CloudCompare forum

Hi, I am new to CloudCompare and processing point cloud data so this could be a very simple question. For part of my undergraduate dissertation I am registering a large mobile lidar dataset with a large airborne lidar dataset using ICP. Both datasets are already well aligned however when I run the ICP registration the resulting T matrix is obviously incorrect.

To register the datasets I have created patches which contain both mobile and airborne data. I have then registered all of the mobile patches against all of the airborne patches. The cloud 2 cloud distances do improve after the registration of the patches but if I applied the given T matrix to the whole dataset both datasets would no longer be aligned.

Any help would be greatly appreciated.

Many thanks,
Max

Do you mean that you apply the registration process on a small patch and then you try to apply the same transformation to the whole cloud?

If yes, then depending on the size of the clouds and how well they match, it's totally possible that the 'local' registration does not apply to the whole cloud properly. Especially in presence of deformations, distortions, etc. This is generally not a good idea to process like this (or you should at least you several patches sampled all over the clouds).

Otherwise did I misunderstood something?