v.outlier removes outliers in a 3D point cloud. By default, the outlier
identification is done by a bicubic spline interpolation of the
observation with a high regularization parameter and a low resolution
in south-north and east-west directions. Those points that differ in
an absolute value more than the given threshold from a fixed value,
reckoned from its surroundings by the interpolation, are considered as
an outlier, and hence are removed.
There is a flag to create a vector that can be visualizated by
qgis. That means that topology is build and the z coordinate is
considered as a category.