I Need Help with Aligning Multiple Point Clouds in CloudCompare

[komyash]

Hello there,

I am new to CloudCompare and have been experimenting with its powerful tools for managing and processing point clouds. I have run into an issue and would appreciate some guidance from the community.

I am working with several large point clouds from a LiDAR survey; each representing different portions of the same area. My goal is to align these point clouds into a unified dataset. While I have managed to import the data without issues; the alignment process is proving to be more challenging than I anticipated.

I selected key points and attempted to align the clouds manually. It works somewhat; but it is tough to achieve high precision.
I have used the Align ICP tool; but the results are not great. I suspect it is because my initial alignments are not close enough.
I attempted transformations to fine tune alignments, but the process feels clunky, and I wonder if I am missing a better workflow.

Is there a recommended method to improve initial alignment accuracy before using ICP?

Also, I have gone through this post; https://www.cloudcompare.org/forum/viewtopic.php?uipath=4083 which definitely helped me out a lot.

Are there specific settings in the ICP tool that might help in my case?

Thanks in advance for your help and assistance.

[daniel]

Hi,

I believe Align + ICP is definitely the best workflow.

And the key for ICP is to properly set the 'overlap' parameter correctly. It's the ratio of points belonging to the 'to-be-aligned' that will likely have a matching point in the reference cloud. If you are unsure, don't hesitate to start with a low/conservative value first, and increase it until it shifts everything in the wrong direction. You also should use the most dense cloud as the 'reference' one.