频域约束的红-黑小波变换图像复原算法

Frequency Restricted Red-Black Wavelet Transform for Image Restoration

提出一种频域约束的红-黑小波变换图像复原新算法.该算法引入频域约束因子,对不同频率的部分系数给予不同程度的约束,保证了对有色噪声的压制效果,保留了图像细节信息.利用频域约束的红-黑小波变换提取图像信息,消除了有色噪声和振铃效应,实现了降质图像复原.运用含噪退化的图像进行了实验.结果表明,该方法与维纳滤波、等功率谱方法相比,信噪比RSN及信噪比提高RS′N两个参数均提高了3倍.

A novel algorithm of frequency restricted red-black wavelet transform for image restoration is presented. Introducing the frequency restricted factor, the algorithm restricts each and every frequency coefficient to retain image information while compressing the noise. The algorithm makes use of the red-black wavelet transform in extracting image from the result, compressing its color noise and ringing effect. Experimental results showed that the algorithm is three times as effective than Wiener filtering, and power spectrum equalization filtering on the R SN and R SN parameters.