We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Astrophysics > Instrumentation and Methods for Astrophysics

Title: Image Improvement and Restoration in Optical Time Series

Abstract: Globular clusters (GCs) are considered strong candidates for hosting free-floating planets. However, since they are not bound to a star, they are undetectable by any conventional planetary detection methods: transit, radial velocity, or direct imaging. Gravitational microlensing, which causes transient brightening of background stars by passing foreground masses, is, on the other hand, an established method of detecting planets and proves promising for application in GCs. By applying differential photometry on the time-series images of GCs, we can extract variability events, build light curves, and inspect them for the presence of microlensing events. However, instrumental anomalies and varying observing conditions over a long observational campaign result in the distortion of the stellar PSF, which affects the subtraction quality and leads to false-positive transient detection. We use the Scaled Gradient Projection (SGP) iterative image reconstruction algorithm to restore stellar shapes in our data. We investigate using a more flexible divergence measure, the $\beta$-divergence, as opposed to the commonly used Kullback-Leibler divergence in SGP, adapting parameter $\beta$ to the data. An extensive empirical study comparing SGP performance under these two objective functions is carried out using several physically-motivated metrics such as Full-Width-at-Half-Maximum (FWHM), ellipticity, and radial profiles. Our experiments on star stamps, extracted from real-sky GC field images, suggest that $\beta$-divergence not only improves the restoration quality but also enhances flux conservation compared to the KL divergence. We find that using $\beta$-divergence in SGP is a promising approach in astronomical image restoration.
Comments: Submitted to MNRAS
Subjects: Instrumentation and Methods for Astrophysics (astro-ph.IM)
Cite as: arXiv:2207.10973 [astro-ph.IM]
  (or arXiv:2207.10973v2 [astro-ph.IM] for this version)

Submission history

From: Yash Gondhalekar [view email]
[v1] Fri, 22 Jul 2022 09:39:51 GMT (7473kb,D)
[v2] Mon, 31 Oct 2022 10:45:29 GMT (6709kb,D)

Link back to: arXiv, form interface, contact.