next up previous contents index Russian
Next: Scan lines averaging Up: Image formation and processing Previous: Averaging data   Contents   Index


Median filtering

Figure 1.8: Median filtering
\includegraphics[width=\textwidth]{median}

Good results can be obtained using median filtering. This non-liner image processing method allow to remove sharp peaks, but leave steps untouched. The last differ this method from averaging. Let's explain this filtering method using one-dimensional example. If we put point's coordinate on horizontal line, and point's value on vertical line, we obtain two-dimensional profile. For filtering point we get values of its neighbors and put them into table. Then we sort this table by growth, and take value in middle table cell as a new value of the point.

Thus if we had a sharp peak in a point, the point's value hit the last cells in the table and does not go to the final image. But steps go through this routine without any change (see pic. 1.8). It is very easy to notice a difference between results of median filtration and averaging (see pic. 1.7, 1.8).


next up previous contents index Russian
Next: Scan lines averaging Up: Image formation and processing Previous: Averaging data   Contents   Index
Filonov 2005-02-04