2 weeks ago
I have a question about the Point Cloud Classification tool. I have found when I change the projection of my point cloud, my output is different, even with the same parameters set in the point cloud classification. Do I need to use a certain projection? The classifer uses meters (for example, under object parameters, you can set Min area and Max area in square meters). I find that I get classified objects that are much smaller than the min area I set, which makes me think it is not reading the units correctly. Thoughts on why this might be?
Thursday - last edited Thursday
Any thoughts from anyone on this? Sometimes I get single pixels classified as building/objects when I go through the point cloud classifier. Why does this happen when I have set a minimum area of 100 m2?
What map projections and units are you using for each of your point cloud files? Do your point clouds also have different vertical projection parameters (spheroid, datum, units)? I would expect different results from the Point Cloud Classification tool if the images are in different projections.
What is the terrain like for your point cloud? For example, is it a mostly forested or an urban area? If the input point cloud does not have any buildings you can disable the building extraction by setting the Plane Offset to a negative value such as -1.0.
Does your input point cloud file have color information? Are you using both the Object and Vegetation parameters?
Hexagon Geospatial Support
Thank you for your reply Stephen! The point clouds as I obtained them are in State Plane, unit US survey feet, spheroid GRS 1980, datum NAD83.
I realize that the results are different with different projections- part of my question is: why? That seems to indicate there are "right" and "wrong" projections to use, so what is the right one? I had thought the classifier tool would read the projection of the file and take that into account in its classification, but that must not be what's happening. If that were happening, it seems that the results should be the same no matter the projection. If it's not considering the projection information, how can it determine things like min and max area of objects? When I run the classifer without changing the projection from State Plane/feet, I get very small blips ID'd as objects/buildings, despite setting the minimum area to 200+ square meters.
My area is mostly rural (Ohio), and I am trying to identify buildings. I do not need to ID vegetation. I do have RGB info, but found it not useful. Sometimes forested areas were ID'd as buildings, though I had hoped RGB encoding the point cloud would prevent this, since the buildings are generally white in the RGB images.
Thank you again for your help! Please let me know if I can provide further clarification. You guys are the best.