<div dir="ltr"><div class="gmail_quote"><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Date: Wed, 06 Aug 2008 11:31:26 +0100<br>
From: Evan Harrell <<a href="mailto:evan.harrell@wadh.ox.ac.uk">evan.harrell@wadh.ox.ac.uk</a>><br>
Subject: [Insight-users] B-Spline Deformable Transform for 3-D point<br>
sets<br><br>
<br>
Hi Everyone,<br>
<br>
I am trying to register two 3-D point clouds in an ICP framework using a B-<br>
Spline deformable transform. I am under the impression that ICP requires the<br>
Levenberg Marquardt Optimizer (not very good with BSplineDeformable because it<br>
takes 8 hours!). The output parameters I get with the following code are all<br>
0's and infinities. Should the BSplineDeformableTransform work for registering<br>
two 3-D point clouds? If so, where have I gone wrong?<br>
<br></blockquote></div><br><br>Hi Evan,<br> If you are interested in the nonrigid registration of 3D point sets, you might<br>want to take a look at <br> <a href="http://code.google.com/p/gmmreg/">http://code.google.com/p/gmmreg/</a><br>
where you can find C++, MATLAB and Python implementations of a robust <br>and fast registration algorithm for both 2D and 3D point sets. The thin-plate splines <br>are used as the transformation model for the nonrigid registration. The C++ code <br>
is not based on ITK but it does use CMake as building tool and vnl as numerical <br>library. <br><br><br>Thanks,<br>Bing<br><br><br><br></div>