An efficient mean filter algorithm

In order to improve the speed of mean filter algorithm and to reduce the redundancy operation, an efficient algorithm is presented in this paper. At first, an assistant array is built up to store the pixels´ value summation of each column. Then, according to the relationship between adjacent windows when filter window is siding along the rows, a recursive formula for calculating the midpoint mean of a new window is schemed out. Meanwhile, the recursive formula for updating assistant array items is obtained based on the relationship between adjacent rows´ pixels in the changing row process of midpoint of window. On the basis of these two recursion formulas, a high efficient mean filter algorithm is achieved. The speed of this method is independent of the variation of window size, which made it suitable for the process of large images with large filter kernel. And it is proved by experiments that the new algorithm is more efficient than the others.