We created a Proof of Concept with FME to detect roofs with the potential for installing solar parks from open data (Lidar). Roofs must have an area bigger than a certain threshold and have an appropriate orientation.
6. 20
22
FME
User
Conference
Describe the Problem
Proof of Concept for detecting roofs that are interesting for
operating solar parks using Open Data
Input:
• Lidar LAS files (open data)
• Building footprints (open data)
Output:
• Only Roofs with area >1000m²
• Orientation
• Slope
• Area
• Address
• Owner
• Company Number
9. 20
22
FME
User
Conference
Solution - Overview
General Steps:
1. Select all roofs with area > 1000m²
2. Clip the PointClouds
3. PointCloud filtering – keep roof points
a) PointCloudStatistics based
b) Outlier detection with PDAL
4. Calculating Normals for roof points with CloudCompare
(command line)
5. RoofPlaneSegmentation (GitHub exe)
6. Plane reconstruction (Python)
7. SurfaceNormalExtractor (FME hub) - updated to Python 3
8. Slope and orientation from SurfaceNormal
9. Join other sources: owner, address, …
24. 20
22
FME
User
Conference
Future Work
● Run on large datasets
● Analyse performance:
● Split large area into batches
● Hardware requirements
● …
● Analyse results for odd roof types
● Investigate Alternatives:
● Open3D for Roof Segmentation - segment_plane()
function
● Roof filtering: necessary?
● Sarthaktum/ roofn3d?
● …