R/smooth_gauss.R

In dfsp-spirit/haze: Smoothing of per-Vertex Data on Triangular Meshes

# This function is currently disabled as the Rvcg::vcgSmoothPVD function can only be found
# in my own branch of Rvcg and it still needs some work and proper tests.

# #' @title Smooth per-vertex data based on a mesh using true Gaussian smoothing.
# #'
# #' @description This function performs Gaussian smoothing of vertices in the geodesic neighborhood of all vertices. This is computationally quite expensive and requires large amounts of memory for large meshes.
# #'
# #' @inheritParams pervertexdata.smoothnn
# #'
# #' @param fwhm scalar double, smoothing kernel full width at half max
# #'
# #' @param trunc_factor scalar double, truncation factor for Gaussian neighborhood, in Gaussian standard deviations. Must be greater than 1.0 to make any sense, leave as is if in doubt. Larger values mean that a larger geodesic neighborhood needs to be computed on the mesh, which is slower.
# #'
# #' @return numerical vector, the smoothed data.
# #'
# #' @seealso \code{\link{pervertexdata.smoothnn}} can be used to approximate Gaussian smoothing with several iterations of nearest neighbor smoothing, and is a lot faster for large meshes.
# #'
# #' @examples
# #' \dontrun{
# #' mesh = rgl::tetrahedron3d();
# #' pvd = rnorm(nrow(mes2$vb), mean = 5.0, sd = 1.0);
# #' pvd_smoothed = pervertexdata.smoothgauss(mesh, pvd, fwhm=5.0);
# #' }
# #'
# #' @keywords export
# pervertexdata.smoothgauss <- function(surface, pvdata, fwhm, trunc_factor=3.5) {
#
#   if(requireNamespace("Rvcg", quietly = TRUE)) {
#     return(Rvcg::vcgSmoothPVD(ensure.tmesh3d(surface), pvdata, fwhm = fwhm, trunc_factor = trunc_factor));
#   } else {
#     stop("The 'Rvcg' package is required to use this functionality. Use 'remotes::install_github('dfsp-spirit/Rvcg', ref='smooth_pervertex_data')' to obtain the correct package version.");
#   }
# }
#