Blind Source Separation Approaches for Exoplanet Signal Extraction

Post-processing is a key step in the extraction of exoplanet signals from direct imaging data. Multiple currently employed techniques (including all variations on principal component analysis and singular value decomposition) belong to a general class of algorithms called Blind Source Separation (BSS). Here we demonstrate two other BSS algorithms that have previously not been tested on exoplanet imaging data: independent component analysis and common spatial pattern filtering. We utilize synthetic data with known signals to evaluate the performance of these techniques and discuss the relative strengths of each approach as a function of the correlation and relative magnitude of various noise sources commonly found in exoplanet imaging data.