ImageMM: Multi-Frame PSF-Aware Deconvolution — Practical Notes & Implementation Diagram

This document summarizes a practical implementation of the multi-frame restoration method described in \citet{Sukurdeep2025}, with several engineering extensions for real astrophotography data. In particular, beyond the white paper’s forward model, MM update, and optional super-resolution path, our implementation now supports \emph{rejection-aware multi-frame deconvolution}: a robust image-integration prepass first identifies per-frame rejected pixels (e.g., satellite trails, airplanes, transient defects, hot-pixel outliers) using the same rejection statistics already employed for stacking, and these rejected pixels are then excluded from subsequent multiplicative updates. We aim to remain faithful to the underlying ImageMM formulation while documenting the practical extensions that improve resilience on real sub-exposure stacks.