R/generate_2d_gaussian.R

In rayimage: Image Processing for Simulated Cameras

#'@title Generate 2D Gaussian Distribution
#'
#'@description Generates a 2D gaussian distribution, with an optional
#'argument to take the gaussian to a user-defined power.
#'
#'@param sd Default `1`. Standard deviation of the normal distribution
#'@param power Default `1`. Power to take the distribution. Higher values will result in a sharper peak.
#'@param dim Default `c(11, 11)`. The dimensions of the matrix.
#'@param width Default `3` (`-10` to `10`). The range in which to compute the distribution.
#'@param rescale_unity Default `FALSE`. If `TRUE`, this will rescale the max value to one. Useful
#'if wanting to plot the distribution with [plot_image()].
#'@import stats
#'@export
#'@examplesIf interactive() || identical(Sys.getenv("IN_PKGDOWN"), "true")
#'image(generate_2d_gaussian(1,1,31), asp=1)
generate_2d_gaussian = function(
  sd = 1,
  power = 1,
  dim = c(11, 11),
  width = 3,
  rescale_unity = FALSE
) {
  if (length(dim) == 1) {
    dim = c(dim, dim)
  }
  mindim = min(dim)
  xy_ratio = dim[1] / dim[2]
  if (xy_ratio > 1) {
    x = seq(-width * xy_ratio, width * xy_ratio, length.out = dim[1])
    y = seq(-width, width, length.out = dim[2])
  } else {
    x = seq(-width, width, length.out = dim[1])
    y = seq(-width / xy_ratio, width / xy_ratio, length.out = dim[2])
  }
  testmat = matrix(0, length(x), length(y))
  if (sd == 0) {
    sd == 0.00001
  }
  if (length(x) > 0 && length(y) > 0) {
    for (i in 1:length(x)) {
      for (j in 1:length(y)) {
        testmat[i, j] = (dnorm(power * x[i], mean = 0, sd = sd) *
          dnorm(power * y[j], mean = 0, sd = sd))
      }
    }
  } else {
    stop("Kernel dimension too small")
  }
  if (rescale_unity) {
    temp = testmat / sum(testmat)
    return(temp / max(temp))
  }
  testmat / sum(testmat)
}