图像平滑作为去除图像中含有噪声的图像增强处理技术，是各种与图像有关软件中必不可少的功能模块。优秀的平滑算法应该在有效处理含有多种噪声的受污图像时，仍具备良好的图像细节保持能力。为了能够在有效地处理含有多种噪声的受污图像的同时保持图像细节，提出了一种结合非线性和线性滤波器的稳健平滑算法，该算法是在分析几种细节保持能力良好的平滑算法(梯度加权平滑、自适应中值滤波、稳健平滑、边缘保持滤波器)的基础上，通过吸收上述几种平滑算法细节保持效果较好的优点，结合它们各自对不同噪声的适应性而得到的。针对含有多种噪声的受污图像，用该算法进行了去噪试验，多次试验结果表明，该算法在稳健地处理这些受污图像的同时，仍可以保持良好的图像细节，同时该算法还可以稳健地处理复杂的含噪图像，且具备良好自适应性，可以为涉及图像处理的软件架构，提供相应的算法支持。

As a key image enhancement technology to remove the noise in images, image smoothing is a necessary function in various image processing software. Excellent smoothing algorithm should be able to deal with images corrupted by various noises and has capability of preserving detailed information. Based on analyzing a few of keeping details smoothing algorithm (gradient weighting filter, self adapted median filter, robust smoothing, edge keeping filter) and according to each adaptation to various noise, a robust smoothing method is presented combining non linear and linear filter. Tests on corrupted image of various noise show that the algorithm can deal with corrupted image robustly and keep well image detail at the same time.