Bastien Laville

About Me

I got my Ph.D. in calculus of variations applied to biomedical imaging, under the supervision of Laure Blanc-Féraud and Gilles Aubert in the Morpheme team. I am now working on spatial observation research, and really eager to pursue collaborations in various fields, especially related to inverse problem, optics, and computational physics as a whole. Feel free to contact me!

PhD Thesis

My thesis focused on off-the-grid methods, leveraging variational principle. It aims to recover curve structures located on a continuum from low-frequency/diffraction-limited measurements; with practical applications in optical science, e.g. biomedical imaging with fluorescence microscopy or ultrasound imaging. See the manuscript and slides.

In the following, see the reconstruction performed by an off-the-grid based method (middle) on an experimental SMLM acquistion stack, v. the mean (left) an other state-of-the-art method, SRRF.

An illustration of the off-the-grid method

Research Interests

Variational principle: calculus of variation, geometric measure theory.
Optics: imager w/ mirrors, leveraging segmented mirror PSF.
Deep Learning: computer vision, Physics Informed NN, PyTorch, JAX.
Differential geometry: Riemannian-based regularisation, Riemannian optimisation .

Publications

Preprints and collaborations

Dynamic off-the-grid reconstruction of curves in the space of position-orientation: theory and algorithms for paths untangling.
With Mr. Théo Bertrand (CEREMADE).
Preprint. To appear, 2024.
[PDF] [Code]

Collaboration with Mr. Cédric Dubois (LAM), we try to leverage continuous methods for stars/clumps localisation throughout different protocols.

Journals

A $\Gamma$-convergence result and an off-the-grid charge algorithm for curve reconstruction in inverse problems.
Bastien Laville, Laure Blanc-Féraud and Gilles Aubert.
Journal of Mathematical Imaging and Vision (JMIV). 2024.
[PDF]

Off-the-grid curve reconstruction through divergence regularisation: an extreme point result.
Bastien Laville, Laure Blanc-Féraud and Gilles Aubert.
SIAM Journal on Imaging Sciences (SIIMS), Volume 16, Issue 2, pp. 867-885. June 2023.
[PDF] [DOI] [Ref]

Off-the-grid variational sparse spike recovery: methods and algorithms.
Bastien Laville, Laure Blanc-Féraud and Gilles Aubert.
Journal of Imaging, Volume 7, Issue 12. December 2021. Review.
[PDF] [DOI] [Code] [TeX]

Chapters and Proceedings

Off-the-grid charge algorithm for curve reconstruction in inverse problems.
Bastien Laville, Laure Blanc-Féraud and Gilles Aubert.
Springer LNCS 14009, 9th International Conference on Scale Space and Variational Methods in Computer Vision. May 2023.
[PDF] [DOI] [Slides] [Code]

Off-the-grid covariance-based super-resolution microscopy.
Bastien Laville, Laure Blanc-Féraud and Gilles Aubert.
International Conference on Acoustics, Speech, & Signal Processing (IEEE ICASSP). May 2022.
[PDF] [DOI] [Slides] [TeX slides] [Code]

National Conferences

Reconstruction de courbes en super-résolution sans-grille.
Bastien Laville, Laure Blanc-Féraud and Gilles Aubert.
GRETSI 2023. September 2023.
[PDF] [Slides] [TeX]

Super-résolution sans-grille pour l'imagerie de fluctuation.
Bastien Laville, Laure Blanc-Féraud and Gilles Aubert.
GRETSI 2022. September 2022.
[PDF] [Slides] [Code] [TeX]

Invited talks

Some developments on off-the-grid curve reconstruction. February 2024, invited speaker at the Institut Mathématiques de Bordeaux, France. [Slides] [Description]
Gridless curve reconstruction: divergence regularisation and untangling by (sub)Riemannian metric. November 2023, invited speaker at the Institut Henri Poincaré, Paris V, France. Workshop closing the ANR CIPRESSI. [Abstract] [Slides]
An extreme point result for off-the-grid curve reconstruction. September 2022, invited speaker at the 11th Applied Inverse Problems Conference in Göttingen, Germany. [Slides]
Off‑the‑grid curve reconstruction: bridge the gap between 0 and 2‑rectifiable measures. March 2023, MIA, Bordeaux, France. [Slides] [TeX]
Prix Pierre Laffitte : super-résolution sans-grille et retombées pour l'industrie. October 2022, Mines Paris-PSL, finalist in contention for the Pierre Laffitte award (Biot, France). [Slides] [TeX]
New off-the-grid approaches. June 2022, invited speaker at the total variation symposium of applieds mathematics conference CANUM (Évian-les-Bains, France). [Slides] [Description]
Off-the-grid static and dynamic super-resolution. December 2021, ANR MicroBlind meeting, Centre International de Rencontres Mathématiques (Marseille, France). [Slides] [TeX]

Services

Automatic - 120 hours (Prépa Polytech).
Admission panel (DUT/M1) - 9 hours.
Mathematics introductory course - 31 hours.
Fourier analysis - 25 hours (L3).

2024 (C) - Inria