Reconstruction of partially sampled STEM-EELS images with atomic resolution

Etienne Monier (1) , Thomas Oberlin (1, 2) , Nathalie Brun (3) , Nicolas Dobigeon (1, 4, 5)

1 IRIT-SC - Signal et Communications
2 ISAE-SUPAERO - Institut Supérieur de l'Aéronautique et de l'Espace
3 LPS - Laboratoire de Physique des Solides
4 IUF - Institut universitaire de France
5 Toulouse INP - Institut National Polytechnique (Toulouse)

Electron microscopy has shown to be a powerful tool to ana-lyze chemical composition of samples. However, acquiring a high qualityimage is hard due to radiation damages which limit the signal-to-noiseratio. One solution, considered in this work, consists in spatially partiallyacquiring the multi-band image and reconstructing it afterwards. Wepropose a reconstruction algorithm, referred to as Fourier sparsity in3D (FS3D), based on a regularization specifically tailored for atomicallyresolved images. Experiments show that the proposed FS3D method leadsto state-of-the-art results with a significantly lighter computational cost.

Domaines: Intelligence artificielle [cs.AI]; Vision par ordinateur et reconnaissance de formes [cs.CV]; Analyse de données, Statistiques et Probabilités [physics.data-an]; Apprentissage [cs.LG]; Traitement des images [eess.IV]; Traitement du signal et de l'image [eess.SP]

Etienne Monier, Thomas Oberlin, Nathalie Brun, Nicolas Dobigeon. Reconstruction of partially sampled STEM-EELS images with atomic resolution. Workshop on Signal Processing with Adaptative Sparse Structured Representations (SPARS 2019), Nicolas Dobigeon, Toulouse INP, France; Cédric Févotte, CNRS, France, Jul 2019, Toulouse, France. ⟨hal-03108208⟩