Morphological & median filter

Haris khan

2016 3 2

0 个回答

4 次查看（30 天）

0 个投票

Let the median value of the increasingly ordered samples of noise-free pixels in the (2Lf + 1) × (2Lf + 1) filtering window W(i, j) centered at (i, j) be m(i, j). The true value r(i, j) of the noise pixel can be estimated by combining the output of the conditional morphological filter with that of the improved median filter, i.e., r(i, j) = conditional opening + w(i, j) · m(i, j) + conditional closing/ w(i,j)+2 where w(i, j) denotes the weighted coefficient of the median value m(i, j). Extensive simulations show that the decreasing value should be assigned to w(i, j) with the increasing noise ratio to yield good restoration results. Therefore, w(i, j) is formulated as: w(i, j) = 1 - R/2R

where R= S/M*N in this formula "S" is detected noise pixel in the image where "M" and "N" is the total numbar of pixel in the horizontal and vertical dimension of images..