The filters.mad filter automatically crops the input point cloud based on the distribution of points in the specified dimension. Specifically, we choose the method of median absolute deviation from the median (commonly referred to as MAD), which is robust to outliers (as opposed to mean and standard deviation).


This method can remove real data, especially ridges and valleys in rugged terrain, or tall features such as towers and rooftops in flat terrain. While the number of deviations can be adjusted to account for such content-specific considerations, it must be used with care.


The sample pipeline below uses filters.mad to automatically crop the Z dimension and remove possible outliers. The number of deviations from the median has been adjusted to be less agressive and to only crop those outliers that are greater than four deviations from the median.



The number of deviations from the median. [Default: 2.0]
The name of the dimension to filter.