子波变换去噪的一种新算法

在图象平滑中，经常采用均值滤波和中值滤波。然而均值滤波对高斯白噪声比较有效而中值滤波对脉冲噪声比较有效。因而寻求对两种噪声都有效的去噪算法是我们所希望的。本文在概要介绍子波变换的基础上，首先从理论分析和实验计算两个方面详细讨论了脉冲信号和高斯白噪声的子波变换特性。然后说明了一种对脉冲噪声和高斯白噪声同时有效的新的去噪算法。最后利用改进的去噪算法对一维信号和二维信号进行了实验，并给出了数字仿真的结果。

A New Denoising Algorithm Based on Wavelet Transform

In inage smoothing, mean filters and median filters are often used. However, mean filtering is onlyeffcetive to Gaussian white noise while median filtering is only effective to impulsive noise. Therefore, new denoisingalgorithms effective to both types of noises are desired. In this paper after briefly introducing wavelet transform,the characteristics of the Waveket transform of pulse signal and white noise are first discussed in detail, from atheoretical point of view as well as an experimental one. Secondly, a new denoising algorithm effective to bothinpulsive noise and white Gaussian noise is illustrated. Finally, the modified denoising algorithm is used to processcontaminated 1D and 2D signal and results of computer simulation are given.