中文摘要：

针对传统维纳滤波算法在处理图像非边界区域效果不佳的问题提出了一种改进算法，进一步提高了输出图像的峰值信噪比。算法对输人的原始图像进行小波分解，然后计算出低频近似矩阵中沿待处理子块中心像素的k个方向祛除极值后模板的方差，以这些方差的倒数作为权重，对k个模板剩余像素灰度的均值进行加权来作为相应位置的输出灰度值。实验结果表明，当方向数k取值越大时，该算法能有效提高对不同比例混合噪声图像的滤波效果，输出峰值信噪比增益较现有算法达到0.25~0.4dB。同时算法能够满足实时处理的需求，可在一般电子类消费产品用于抑制图像中混合噪声。

英文摘要：

For the poor performance of the traditional wavelet domain wiener filtering algorithm on the non-boundary region,an improved new algorithm is suggested to further increase the PSNR(peak signal to noise ratio)value of output image.The algorithm uses the wavelet to decompose the input original image,and then calculates the variance of the template without extreme value from k kinds of directions of the sub block center pixel in the low frequency approximation matrix.By using the reciprocal of these variance as weight,weighted average of the remaining pixels of the template comes as the output gray value of the corresponding position.The experimental results show that this algorithm can effectively improve the filtering effect of mixed noise image in different proportion when k has a high value,and the PSNR value of the output images is increased by0.25?0?4dB than that of other similar methods.Meanwhile,the processing speed of the algorithm can meet the needs of real-time requirements,and be used to sup-press mixed noise in general electronic consumer products.