Using Data Imputation for Signal Separation in Highcontrast Imaging
Abstract
To characterize circumstellar systems in highcontrast imaging, the fundamental step is to construct a best pointspread function (PSF) template for the noncircumstellar signals (I.e., starlight and speckles) and separate it from the observation. With existing PSF construction methods, the circumstellar signals (e.g., planets, circumstellar disks) are unavoidably altered by overfitting and/or selfsubtraction, making forward modeling a necessity to recover these signals. We present a forward modelingfree solution to these problems with data imputation using sequential nonnegative matrix factorization (DIsNMF), which first converts this signal separation problem to a "missing data" problem in statistics by flagging the regions that host circumstellar signals as missing data, then attributes PSF signals to these regions. We mathematically prove it to have negligible alteration to circumstellar signals when the imputation region is relatively small, which thus enables precise measurement of these circumstellar objects. We apply it to simulated pointsource and circumstellar disk observations to demonstrate its proper recovery of them. We apply it to Gemini Planet Imager K1band observations of the debris disk surrounding HR 4796A, finding a tentative trend that the dust is more forward scattering as the wavelength increases. We expect DIsNMF to be applicable to other general scenarios where the separation of signals is needed.
 Publication:

The Astrophysical Journal
 Pub Date:
 April 2020
 DOI:
 10.3847/15384357/ab7024
 arXiv:
 arXiv:2001.00563
 Bibcode:
 2020ApJ...892...74R
 Keywords:

 1858;
 363;
 1300;
 313;
 1943;
 1913;
 Astrophysics  Instrumentation and Methods for Astrophysics;
 Astrophysics  Earth and Planetary Astrophysics;
 Astrophysics  Solar and Stellar Astrophysics;
 Statistics  Machine Learning
 EPrint:
 18 pages, 9 figures, ApJ published. Modified AASTeX template at https://github.com/seawander/aastex_pwned