Cloud to Cloud registration RMS accuracy

[arjun]

Hi,

I've been using the cloud to cloud registration tool to register a number of scans together of a mine. We have used spheres inside but I am finely registering them together first and then picking the sphere to georeference them .

Once aligned CC was giving me an RMS of approx 0.080 - 0.093 when aligning each scan. My question is how do I interpret this in terms of accuracy? is that good? Visually they look great but is there a way to actually calculate the error between the scans in mm?

Thanks

[daniel]

Well, it's hard to tell because it highly depends on your data, your scanner, etc.

8 to 9 cm error seems neither bad, neither good. There may be room for improvements? (have you use the ICP Fine registration tool? If yes have you set the 'overalp' parameter correctly?).

To check the registration you can use the C2C distance computation tool (and focus on the overlapping parts).

Don't hesitate to post snapshots or even send me some data samples if you need more help.

[arjun]

Thanks Daniel,

I havent used the iCP fine registration tool, we thought it would be best to just fit the spheres as they are coordinated. If we then carried out a fine registration we are assuming that, that additional movement would affect the accuracy away from the spheres. Would we be right there?

I'll use the C2C distance computation as a comparison. Thanks for your input.

Yes that would be really helpful if we could send you some snapshots and samples if helpful.

Thanks

[daniel]

Oh, if you have used the point-pair based registration method, then it's the best solution indeed.