<div dir="ltr"><div>Hi Elvis,</div><div><br></div><div>The official "starting point" is probably the wiki:</div><a href="http://www.vtk.org/Wiki/VTK">http://www.vtk.org/Wiki/VTK</a><div><br></div><div>There's also a mailing list where all the developers hang out:</div><div><a href="http://www.vtk.org/mailman/listinfo/vtk-developers">http://www.vtk.org/mailman/listinfo/vtk-developers</a></div><div><br></div><div>Regarding the anisotropic scaling, that could go directly into vtkLandmarkTransform alongside the current RigidBody, Similarity, and Affine settings.</div><div><br></div><div>By the way, I've been meaning for ages to fix the "Affine" code... it is implemented with a matrix inversion and it fails for certain landmark configurations. It should be using a generalized inverse of some kind to make it more robust.</div><div><br></div><div>Cheers,</div><div> - David</div><div><br><div class="gmail_extra"><br><div class="gmail_quote">On Wed, Apr 13, 2016 at 7:42 AM, Elvis Chen <span dir="ltr"><<a href="mailto:elvis.chen@gmail.com" target="_blank">elvis.chen@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div dir="ltr">hi all,<div><br></div><div>I have an algorithm that solves Orthogonal Procrustes Analysis (i.e. vtkLandmarkTransform) with anisotropic scales. That is, given two sets of homologous points, X, and Y, my algorithm solves the following registration problem:</div><div><br></div><div>Y = R S X + t</div><div><br></div><div>where R is 3x3 orthonomal rotation, t the translation, and S = diag(Sx,Sy,Sz) where the scales are anisotropic. It is the generalization of the similarity mode (isotropic scale) in vtkLandmarktransform.</div><div><br></div><div>I have further extended/incorporated it into an ICP framework. The publication can be found at <a href="http://link.springer.com/article/10.1007/s11548-015-1199-9" target="_blank">http://link.springer.com/article/10.1007/s11548-015-1199-9</a>, and an initial implementation (C++/Matlab) can be found at <a href="https://github.com/chene77/RobartsICP" target="_blank">https://github.com/chene77/RobartsICP</a></div><div><br></div><div>I would like to implement them as vtk classes and would appreciate any help and tips. I am a typical vtk USER and not necessarily a developer; I am sure I can and need to learn quite a bit.</div><div><br></div><div>To start, I am thinking of subclassing vtkLandmarkTransform. I would appreciate any opinion/suggestions as on:</div><div><br></div><div>- is this the best place to start, and</div><div>- any coding guideline/documentation,</div><div><br></div><div>Thank you,</div><div><br></div><div>--</div><div>Elvis Chen</div><div>Robarts Research Institute </div></div>
</blockquote></div><br></div></div></div>