基于均值查找的快速中值滤波算法

针对传统中值滤波算法时间复杂度高、运行速度慢,难以满足大型图像数据实时处理的问题,提出了一种快速中值滤波算法,将确定中值元素的过程由排序运算转换为基于均值对集合的二分查找,算法不依赖于滤波窗口的形状以及相邻窗口的相关信息,有效提高了中值滤波的执行效率,使传统中值滤波算法的时间复杂度由O(nln n)下降至O(n).实验中,该算法应用于大型图像序列的滤波处理,其运算速度提高到传统中值滤波算法的3倍以上,并且算法运行时间仅随滤波窗口大小线性增长,可以满足大尺度滤波窗口对大型图像数据实时处理的需求,具有显著的实际应用价值.

Fast Median Filtering Algorithm Based on Mean Searching

In order to solve the problem of traditional median filter that filtering algorithms are very time consuming and can't meet the need of real time processing, a novel fast median filtering algorithm was proposed, which changed the process of finding median element from quick sorting into binary searching. The new algorithm didn't rely on filtering window shape and the relative information of adjacent windows, and effectively reduced the time complexity of standard median filter from O(nln n) to O(n). Experiments showed that the new algorithm is at least three times faster than standard median filter and computing time is increasing linearly with filtering window size, which proved that it has special value in large-scale image real time processing.