Blind and robust reconstruction of adaptive optics point spread functions for asteroid deconvolution and moon detection

Anthony Berdeu,Férréol Soulez,Kate Minker,Benoit Carry,Guillaume Bourdarot,Antoine Kaszczyc,Maud Langlois

DOI: https://doi.org/10.1117/12.3019874

2024-10-11

Abstract:Initially designed to detect and characterize exoplanets, extreme adaptive optics systems (AO) open a new window on the solar system by resolving its small bodies. Nonetheless, despite the always increasing performances of AO systems, the correction is not perfect, degrading their image and producing a bright halo that can hide faint and close moons. Using a reference point spread function (PSF) is not always sufficient due to the random nature of the turbulence. In this work, we present our method to overcome this limitation. It blindly reconstructs the AO-PSF directly in the data of interest, without any prior on the instrument nor the asteroid's shape. This is done by first estimating the PSF core parameters under the assumption of a sharp-edge and flat object, allowing the image of the main body to be deconvolved. Then, the PSF faint extensions are reconstructed with a robust penalization optimization, discarding outliers on-the-fly such as cosmic rays, defective pixels and moons. This allows to properly model and remove the asteroid's halo. Finally, moons can be detected in the residuals, using the reconstructed PSF and the knowledge of the outliers learned with the robust method. We show that our method can be easily applied to different instruments (VLT/SPHERE, Keck/NIRC2), efficiently retrieving the features of AO-PSFs. Compared with state-of-the-art moon enhancement algorithms, moon signal is greatly improved and our robust detection method manages to discriminate faint moons from outliers.

Signal Processing,Instrumentation and Methods for Astrophysics

What problem does this paper attempt to address?

The problems that this paper attempts to solve are: **In the Adaptive Optics (AO) system, due to the imperfect correction of the Point Spread Function (PSF) caused by atmospheric turbulence and other factors, how to directly reconstruct a high - quality AO - PSF from data without prior knowledge, and remove the bright halo caused by PSF spreading, thereby improving the imaging quality of asteroids and the detection ability of faint satellites**.

Specifically, the paper focuses on the following aspects:

1. **PSF Reconstruction Problem**:
- Traditional methods rely on reference PSF or instrument parameters, but these methods are not effective in the face of random turbulence.
- The paper proposes a blind reconstruction method that can directly reconstruct the AO - PSF from observational data without any prior information.

2. **Halo Removal and Deconvolution Problem**:
- The correction of the AO system is not perfect, resulting in a bright halo in the image, which will mask faint satellites close to the main celestial body.
- The paper estimates the core parameters of the PSF and performs deconvolution, and then uses a robust optimization method to reconstruct the faint extended part of the PSF to remove the halo.

3. **Faint Satellite Detection Problem**:
- After removing the halo, the paper proposes a detection method, which uses the reconstructed PSF and the identified outliers (such as cosmic rays, bad pixels, etc.) to detect faint satellites in the residual map.

### Formula Summary

1. **Forward Model**:
\[
d=(o \star p)+n
\]
where \(d\) is the observational data, \(o\) is the extended target, \(p\) is the long - exposure PSF, and \(n\) is the noise term.

2. **Regularized Cost Function**:
\[
(\tilde{o}, \tilde{p}) \leftarrow \arg\min_{o \geq 0, p \geq 0} C(o, p)=D_{wls}(d, o \star p, w)+\mu_{obj} R_{obj}(o)+\mu_{psf} R_{psf}(p)
\]
where:
- \(D_{wls}\) is the weighted least - squares difference:
\[
D_{wls}(\phi_1, \phi_2, w) \triangleq \frac{1}{2} \sum_x w(x)(\phi_1(x)-\phi_2(x))^2
\]
- \(R_{obj}\) is the regularization term of the object, which encourages a smooth and sharp - edged object:
\[
R_{obj}(\phi) \triangleq \sum_x \left[ \sqrt{([\nabla_1 \phi(x)]^2+[\nabla_2 \phi(x)]^2+[\epsilon_{obj}]^2)}-\epsilon_{obj} \right]
\]
- \(R_{psf}\) is the regularization term of the PSF, which is applied to the gradient of the logarithmic PSF:
\[
R_{psf}(\phi) \triangleq \sum_x \left( [\nabla_1 [\ln(\phi(x))]]^2+[\nabla_2 [\ln(\phi(x))]]^2 \right)
\]

3. **Weight Update**:
\[
w(x)=
\begin{cases}
0 & \text{if } w_\rho(x) \leq \bar{w}_\rho \\
\frac{1}{\eta d_{mod}(x)+v_{ron}} & \text{otherwise}
\end{cases}
\]
where \(w_\rho(r)\) is the robust weight:
\[