This analytic semantically segments, i.e. classifies, a point cloud. It runs as a fully automatic self-service triggered by the user. It has been optimized for Power & Utilities case to provide fast results. This algorithm was especially optimized for Power Distribution Lines assets.
This tutorial details the step-by-step sequence to obtain and view a classified point cloud. Limitations are also stated in last section.
1. Output classes
Seven classes are obtained :
|CLASS||LAS CLASS NUMBER|
|Wire - Conductor||14|
A .las point cloud, without classification, or carrying an anterior classification, that will be overriden during the process.
3. Generating the classification
Step 2 - From left bar, go to the Analytics section :
Step 3 - Open P&U BASIC ANALYTICS section, go to Automatic Point Cloud Classification :
Step 4 - Press LAUNCH to start to select the input file :
Step 5 - Choose among the point clouds available in your survey, onto which one you want to run the classification :
Step 6 - Press Launch automatic point cloud classification, then Finish :
4. Progress & Completion Status
You can check status anytime from the analytic LAUNCHED tab :
A notification is equally issued at completion :
5. Browsing the classified point cloud
Exiting Analytics section, or from any other section in the platform, follow sequence below :
Step 1 - Go back to Layers section :
Step 2 - Select 3D view from the toolbar on right side :
Step 3 - Go to BASE LAYERS section, to find your newly classified point cloud :
Step 4 - Click on the layer to open the information panel on right side ; a colored chart allows to identify classes. Use zoom functions to examinate the point cloud :
6. Exporting the point cloud
In case you need to export classified point cloud for third-party usage :
Step 1 - Go to Downloads section from the left bar :
Step 2 - Select and export your file :
7. Renaming or Deleting Classified Point Cloud
Two methods are available :
1. From the drop-down menu, on Layer :
or 2. From the information panel drop-down menu :
8. Known Limitations
8.1. Very Low Density Case
For point clouds containing zones with a very low point density, there may be some locally improperly classified items. In example hereafter, a portion of power lines is seen green as classified into vegetation. Please ensure to provide dense enough point clouds for maximal results.
8.2. Power Transmission Lines Case
The algorithm employed in this analytic was trained on Distribution Lines models. While running it on Transmission Lines, abnormal detections of lines as vegetation may be encountered.
8.3. Noisy Point Clouds
Point Clouds with noise, especially with buildings and close vegetation, may result in buildings borders sometimes classified as vegetation, as in example below :