Reconstruct and threshold starlet scales to produce a region mask

Usage

threshold_starlet_regions(
  dec,
  keep_scales = 2:5,
  include_coarse = FALSE,
  denoise_k = 3,
  mode = c("soft", "hard"),
  threshold = c("mad", "abs"),
  k = 3.5,
  tau = NULL,
  positive_only = TRUE,
  per_scale_positive = TRUE,
  k_hi = 5,
  k_lo = 3,
  area_min = 12,
  keep_negatives = FALSE,
  ring_nbins = 8
)

Arguments

dec: sagui_starlet object (or a matrix; if matrix, J inferred from keep_scales)
keep_scales: integer vector (e.g., 3:6)
include_coarse: logical
denoise_k: NULL or numeric (pre-recon per-scale)
mode: "soft" or "hard" (per-scale)
threshold: "mad" for k·MAD, or "abs" for absolute tau
k: numeric; multiplier for MAD (unused; kept for compat)
tau: numeric; absolute threshold when threshold == "abs"
positive_only: logical; TRUE uses rec>thr; FALSE uses |rec|>thr
per_scale_positive: logical; clamp each reconstructed starlet scale to positive values before thresholding.
k_hi: numeric; high MAD threshold used to define seed pixels.
k_lo: numeric; low MAD threshold used to define candidate pixels.
area_min: integer; minimum connected-component area to keep.
keep_negatives: logical; retain negative reconstruction values.
ring_nbins: integer; number of radial bins for background MAD.

Value

list(soft_rec, hard_rec, mask, sigma, threshold, seeds, candidate)