Partial-Attribution Instance Segmentation for Astronomical Source Detection and Deblending
This addresses the challenge of accurately separating individual stars or galaxies in astronomical images, which is crucial for astronomers but remains an open problem due to data complexities.
The paper tackles the problem of astronomical source detection and deblending in images with overlapping sources, high dynamic range, and low signal-to-noise, by introducing Partial-Attribution Instance Segmentation, a new approach that makes the process tractable for deep learning models, with a novel neural network implementation provided as a demonstration.
Astronomical source deblending is the process of separating the contribution of individual stars or galaxies (sources) to an image comprised of multiple, possibly overlapping sources. Astronomical sources display a wide range of sizes and brightnesses and may show substantial overlap in images. Astronomical imaging data can further challenge off-the-shelf computer vision algorithms owing to its high dynamic range, low signal-to-noise ratio, and unconventional image format. These challenges make source deblending an open area of astronomical research, and in this work, we introduce a new approach called Partial-Attribution Instance Segmentation that enables source detection and deblending in a manner tractable for deep learning models. We provide a novel neural network implementation as a demonstration of the method.