The code and software icp results are inconsistent

[darkdelting]

software:

code:

[dgirardeau]

Sorry, I am not sure to fully understand the issue? Do you mean that the default parameters of the CCCoreLib library are different from the CloudCompare GUI version? (in which case that's definitely possible).

[darkdelting]

Sorry, I am not sure to fully understand the issue? Do you mean that the default parameters of the CCCoreLib library are different from the CloudCompare GUI version? (in which case that's definitely possible).

Yes , I use the CloudCompare v2.13.2 and I'm not sure if it's the problem I'm using.

[dgirardeau]

Well, CCCoreLib and CloudCompare are 2 different projects. They don't need to use the same default parameters.

[darkdelting]

Well, CCCoreLib and CloudCompare are 2 different projects. They don't need to use the same default parameters.

Do you mean , That is CCCoreLib and CloudCompare are 2 different projects and use the same algorithm, but default parameters is different ? what is CloudCompare default parameters . Can I align the performance of the two algorithms?

[dgirardeau]

You can see how CloudCompare calls CCCoreLib here:

• https://github.com/CloudCompare/CloudCompare/blob/master/qCC/mainwindow.cpp#L3798
• https://github.com/CloudCompare/CloudCompare/blob/master/qCC/ccRegistrationTools.cpp#L46

Most of the parameters are set via the GUI, so it may not be simple to find the default values easily (you can also look at the dialog's ui file).

[darkdelting]

Thanks for your anwser!! As you can see from the code, the parameters have been recalculated in the backend ，which is diffrent from ui

[dgirardeau]

Well, 'baseIncrement' is just for memory allocation... We then simply update the overlap ratio to only consider points within the 'max search distance' (if any). Because the UI let's you play with various conflicting concepts.