{"id":1523,"date":"2024-10-10T15:18:47","date_gmt":"2024-10-10T13:18:47","guid":{"rendered":"https:\/\/wordpress-test.app.u-pariscite.fr\/diip\/?p=1523"},"modified":"2024-10-25T11:02:57","modified_gmt":"2024-10-25T09:02:57","slug":"investigating-diffusion-models-for-astronomical-image-deconvolution-boosting-the-synergy-between-euclid-and-lsst","status":"publish","type":"post","link":"https:\/\/wordpress-test.app.u-pariscite.fr\/diip\/investigating-diffusion-models-for-astronomical-image-deconvolution-boosting-the-synergy-between-euclid-and-lsst\/","title":{"rendered":"Investigating Diffusion Models for Astronomical Image Deconvolution – boosting the synergy between Euclid and LSST"},"content":{"rendered":"

[et_pb_section fb_built=”1″ _builder_version=”3.22″][et_pb_row _builder_version=”3.22″ background_size=”initial” background_position=”top_left” background_repeat=”repeat”][et_pb_column type=”1_4″ _builder_version=”3.0.47″][et_pb_text _builder_version=”3.22.1″ background_color=”#072c72″ border_color_all=”#3255c9″ text_orientation=”right” background_layout=”dark” custom_padding=”20px|15px|15px|” z_index_tablet=”500″]<\/p>\n

2024 <\/em><\/p>\n

Master’s Projects<\/p>\n

@Physics\/Astronomy<\/span><\/p>\n

[\/et_pb_text][et_pb_text _builder_version=”3.22.1″ text_orientation=”right” z_index_tablet=”500″]<\/p>\n

#diffusion models<\/p>\n

#image processing<\/p>\n

#deconvolution<\/p>\n

 <\/p>\n

\ud83c\udf89This project has been accepted at the ML4Physics workshop at NeurIPS!<\/p>\n

[\/et_pb_text][\/et_pb_column][et_pb_column type=”3_4″ _builder_version=”3.0.47″][et_pb_text _builder_version=”3.22.1″ z_index_tablet=”500″]<\/p>\n

Project Summary<\/strong><\/h3>\n

This project tackles the problem of deconvolving astronomical images to uncover the intrinsic properties of celestial objects, especially in ground-based observations. It explores the use of diffusion models (DMs) and the Diffusion Posterior Sampling (DPS) algorithm to address this challenge. Score-based DMs, trained on high-resolution cosmological simulations, are used in a Bayesian framework to compute posterior distributions based on observations. By incorporating redshift and pixel scale, the method adapts to various datasets. Tests on Hyper Supreme Camera (HSC) data achieve resolutions comparable to Hubble Space Telescope (HST) images, while also quantifying uncertainties and identifying prior-driven features for scientific use.<\/p>\n

 <\/p>\n

Alexandre Boucaud
<\/strong>aboucaud@apc.in2p3.fr<\/h3>\n