Building Filter and Extraction for dense datasets

Requires an LP360 Geospatial Advanced or above, LP360 Drone, or LP360 Land Standard license.

Some LP360 Point Cloud Tasks (PCT's) were originally designed for older manned lidar datasets, which have significantly lower point density than modern sUAS datasets. TrueView and other sUAS sensors have the capability of producing much denser point clouds, so these processes can take much longer if running on the entire dataset. A common strategy and workflow for processing high density datasets is to thin the point cloud first before running some tasks on the dataset, such as the Building Filter and Extractor. This involves running a few PCT's in sequence to greatly improve processing speed. This workflow is briefly reviewed below in sequential order:

Note: Ground Classification must be performed prior to building extraction.

Data Thinning and Filtering

• Run the Classify by Statistics PCT to thin the point cloud to a roughly 8 points per square meter level.
  
  • Input Classes: 0 (Unclassified) and 1 (Ground); Destination Class: Model KeyPoints
  • Cell Size: Start with 1 ft, but consider 0.5 ft for more detail in building classification.
• Note: Performing a Height Filter before executing the next step (Planar Point Filter PCT) to reclassify all features taller than your tallest building and all features shorter than your shortest building (i.e. vegetation, powerlines, etc.) can also help to decrease the size of the input class.

Building Filter 

Building filtering is also referred to as Building Classification. Run the Building Filter (Planar Point Filter PCT) on the thinned dataset. This initial classification may need some cleanup/manual classification before proceeding to the next step.

• Input Class: Model KeyPoints
• Use the Profile View to estimate the Minimum and Maximum height above ground* for your buildings. Round down slightly for shortest building, round up slightly for tallest building

Building Extractor

Run the Building Extractor (Point Tracing and Squaring PCT) on the building classification of the thinned dataset. If you have a number of buildings, use Feature Analyst to QA/QC the building extraction results.

• Default Task: Default Building Extractor
• Boundary Points: Buildings
• Output: set path > Polygons representing building footprints.

Proximity Classifier

Run the Proximity Classifier PCT to incorporate the classification of the thinned class subset back into the original dataset.