中文摘要：

为了有效滤除图像高斯噪声,将数学形态学与小波域增强相结合,提出了一种高斯噪声新型滤波算法.该算法首先将噪声图像进行二维小波分解,得到低频和高频子图像;然后保留低频子图像不变,对各高频子图像根据其噪声分布特点分别设计出多角度、多结构逐级形态学滤波器进行滤波处理,并进行小波分解系数重构;最后对经过形态学滤波后的图像进行2层小波分解,通过设计出一种新型小波增强函数对不同幅值的小波系数进行不同程度的收缩处理,在此基础上进行分解系数重构.将自适应中值滤波与数学形态学滤波与本文算法进行比较,实验证明本文滤波算法其去噪效果优于前两种算法.

英文摘要：

In order to filter the Gaussian noise in digital image,combining the Mathematical morphology and Wavelet domain enhancement,a new filter algorithm is put forward.Firstly,the noise image is conducted two-dimensional wavelet decomposition,obtaining high-frequency and low-frequency sub image.Then keep the low-frequency sub image unchanged,according to the characteristics of the Gaussian noise distribution in each high-frequency sub image,the multi-angles,multi-structure mathematical morphology filters are designed to filter out the Gaussian noise,then the wavelet coefficient are reconstructed.Finally,the image after mathematical morphology filtering are conducted two layer wavelet decomposition,a new wavelet domain enhancement function is designed so as to contract the different amplitude wavelet coefficients in different degree,then the wavelet coefficient are reconstructed.The adaptive average filter and mathematical morphology and the new filter algorithm in this paper are applied to denoising the Gaussian noise in digital image respectively,the experiment results show that the new filter algorithm in this paper is better than the others.