Machine Learning for High-Contrast Imaging

ESO is carrying out R&D to prepare for the ELT Planetary Camera and Spectrograph (PCS) which is an instrument dedicated to the detection and characterization of low-mass Exoplanets around nearby stars. As part of this work, a laboratory setup called GHOST has been developed which mimics the eXtreme Adaptive Optics (XAO) system and provides an imaging channel.

Optimization and Post-processing of XAO-corrected images are important aspects of high-contrast imaging. The student will review existing algorithms and explore avenues to improve those using data generated by GHOST. Machine learning methods could help to improve both, the optimization of the instrumental point spread function which is a non-linear function of the instrument’s internal optical aberrations, and its post-observational calibration.

The student should be knowledgeable in Python or similar data analysis tools (e.g. Matlab) and have a basic understanding of optical image formation and machine learning.


Supervisor: Dr. Markus Kasper