Wavelet based speckle suppression for exoplanet imaging - Application of a de-noising technique in the time domain

Context. High-contrast exoplanet imaging is a rapidly growing field as can be seen through the significant resources invested. In fact, the detection and characterization of exoplanets through direct imaging is featured at all major ground-based observatories. Aims. We aim to improve the signal-to-noise ratio (SNR) achievable for ground-based, adaptive-optics assisted exoplanet imaging by applying sophisticated post-processing algorithms. In particular, we investigate the benefits of including time domain information. Methods. We introduce a new speckle-suppression technique in data post-processing based on wavelet transformation. This technique explicitly considers the time domain in a given data set (specifically the frequencies of speckle variations and their time dependence) and allows us to filter-out speckle noise. We combine our wavelet-based algorithm with state-of-the-art principal component analysis (PCA) based PSF subtraction routines and apply it to archival data sets of known directly imaged exoplanets. The data sets were obtained in the L filter where the short integration times allow for a sufficiently high temporal sampling of the speckle variations. Results. We demonstrate that improvements in the peak SNR of up to forty to sixty percent can be achieved. We also show that, when combined with wavelet-denoising, the PCA PSF model requires systematically smaller numbers of components for the fit to achieve the highest SNR. The improvement potential is, however, data set dependent or, more specifically, closely linked to the field rotation available in a given data set: larger amounts of rotation allow for a better suppression of the speckle noise. Conclusions. We have demonstrated that by applying advanced data post-processing techniques, the contrast performance in archival high-contrast imaging data sets can be improved.