Spectral Density

The joint distribution of reflectance against wavelength across every pixel in the cube, drawn as a two-dimensional histogram: a luminous band of probability mass with the mean spectrum threaded through it.

Usage

hsa_spectral_density(
  cube,
  nbins = 128L,
  limits = NULL,
  transform = c("log1p", "identity"),
  normalise = c("none", "band"),
  palette = "mako",
  show_limits = FALSE,
  probs = c(0.02, 0.98)
)

Arguments

cube

An hsi_cube (from hyperspectR) or a 3-D array with dimensions (rows, cols, bands).

nbins

Number of reflectance bins. Default 128.

limits

Numeric length-2 reflectance range to bin over. NULL (default) uses the 0.1st and 99.9th percentiles of the finite data, so a handful of saturated pixels cannot flatten the whole figure.

transform

Count transform: "log1p" (default) or "identity". Reflectance histograms are heavy-tailed, and on a linear count scale only the mode is visible. Colourbar labels are back-transformed, so the legend still reads true counts.

normalise

"none" (default) or "band". Per-band normalisation stops a band with many masked pixels from reading as empty.

palette

Palette name passed to hsa_palette().

show_limits

Draw horizontal rules at the percentile limits the image functions would use. Default FALSE.

probs

Percentiles for stretch = "percentile". Default c(0.02, 0.98).

Value

A ggplot2 object.

Details

This is the only composition here that is not an image, and it is the one the others should be read against. Every rendering in this package applies a contrast stretch; this figure shows the distribution that stretch is being applied to, so a reader can see whether the limits are reasonable or whether a striking image is the product of an aggressive one. Passing show_limits = TRUE draws those limits directly onto the distribution.

Values outside limits are dropped, not clipped. Clipping would pile their mass into the end bins and manufacture bright edges at the extremes, which is exactly the kind of artefact this figure exists to expose.

Examples

cube <- hsa_demo_cube()
hsa_spectral_density(cube)