uniDEM supports numerous smoothing algorithms to eliminate noise and first surface DEM artifacts.

Blur

Several kernel based blurring options are available in unidem including boxcar, gaussian and percentile. Pixels in a selection will be modified while neighbours are included in the calculation.

1. Boxcar

   Any pixel that is selected will be smoothed by a square kernel defined by the kernel size which is convolved over the pixel and its neighbours.

   Default shortcut: none

2. Percentile

   Any pixel that is selected will be smoothed by the value within its kernel neighbourhood that is closest to the selected percentage value when ordered from lowest to highest.
   For example filtering at 50% is equivalent to a median filter.

   Default shortcut: none

3. Gaussian

   Any pixel that is selected will be smoothed by a gaussian shaped kernel defined by the kernel size which is convolved over the pixel and its neighbours providing more weighting to the central values of the kernel than the boxcar blur would.

   Default shortcut: none

Constrain Slope

Constrain Slope can be used to grow or erode selected regions beneath the Medium Gradient.

Default shortcut: none

Remove Spikes

Used to remove single pixels spikes where the central pixel differs from the average of its neighbours.
Two modes exist:

• Absolute - where the average difference between the pixel and its neighbours exceeds a fixed threshold
• Relative - where the average difference between the pixel and its neighbours exceeds some multiple of the standard deviation (σ) of the area

Default shortcut: none