Boulders and river processing problems in Cloudcompare

[soyukkin]

Hello everyone. I’m a final year student working on my final year project in Hong Kong.

My project is to scan a river which has been modified by human and there are some rock boulders nearby it. My jobs are to compare the point cloud data collected by airborne, handheld and backpack LiDAR.

I have four focus. They are the 1) boulder distribution, 2) boulder dimension 3) delineation of river 4) geometry of river. I’m still looking for alternative methodology to give better results.

For boulder distribution, the major issue is to identify the boulders from point cloud. I know that I can extract the boulders manually by looking at the point cloud. But I’m looking for an automatical way to extract the boulders. I have used the qCANUPO algorithm but it didn’t work at all. Im guessing the steep terrain is limiting the algorithm to function. Are there other methods for me to automatically pick up the boulders from point cloud? Or there are some ways to visualise the elevation in point cloud so that I can identify the protruding boulders?

For boulder dimension, I have used the point picking tools to manually measure the axes for boulders. That works pretty well. But are there also other methods that can automatically measure the axes of an object?

For the river, it is very challenging. I have to identify the boundary of this river from point cloud but I have no idea how it works. I know that water makes the point cloud disappear and decreases the point density. Is there a way to visualise the point cloud density as a scalar field? Or there are other methods to help me identify the river boundary?

Also, when I draw a polyline on the point cloud and export it. It has an elevation higher than the original position. Is there a method to prevent this problem? Because I want to draw lines on my point cloud.

If u have any experience on processing boulders and rivers point cloud data, please let me know and share your ideas below. That would be great. Thanks!

[daniel]

1) CANUPO could work, but it's probably a matter of using the right scales (and it depends on the scale variability of your boulders by the way). It's also a matter of training it properly...

2) For the dimensions, I don't see any obvious answer. Unless you assume boulders are equivalent to spheres and you can fit a sphere on the segmented boulders

3) You can compute the density with https://www.cloudcompare.org/doc/wiki/i ... c_features

4) Not sure to understand your issue. We would need more context / data / examples to help you.