Abstract: This paper presents a new automatic smoothing approach for color images.The method analyzes the human eyes by spatial frequency and Contrast Sensitivity Function（CSF），and defines the Gaussian mask by the viewing distance automatically.Then it utilizes the human visual mechanism of color detail blindness and simultaneous contrast，therefore decides the Gaussian mask according to the color difference and location between adjacent pixels.The algorithm can preserve the edges and details，while quickly smooth the image，and the results show its validity.

Key words: Gaussian filtering, image smoothing, color difference, bilateral filtering, Contrast Sensitivity Function（CSF）

摘要： 提出了一种新的彩色图像自动平滑方法。该方法首先根据人眼视觉系统的空间模糊特性，由观测距离自动确定初始高斯模板；并进一步从人眼视觉感知中的彩色细节失明及同时对比特性出发，利用象素间的色差和位置关系确定最终的高斯权值矩阵。算法对图像进行平滑的同时，可以保持图像细节和边缘不变。实验结果表明了算法的有效性。

关键词: 高斯滤波, 图像平滑, 色差, 双边滤波, 视觉灵敏度函数