ICP registration - Deviations between source and target file

[Sralelb]

Checklist

• I have searched for similar issues.
• For Python issues, I have tested with the latest development wheel.
• I have checked the release documentation and the latest documentation (for master branch).

My Question

I used the example for icp registration from:
http://www.open3d.org/docs/0.7.0/tutorial/Basic/icp_registration.html
to compare cad data. I did not change that much of the example code, basically played around with "threshold" parameter and added a convergencecriteria with "max_iteration=2000".
My example is basically only a 2d-problem. I want to compare both contures and find a good fit solution instead position it by hand.
The result looks even promising but there are some mistakes if you have a closer look. Obviously it seems to me that the angle between contiur 1 and 2 is not correct yet (look at the red marked spots). If have attached a snapshot of both contures before point-to-point ICP and afterwards, as well as my used data.

Aufriss_Original_Punkte.txt
Profilscan Reklamation_Punkte.txt
.

[theNded]

ICP only guarantees local optimal solutions. Unfortunately, to the best of my knowledge, I don't think any automatic algorithms can perfectly align your surfaces due to their non-linearity, high local similarity, and partial incompleteness.