J-comb: An image fusion algorithm to combine observations covering different spatial frequency ranges

TL;DR

J-comb is a new image fusion algorithm that combines high and low-resolution astronomical images, ensuring Gaussian beam responses for easier multi-wavelength comparison and outperforming existing methods in accuracy and detail.

Contribution

The paper introduces J-comb, a novel linear image fusion algorithm that guarantees near-Gaussian beam responses and effectively handles non-overlapping Fourier domains, improving astronomical image analysis.

Findings

J-comb outperforms CASA-feather and MIRIAD-immerge in benchmarks.

Successfully applied to Orion A data, revealing finer details.

Produces high-resolution dust temperature and density maps.

Abstract

Ground-based, high-resolution bolometric (sub)millimeter continuum mapping observations on spatially extended target sources are often subject to significant missing fluxes. This hampers accurate quantitative analyses. Missing flux can be recovered by fusing high-resolution images with observations that preserve extended structures. However, the commonly adopted image fusion approaches do not maintain the simplicity of the beam response function and do not try to elaborate the details of the yielded beam response functions. These make the comparison of the observations at multiple wavelengths not straightforward. We present a new algorithm, J-comb, which combines the high and low-resolution images linearly. By applying a taper function to the low-pass filtered image and combining it with a high-pass filtered image using proper weights, the beam response functions of our combined images are guaranteed to have near-Gaussian shapes. This makes it easy to convolve the observations at multiple wavelengths to share the same beam response functions. Moreover, we introduce a strategy to tackle the specific problem that the imaging at 850 um from the present-date ground-based bolometric instrument and that taken with the Planck satellite do not overlap in the Fourier domain. We benchmarked our method against two other widely-used image combination algorithms, CASA-feather and MIRIAD-immerge, with mock observations of star-forming molecular clouds. We demonstrate that the performance of the J-comb algorithm is superior to those of the other two algorithms. We applied the J-comb algorithm to real observational data of the Orion A star-forming region. We successfully produced dust temperature and column density maps with ~10" angular resolution, unveiling much greater details than the previous results.