Convolutional neural network exploiting pixel surroundings to reveal hidden features in artwork NIR reflectograms

Anno: 2022

Autori: Karella T.; Blazek J.; Striova J.

Affiliazione autori: Acad Sci Czech Republ, Inst Informat Theory & Automat, Dept Image Proc, Vodarenskou Vez 4, Prague 18000, Czech Republic; CNR, Natl Res Council, Natl Inst Opt INO, Largo Enr Fermi 6, I-50125 Florence, Italy.

Abstract: Near-infrared reflectography (NIR) is a well-established non-invasive and non-contact imaging technique. The NIR methods are able to reveal concealed layers of artwork, such as a painter’s sketch or repainted canvas. The information obtained may be helpful to historians for studying artist technique, attributing an artwork reconstructing faded details. Our research presents the improved method previously devel-oped that reveals the hidden features by removing the information content of the visible spectrum from NIR. Based on convolutional neural networks (CNN), our model estimates the transfer function from vis-ible spectra to NIR, which is nonlinear and specific for painting materials. Its parameters are learnt for particular paintings on the subsamples randomly selected across the canvas, and the model is further utilised to enhance the whole artwork. In addition to the previously developed model, our algorithm exploits each pixel’s surroundings to estimate its NIR response. This leads to more precise results and increased robustness to various noises. We demonstrate higher accuracy than the previous method on the historical paintings mock-ups and higher performance on well-known artworks such as Madonna dei Fusi attributed to Leonardo da Vinci.(c) 2022 Consiglio Nazionale delle Ricerche (CNR). Published by Elsevier Masson SAS. All rights reserved.

Giornale/Rivista: JOURNAL OF CULTURAL HERITAGE

Volume: 58      Da Pagina: 186  A: 198

Maggiori informazioni: We are grateful to the National Institute of Optics (INO-CNR) for sharing this rare dataset for the presented analysis. This research was supported by Strategy AV21 of the Czech Academy of Sciences Hopes and Risks of the Digital Era and GACR GA21-03921S.
Parole chiavi: Signal separation; Concealed features visualization; Artwork analysis; Infrared reflectography; Convolutional neural networks
DOI: 10.1016/j.culher.2022.09.022

Citazioni: 2
dati da “WEB OF SCIENCE” (of Thomson Reuters) aggiornati al: 2025-03-16
Riferimenti tratti da Isi Web of Knowledge: (solo abbonati)
Link per visualizzare la scheda su IsiWeb: Clicca qui
Link per visualizzare la citazioni su IsiWeb: Clicca qui