Документ взят из кэша поисковой машины. Адрес
оригинального документа
: http://www.spm.genebee.msu.ru/manual/en/node13.html
Дата изменения: Fri Feb 4 20:38:59 2005 Дата индексирования: Mon Oct 1 20:28:50 2012 Кодировка: |
There is some noise component in every surface image. To reduce or remove this noise component height in every point can be changed to average height value of its closest neighborhood (see pic. 1.7). If it doesn’t work - noise level is too high - more complex method should be applied. You can, for instance, remove high-frequency part from the initial signal. In fact, the averaging we described before is a kind of high-frequency noise filtering. The next thing you can do is to increase averaging region's size. Let's look at one-dimensional (a sting, not a matrix) example of such averaging operation.
Let's draw a plot, on which X-axis corresponds to coordinate of a point, and Y-axis - to point value. As a result we have string profile. This profile before and after filtration is represented on pic. 1.7.