Precision speckle pattern reconstruction for high contrast imaging

TL;DR

This paper presents a method for high contrast imaging that reconstructs speckle patterns using phase retrieval to improve detection sensitivity, achieving a flux ratio of 5×10⁻⁷ at small angular separations.

Contribution

It introduces a novel approach combining physical speckle reconstruction with optimal detection statistics, correcting biases and analyzing sensitivity loss in simulations.

Findings

Achieves a 5σ detection limit of 5×10⁻⁷ flux ratio.

Corrects bias from PSF reconstruction up to 10,000 images.

Demonstrates effectiveness with simulated observations of Sirius A.

Abstract

In High Contrast Imaging, a large instrumental, technological and algorithmic effort is made to reduce residual speckle noise and improve the detection capabilities. In this work, we explore the potential of using a precise physical description of speckle images, in conjunction with the optimal detection statistic to perform High Contrast Imaging. Our method uses short-exposure speckle images, reconstructing the Point Spread Function (PSF) of each image with phase retrieval algorithms. Using the reconstructed PSF's we calculate the optimal detection statistic for all images. We analyze the arising bias due to the use of a reconstructed PSF and correct for it completely up to its accumulation over $10^4$ images. We measure in simulations the method's sensitivity loss due to overfitting in the reconstruction process and get to an estimated 5$\sigma$ detection limit of $5\times 10^{-7}$ flux ratio at angular separations of $0.1 -0.5^{\prime\prime}$ for a $1h$ observation of Sirius A with a 2m-telescope.