Pete Gadomski
Pete Gadomski
To facilitate scientists using this code, we should add bindings to commonly-used higher-level languages. For Python, try boost.python?
They're broken ATM.
Several users have reported that this C++ implementation is slower than the Matlab implementation.
We might be able to get away with lower error tolerances in some cases, so we should let the user set fgt's epsilon value.
For large sigma2 values, it can be better to use ifgt, then switch to direct-tree when the sigma2 values get small enough. I used to have switching built-in, but that...
PDAL could be used to ingest points — since there's PDAL methods to turn point views into Eigen matrices, it shouldn't be too hard to add an API call or...
This will be a part of our public API that can be used by downstreams.
Andriy's reference implementation came with a nonrigid_lowrank transformation, which used some tricks to speed up the very slow nonrigid registration. I think we could use https://github.com/m-reuter/arpackpp to do the eigenvalue...