Improving Registration
From the desktop version of the software, you can improve alignment further by using these tools:
Selective Alignment
When using [Visual Alignment] to align scans together, you can selectively mask only the areas to use for alignment. For example, removing high-variance aspects like tree foliage, grass, people or traffic to guarantee a more accurate result.
You'll see this terminology thrown around so we will clarify it here: Each scan has a 'visual' version and a 'cloud-to-cloud' version. 'Visual' refers to the point cloud output that you see in the main viewport. 'Cloud-to-cloud' refers to a version of the point cloud used only for alignment purposes.
When masking/deleting points, you have the option to affect one or both, so pick accordingly where appropriate. For alignment, masking will only be affecting the Cloud-to-cloud.
Find the masking tool at the top, select between different selection modes. Use the options at the bottom to delete or restore points relative to the selection to get your desired result:
Once the masking has been made, running optimisation may yield better results. In the example below, tree foliage was deleted so that it does not introduce noise into the alignment, note the better error rate of 0.002m, compared to 0.004m without the masking.
Last updated