topic Re: Extract Rails from Point Cloud (3D Analyst Tool) in ArcGIS Pro Questions

Extract Rails from Point Cloud (3D Analyst Tool)

offi24 — Wed, 02 Jul 2025 07:59:19 GMT

Hello,

Does someone has experience with the Extract Rails from Point Cloud Tool. I have a few point clouds (from mobile mapping and UAV) and want to extract the rails out of it. I tried to us the workflow described in the following link (https://www.esri.com/arcgis-blog/products/arcgis-pro/3d-gis/extract-rails-and-centerlines-from-mobile-lidar-point-clouds). But already with the tool: Classify LAS by Height, I couldnt get an accurate classification. Could someone help me classify my point clouds.

Thanks in advance and best regards

Eric

Re: Extract Rails from Point Cloud (3D Analyst Tool)

gti_esri — Wed, 09 Jul 2025 13:34:06 GMT

Hi @offi24 ,

Yes I tried it yesterday...

On my side, I managed the classification in this way:

Input data = lasd file with pyramids, all points are unclassified and the adjustment of point cloud is done.

1) Classified the ground (arcpy.ddd.ClassifyLasGround)

2) Classified the noise

3) Draw rectangle around the railway area (keep only ballast points) - (See 2D View attachment)

4) Use Classify LAS by Height...I run 4 iteration with this tool, from 10cm --> 17 cm (for Europe railway)

5) Then you should be able to use the last tool for the rails extraction (ExtractRailsFromPointCloud)

Let me know if I can help more?

Re: Extract Rails from Point Cloud (3D Analyst Tool)

offi24 — Tue, 15 Jul 2025 08:23:54 GMT

Hi @gti_esri ,

Thanks for your help, the workflow is working only for my UAV captured data and not for the data that was provided by a mobile mapping survey, where the LIDAR was fixed on a train. You worked with UAV-data? And is your point cloud a real LIDAR point cloud or are they extruded from photographic data?

Best regards

Re: Extract Rails from Point Cloud (3D Analyst Tool)

gti_esri — Fri, 29 Aug 2025 14:13:27 GMT

Hello@offi24 : sorry for my late answer.

I'm experiencing the same issue when working with Lidar data collected from mobile mapping systems. The datasets I’m using lack essential attributes such as return intensity, number of returns, and return order. However, these attributes are critical for performing accurate point cloud classification.