weightmask-dev

name	weightmask-dev
description	Develop weightmask (weight/confidence maps and defect masks) — YAML config surface, quality-bit contract, streak detection, synthetic benchmarks
metadata	{"project":"weightmask","stack":"python, numpy, pixi"}

weightmask development

Builds weight maps, confidence maps, and defect masks for astronomical FITS images, aimed at stacking/coaddition: usable mask plane, inverse-variance estimate, and separate outputs for key contaminants. See docs/ and TODO.md.

Contract

• Canonical YAML config surface: flat_masking, sep_background, sep_objects, streak_masking, variance.
• NumPy array/header contract with named quality bits, explicit set_means_flagged mask polarity, and versioned provenance metadata.
• Primary outputs: weight or confidence map, combined bitmask, inverse-variance map, sky map, per-contaminant masks (--individual_masks).

Detection

• Default streak detector is auto_ground (percentile rescaling + smoothing, compact-source suppression, multi-scale Canny + Hough/KHT segment extraction, trail-aligned strip refinement, MRT-like Radon rescue, optional sparse RANSAC).
• frangi_legacy remains a COMPARISON path only — do not regress it to default.
• Bad pixels from flats, saturation/bleed trails, cosmic rays, objects, and linear streaks.

Benchmarks (statistics-principled)

• pixi run benchmark-synthetic — synthetic suite vs baselines (--with-baselines): inject known defects and verify recovery statistics.
• pixi run benchmark-megacam — Megacam real data
• pixi run benchmark-acs — ACS comparison
• pixi run test, pixi run lint, pixi run format

CLI

weightmask science.fits --config weightmask.yml; MEF inputs supported.