中文摘要：

提出一种新的利用图像灰度值自然对数比率邻域滤波的图像平滑方法。从现有邻域滤波模型出发，分析了其去噪原理，讨论了其规一化因子中像素灰度值相似函数存在的问题。在此基础上针对具有代表性的双边滤波模型进行修改，提出新的灰度值自然对数比率相似函数模型。新的灰度值自然对数比率双边邻域滤波模型不仅保持了原有双边邻域滤波模型较好的平滑效果，而且保留了尽可能多的细节部分，同时对于彩色图像还保持了原有图像的色彩过渡。实验结果表明，该方法达到了理想的去噪效果。

英文摘要：

A new method based on gray-level value natural logarithm ratio neighborhood filters was presented for image de-noising. Starting from the present neighborhood filters model, the denoising principles and the existing problem of the gray-level value in normalization factor were analyzed. As a solution, a new gray-level value natural logarithm ratio range filter kernel was pointed out for the Bilateral filtering, a representative method of Neighborhood filters. The new gray-level value natural logarithm ratio Bilateral Neighborhood filtering method can not only well restrain noise but also keep weak edges and details of an image as many as possible, and preserve the original color transition of color images. Experimental results show that the effectiveness for image de-noising of the new method is better.