中文摘要：

结合频率域滤波去噪方法，对比低通滤波图像，按照灰色关联分析方法筛选出遥感图像中可能存在的噪声像元，优化噪声窗口灰度序列，并以此建立灰色线性模型，在参考传统中值滤波思想的基础上对含噪遥感图像进行去噪处理。大量仿真试验分析表明，该方法在处理脉冲噪声方面具有良好的去噪效果，且具有保持图像信息质量的特点。此外，试验中发现，采用该方法处理含有混合噪声的遥感图像时，根据具体图像含噪情况，合理地选取阈值，在一定程度上可以提高其削弱图像中混合噪声的水平。

英文摘要：

The noise signal of image quality is caused by internal and external disturbance during receiving, transferring and decoding data of remote sensing image data.In this paper,combined with the denoising method for frequency domain and compared to the low-pass filtered image,the grey relational analysis is employed to se-lect the suspect noise pixels and optimize its spatial filtering window pixels series which is used to build a grey linear model to process the noise in the remote sensing image by reference to the traditional median filtering method.According to simulation experiment analysis,this method has not only a superior effect for removing impulse noises,but also can keep the quality of image detail information well.In addition,this combination al-gorithm can improve the level of reducing mixed noise in some extent if the threshold can be selected reasonably by considering the specific situation of noisy image.