一种新的小波域维纳滤波图像去噪算法

A new image denoising algorithm based on wavelet domain Wiener filtering

针对传统小波去噪和维纳滤波存在的不足,提出了一种新的小波域维纳滤波图像去噪算法。首先,对含噪图像进行小波分解,依靠对角细节子带小波系数对噪声方差进行估计;然后,引入噪声方差修正因子,并根据不同子带小波系数的统计特性,在低频子带和高频子带分别选择合适的维纳滤波模板尺寸,使维纳滤波在对小波系数进行滤波处理时具备了更加精确的噪声方差估计和更加合理的滤波模板尺寸;最后,对维纳滤波处理后的小波系数进行小波重构,获得了去噪后的图像。试验结果表明:该算法兼具小波去噪的多分辨率分析特性和维纳滤波的自适应特性,能有效提高去噪后图像峰值信噪比,去噪效果优于小波去噪和维纳滤波。

To overcome the shortcomings of traditional wavelet denoising and Wiener filtering, a new image denoising algorithm based on wavelet domain Wiener filtering is proposed. Firstly, the noisy image is decomposed by wavelet transform and the estimated noise variance is made through the wave- let coefficients of diagonal detail subband. Then the noise variance modified factor is introduced, proper template sizes of Wiener filter are selected for low frequency subbands and high frequency sub- bands concerning the statistic characteristic of wavelet coefficients; therefore, wavelet coefficients are filtered by Wiener filtering with more accurate noise variance estimation and a more reasonable filter template size. Finally the denoising image is obtained through the reconstruction of wavelet coeffi- cients filtered by Wiener filtering. Experimental results show that the new algorithm has the advanta- ges of multi-resolution analysis characteristics of wavelet denoising and adaptive characteristics of Wiener filtering and that the denoising performance in sense of peak signal-to-noise ratio （PSNR） is superior to wavelet denoising and Wiener filtering.