Abstract: This paper proposes the algorithm of histogram specification based on multi-means normal distribution function.In order to get the function，the average of image gray is treated as the first mean that can produce a series of other means，and the contrast of image is taken as variance controlled by Gaussian peak gain.The multi-means normal function is taken as histogram transformation，and the image histogram is constrained into the form of multi-means normal distribution.Experiment shows that the algorithm can not only enhance the image contrast selectively and locally but also improve mean of image，thereby to get the best optical effect.

Key words: multi-means, normal distribution, histogram specification, image enhancement

摘要： 提出一种基于多均值正态分布的直方图规定化算法，以图像的灰度平均值作为首个高斯峰均值，计算出后续值，以图像的对比度作为所有高斯峰的均方差，通过峰值增益加以调节。以形成的多均值正态分布作为直方图变换函数，将直方图约束成多均值正态分布形式。实验证明，基于多均值正态分布规定化处理，不仅有选择地对某灰度范围进行局部对比度增强，而且能改善图像的灰度平均值，从而增强图像的视觉效果，层次丰富，细节明晰。

关键词: 多均值, 正态分布, 直方图规定, 图像增强