Abstract: As the image enhancement techniques for image denoising，the traditional image smoothing methods can improve the signal-to-noise ratio（SNR） of image，but at the meantime also blur the image.For overcoming these disadvantages，an improved adaptive Gaussian filter is introduced.The filter，which combines the properties of Gaussian filter and Gradient Inverse Weighting Filter，takes simultaneously the space distance and pixel distance into account，so as to choose the pixels and their weights for local smoothing.The filter maintains the local image characteristic，expecially on the edges and details，while it depresses the computational performance.The experiments compare the performance of the filter with other filters，and the results demonstrate the validity of the filter.

Key words: image smoothing, Gaussian filter, space distance, pixel distance, adaptive weight

摘要： 作为去除图像中噪声的图像增强技术，常用的图像平滑方法在提高局部信噪比的同时，也使图像产生模糊。为克服上述缺点，引入了自适应高斯滤波器，它结合了高斯滤波器和梯度倒数加权滤波器的特点，同时考虑了图像局部的空间距离和像素距离，以确定参与局部平滑的像素及其权值。该滤波器算法牺牲了简单平滑滤波器的计算性能，但很好地保留了图像的局部特点，特别是边缘和细节。实验比较了该方法与其他常用滤波器的性能，结果证实了该方法的有效性。

关键词: 图像平滑, 高斯滤波器, 空间距离, 像素距离, 自适应权值