Lidar360 ALS Forestry - point classification using colour values?

[Cameron Dowd]

Hello,

I am using the software for forestry in coastal areas, where we have collected colourised point clouds using UAVs. The area has section of water, sand, and small shrubs and bushes. We want to count and measure some of these smaller pieces of vegetation (between 0.6m to 1m high, maybe 0.6m in diameter). We've had issues using the standard ground classification tools in the program to robustly classify ground points however, typically it classifies too many of the small vegetation as ground as well.

Is there a workflow in Lidar360 where we can classify ground points based on the point colour? Or assign points to specific classifications based on their point colour? The ground points and vegetation points are always significantly different colours, which would help a lot to improve how robustly we detect the smaller vegetation.

Thanks,
Cameron

[Solaire Guo]

Hello Cameron,

Thank you for your feedback!
Apologies, but currently LiDAR360 does not support classification based on point cloud color.
However, if you would like to classify lower vegetation, you can try using the "Classify by Height Above Ground" function, which allows you to extract lower vegetation based on parameters you set.
Alternatively, you can use the "Classify by Deep Learning" function to perform an overall classification of the point cloud and then manually fine-tune it using the Classification Editor.

If you need any further assistance, please feel free to reach out.

Best regards,

[Cameron Dowd]

Thanks for the response. My challenge is that the automatic ground detection often classifies the smaller shrubs as ground, so I wanted a more robust way to classify. I have not yet gotten the deep learning model to work robustly either. Is there an easy method to filter points by colour so I can select only points of a certain colour to manually classify as vegetation or ground? I tried with the RGB visualisation option but it didn't seem to limit my selections to only that colour.

As an interim I will try to classify the points in other software based on colour.