈值法对小波系数进行缩减。实验结果表明本算法明显优于传统的滤波算法,具有较高的信噪比,是一种有效、可行的图像去噪方法。Although wavelet shrinkage shows its excellent performance in operation, it kills too many useful image information. An image denoising algorithm is proposed based on soft thresh holding using neighbor wavelet coefficients. In this paper, the relativities of filter coefficients and neighbor wavelet coefficients are considered adequately. Firstly, a coefficient sliding window of 3 ~ 3 pixel is adopted to perform mean sampling of the noise coefficient, and then the dual threshhold method is used to reduce wavelet coefficients. Experimental results show that the proposed algorithm is more robust than the traditional filter algorithms.

关 键 词：魏纳滤波 小波缩减法 系数相关性 峰值信噪比 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]