Build an adaptive multi-band/spatial support mask — build_adaptive

Builds a foreground support from band-wise or wavelength-slice detection evidence rather than from a single white-light image. This is useful for IFU cubes where the source is persistent over many wavelength channels but the background may vary spatially or spectrally.

Usage

build_adaptive_support(
  input,
  bands = NULL,
  transform = c("none", "asinh", "signed_log1p", "log1p", "copula_uniform",
    "copula_gaussian"),
  sky_method = c("border", "all"),
  border_fraction = 0.1,
  z_threshold = 3,
  min_band_persistence = NULL,
  single_band_z = Inf,
  smooth_sigma = 0
)

Arguments

input: A FITS-like object with imDat, or a 3-D cube array.
bands: Wavelength slices used to build the support. May be NULL for all slices, a numeric vector of indices, or a character vector matching cube dimnames along the third dimension.
transform: Per-slice image transform before robust sky normalization. Options are "none", "asinh", "signed_log1p", "log1p", "copula_uniform", and "copula_gaussian".
sky_method: Pixels used for robust sky/noise estimation. "border" uses image borders; "all" uses all finite pixels.
border_fraction: Fraction of rows/columns used as image border when sky_method = "border".
z_threshold: Per-slice detection threshold in robust sigma units.
min_band_persistence: Minimum number of selected slices above z_threshold. If NULL, defaults to two slices when possible.
single_band_z: Optional single-slice rescue threshold. Use Inf to disable.
smooth_sigma: Optional Gaussian smoothing applied to each significance map before thresholding. Requires EBImage or imager.

Value

A list with collapsed, reconstruction, mask, evidence, evidence_count, z_maps, band_stats, and metadata.