Classify by Statistics is a Point Cloud Task in LP360 that can be created by selecting the “Add Task…” option under the Point Cloud Tasks menu. It is primarily useful for general data thinning – especially useful for very dense data such as that derived from imagery (dense image matching, DIM, data). For example, you can set the grid size to 2 meter and choose a median selector, such as one point. You will then have a 0.5 points/m2 point cloud based on the median points from the dense data. Another example, I can set the grid size to 1 meter and choose a median selector, two points. I will then have a 2 points/m2 point cloud based on the median points from the dense data.
The task creates cells of size ‘Cell size’ by ‘Cell size’. For each cell it calculates the statistics on the Z value (elevation). It then keeps the ones you check:
|Min||Point with lowest Z value; if more than 1 point with lowest Z picks a random one to keep.|
|Max||Point with highest Z value; if more than 1 point with highest Z picks a random one to keep.|
|Median||Point with median Z value.|
|Random||n points selected at random that are not the Min, Max or Median points if those options are selected.|
So if you set the task to keep Min, Max, Median and 5 random points and a cell size of 1 m you will end up with 8 ppsm in every cell where there is at least 8 points. For cells with less than 8 points you will keep them all. It is a very good way to thin dense data sets. In the US we often set these to generate 2 ppsm or 8 ppsm clouds since these values correspond to the USGS quality level densities for airborne lidar.