关键词: 中值滤波, 符号检验法, 正态分布, Euclid距离

Abstract: This paper studies image median filtering and several fast algorithm，then designs and implements a novel median filtering algorithm based on proved sign test method.The new algorithm does not need to sort the pixel values in neighbourhood so as to greatly speed up the operation of median filtering.On the other hand，this algorithm employs a new statistics to solve the shortcomings of original sign test method，especially when samples of two massing are almostly the same.At last，the improved sign test method is employed to better the image median filtering algorithm.Algorithm analysis and a lot of experiment results show that the new algorithm can not only speed up the operation of image median filtering，but also retain more information of image edge，figure and texture，compared to several other fast algorithm of image median filtering.

Key words: median filtering, sign test method, normal distribution, Euclid distance