Approaches to Vegetation

This work was focused on uses for intervisibility computations, but has other applications. 

Two grids. You can use two of the following three grids, and compute the third on the fly:
  1. DTM: digital terrain model, or bare earth.
  2. DSM: digital surface model, or first return
  3. HAG, or height above ground (DSM- DTM). 

You could also try creating  vegetation height grids from medium scale data sets.

Point cloud Software can search the neighborhood of a point to find vegetation, getting either presence/absence or a density.
Vegetation density voxel grids These provide indexing for rapid access, and significant compression

Create major categories ground, vegetation, buildings, and power lines.  I have not actually seen a point cloud with classified power lines, but  LAS 1.4 has new categories for them, so they will be appearing.  They are currently being created with the Reserved classifications, which will be different for every program that classifies data.

Classify point cloud  
Create DTM  
Split LAS file into classification layers
  • Unclassified
  • Ground
  • Vegetation
  • Buildings
  • Power lines
  • Other (high/low noise, water; consider deleting)
Remove ground returns
  • Replace with DTM
Extract buildings
  • Use grid, either separate or combine with DTM
  • Use point cloud
Remove vegetation
  • Replace with voxel density grids
Deal with unclassified points
  • Improve classification
  • Delete (low importance "ground clutter")
  • Combine with vegetation in voxel density grids
  • Create separate voxel density grids

last revision 8/8/2012