morphmet wrote: > Hello, > > I am interested in trying 3D thin-plate splines for registration of two 3D > surfaces, and was wondering if someone has such a code available. > > Thank you > Ira > >
Your question is a bit ambiguous. Do you really mean TPS for registration, i.e., the construction of relative warp scores plus uniform component for subsequent MV analysis, or do you want to (just) look at the splines? Do you really want "code" to incorporate into your own software development, or are you looking for programs that produce TPS and/or their associated numerical values? If you are programming, 3D TPS is a straightforward extension of 2D. Just add the appropriate coordinate rows/cols to matrices and use |r| instead of r^2log(r^2) in the construction of the L-matrix. Explained in Bookstein's Orange Book. Also, discussed in my Intro chapter to "Modern Morphometrics in Physical Anthropology." If it is software you seek, the three programs with which I am somewhat familiar that produce 3D TPS are: Morpheus et al. (Windows) - does the 3D superimposition and produces plots of highly customizable, 3 to m splines stacked along a specified axis. Old 16-bit code and poorly documented now. This is the program I use most because of familiarity. The author is not that bright, but can be somewhat helpful and responsive to persistent e-mail pleas for assistance. Morphologika (Windows) - I am recommending this to more and more folks because of its internal production of PCAs and visualizations of the same (and other new features). It handles 3D data and produces a 3D TPS. It's been a while since I played with that, but I think its positioning might have been reasonably dynamic. Edgewarp (Linux and, I think, Mac (unix-based OS-X) and, maybe, Windows + cygwin) - steep learning curve, limited documentation, but leading-edge features and direct support for volumetric data. It supports interactively scaled, positioned, and oriented splines in space, again with support for volumetric data, and can generate "film strips" of splines moving and deforming along anatomical paths. Links to all can be found in the software section at http://life.bio.sunysb.edu/morph You should be aware, or will soon find out, that 3D TPS are not nearly as satisfying as their 2D counterparts. The deformation represented is an interpolation of volumetric changes that cannot be displayed very well as 2D projections. Interactive computer displays help. Static print displays are even more limited. The best static results tend to be splines for some plane of interest through a space or volume. Best, dslice ([EMAIL PROTECTED]) -- Dennis E. Slice Department of Anthropology University of Vienna ======================================================== -- Replies will be sent to the list. For more information visit http://www.morphometrics.org
