Fusion between ground- and space-based mid-infrared observations

TL;DR

This paper introduces a novel method using Non-negative Matrix Factorization to merge space- and ground-based mid-infrared astronomical data, enhancing sensitivity, spectral coverage, and angular resolution.

Contribution

The paper presents a new NMF-based technique for combining mid-infrared data from different telescopes, improving data quality and resolution in astronomical observations.

Findings

Proven efficiency on real mid-IR data

Enhanced sensitivity and resolution through data merging

Applicable to various hyperspectral datasets

Abstract

Mid-infrared astronomy (operating at wavelengths ranging from 2 to 25 $\mu$m) has progressed significantly in the last decades, thanks to the improvement of detector techniques and the growing diameter of telescopes. Space observatories benefit from the absence of atmospheric absorption, allowing to reach the very high sensitivities needed to perform 3D hyperspectral observations, but telescopes are limited in diameter ($< 1$ meter) and therefore provide observations at low angular resolution (typically a few seconds of arc). On the other hand, ground-based facilities suffer from strong atmospheric absorption but use large telescopes (above 8m diameter) to perform sub-arcsecond angular resolution imaging through selected windows in the mid-infrared range. In this Paper, we present a method based on Lee and Seung's Non-negative Matrix Factorization (NMF) to merge data from space and ground based mid-infrared (mid-IR) telescopes in order to combine the best sensitivity, spectral coverage and angular resolution. We prove the efficiency of this technique when applied to real mid-IR astronomical data. We suggest that this method can be applied to any combination of low and high spatial resolution positive hyperspectral datasets, as long as the spectral variety of the data allows decomposition into components using NMF.