How to Perform Ground Classification in LP360

In this 20-minute video, learn how LP360 can quickly and efficiently create ground classifications from LiDAR data.

LP360 is a powerful software suite designed for processing and analyzing geospatial data, specifically LiDAR point cloud data. Geospatial professionals can benefit from using LP360 for a variety of reasons, including ground classification.

In this informative video below, Martin Flood, Vice President of Special Projects at GeoCue, provides a demonstration of doing a ground classification in LP360 software with LIDAR data collected with a TrueView 535 3D imaging sensor. 

Ground Classification with LiDAR Data

In the demonstration, recorded live from the floor at Geo Week 2023, Martin Flood explains how LP360 can process LiDAR data to create a point cloud, which can be classified using different algorithms to identify the objects in the data set. Ground classification is a prime product generated from LIDAR data, especially when conducting topo work and generating one or two-foot contours. The standard unclassified data set used for this demonstration was collected over a test range at the GeoCue office in Huntsville, Alabama.

Flood points out different objects in the data set, such as trees, vegetation, the parking lot, buildings, and cars parked around. The goal is to extract and identify returns from the ground and build the ground surface without the buildings, vegetation, and other objects that have been classified. LP360 uses an adaptive TIN approach for ground classification, which starts from the lowest surface in the point cloud and merges or molds it to find the lowest continuous surface in the points. Flood mentions that there are user parameters to tweak, which he explains during the actual running of the point cloud task.

LP360 ground classification

Thinning the Data

In the demonstration, Flood also discusses data density, which is a challenge when working with LIDAR data. For instance, Trueview 535 data from drone LIDAR systems is very dense with millions of points collected from the ground and trees. This is often more than needed to generate a good ground surface, especially when generating one or two-foot contours. Flood explains that users can thin the data to reduce data density, making it more manageable for downstream workflows.

He also notes that the United States Geological Survey (USGS) has data standards for LIDAR data, with different quality levels for data sets. Flood plans to thin the data to what’s called QL one or quality level one USGS data, which is equivalent to about eight points per square meter, much thinner than drone data, but still very high resolution for generating two-foot contours or something similar.

LP360 Workflow

Throughout the video demonstration, Martin walks through the LP360 workflow and provides insight into the challenges and techniques involved in generating ground classification. He reviews the importance of filtering low points in point cloud data when using adaptive tin algorithms to build ground surfaces and demonstrates the step-by-step process of running different point cloud tasks in a sequence. He also showcases the live view filter feature to visualize and analyze the classified points.

When the classification is complete, he then reviews several automated and manual tools in LP360 to help clean and sort the data, to create the final deliverables such as contours or digital surface models.  

Share

GeoCue Support has written 707 articles