MAYONNAISE: a morphological components analysis pipeline for circumstellar disks and exoplanets imaging in the near infrared

TL;DR

This paper introduces MAYONNAISE, a pipeline for imaging circumstellar disks and exoplanets in the near-infrared that improves detection sensitivity, preserves disk morphology, and separates planetary signals from disks.

Contribution

The pipeline jointly estimates starlight residuals, disks, and planets in high-contrast images, overcoming limitations of current methods in sensitivity and morphological distortion.

Findings

Detects disks at contrasts above 3×10⁻⁶

Preserves diverse disk morphologies including face-on disks

Effectively separates planets from disks even when close together

Abstract

Imaging circumstellar disks in the near-infrared provides unprecedented information about the formation and evolution of planetary systems. However, current post-processing techniques for high-contrast imaging using ground-based telescopes have a limited sensitivity to extended signals and their morphology is often plagued with strong morphological distortions. Moreover, it is challenging to disentangle planetary signals from the disk when the two components are close or intertwined. We propose a pipeline that is capable of detecting a wide variety of disks and preserving their shapes and flux distributions. By construction, our approach separates planets from disks. After analyzing the distortions induced by the current angular differential imaging (ADI) post-processing techniques, we establish a direct model of the different components constituting a temporal sequence of high-contrast images. In an inverse problem framework, we jointly estimate the starlight residuals and the potential extended sources and point sources hidden in the images, using low-complexity priors for each signal. To verify and estimate the performance of our approach, we tested it on VLT/SPHERE-IRDIS data, in which we injected synthetic disks and planets. We also applied our approach on observations containing real disks. Our technique makes it possible to detect disks from ADI datasets of a contrast above $3\times10^{-6}$ with respect to the host star. As no specific shape of the disks is assumed, we are capable of extracting a wide diversity of disks, including face-on disks. The intensity distribution of the detected disk is accurately preserved and point sources are distinguished, even close to the disk.