Unmixing microinfrared spectroscopic images of cross-sections of historical oil paintings

2026-03-03

AI Summary

This study addresses the spectral mixing problem in attenuated total reflection micro-Fourier transform infrared (ATR-μFTIR) hyperspectral images of historical oil painting cross-sections, which arises from complex multilayer structures, material heterogeneity, and degradation. To tackle this challenge, the authors propose an unsupervised convolutional autoencoder framework for blind source unmixing. The method jointly estimates endmember spectra and their corresponding abundance maps by modeling image patches and introduces a novel weighted spectral angle distance (WSAD) loss function. This loss adaptively assigns band-specific weights based on spatial flatness, neighborhood consistency, and spectral smoothness, thereby enhancing robustness against atmospheric interference and acquisition artifacts. Experiments on cross-section samples from the Ghent Altarpiece demonstrate that the approach effectively disentangles mixed spectra, significantly improving the accuracy, interpretability, and scalability of material identification.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Machine Learning: Unsupervised & Self-Supervised Learning
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Web Mining and Content Analysis: Bridging structured and unstructured data

Abstract

Spectroscopic imaging (SI) has become central to heritage science because it enables non-invasive, spatially resolved characterisation of materials in artefacts. In particular, attenuated total reflection Fourier transform infrared microscopy (ATR-$\mu$FTIR) is widely used to analyse painting cross-sections, where a spectrum is recorded at each pixel to form a hyperspectral image (HSI). Interpreting these data is difficult: spectra are often mixtures of several species in heterogeneous, multi-layered and degraded samples, and current practice still relies heavily on manual comparison with reference libraries. This workflow is slow, subjective and hard to scale. We propose an unsupervised CNN autoencoder for blind unmixing of ATR-$\mu$FTIR HSIs, estimating endmember spectra and their abundance maps while exploiting local spatial structure through patch-based modelling. To reduce sensitivity to atmospheric and acquisition artefacts across $>1500$ bands, we introduce a weighted spectral angle distance (WSAD) loss with automatic band-reliability weights derived from robust measures of spatial flatness, neighbour agreement and spectral roughness. Compared with standard SAD training, WSAD improves interpretability in contamination-prone spectral regions. We demonstrate the method on an ATR-$\mu$FTIR cross-section from the Ghent Altarpiece attributed to the Van Eyck brothers.

Problem

• hyperspectral unmixing
• ATR-μFTIR
• paint cross-sections
• spectral mixing
• heritage science

Innovation

• unsupervised CNN autoencoder
• blind unmixing
• ATR-μFTIR hyperspectral imaging
• weighted spectral angle distance
• patch-based modeling