Current Research


Processing of gigantic point clouds

Nowadays, LiDARs are able to generate tens of millions of points from a single acquisition. The main benefit of this considerable change lies in their capacity to capture extremely detailed surfaces. But this comes with a severe complexity in time and in memory, making the reconstruction very challenging. In the context of the PhD of Arnaud Bletterer (2014-2018), we proposed a local graph-based structure to deal with the set of LiDAR acquisitions of a digitization campaign. We showed how this structure is particularly suitable for resampling gigantic points clouds: it reduces the number of points drastically without altering the visual quality of large and complex sites, whatever the number of acquisitions.

One Related Publication:

A. Bletterer, F. Payan, M. Antonini, A. Meftah, Out-of-core Resampling of Gigantic Point Clouds, Symposium on Geometry Processing 2017- Posters, The Eurographics Association, Paris, July 7-11, 2017.


Progressive Decomposition of meshes for Geosciences

I have been collaborating with IFP Energies Nouvelles since 2014. We proposed a novel progressive decomposition for massive structured hexahedral meshes. Our scheme generates a hierarchy of meshes at increasing levels of resolution, while ensuring a geometrical coherency over the resolutions. Our main contribution is a lossless and reversible wavelet filtering that takes into account the geometrical discontinuities in order to preserve them whatever the resolution, but also manages the categorial properties that are generally associated to the hexahedra in geosciences during analysis.

One Related Publication:

J.-L. Peyrot, L. Duval, S. Schneider, F. Payan, M. Antonini, (H)exaschrink: Multiresolution Compression of Large Structured Hexahedral Meshes with Discontinuities in Geosciences, IEEE International Conference in Image Processing (ICIP), September, 2016.


Surface Sampling and Semi-Regular Reconstruction

During Jean-Luc Peyrot's PhD (2011-2014), I worked on the resampling of surface meshes, and then on the semi-regular reconstruction of surfaces acquired by stereoscopic systems. Our final objective was to develop an acquisition system that provides directly semi-regular output, instead of the classical point clouds ( that must be subsequently cleaned, triangulated, and remeshed, if users want a discretized surface with a semi-regular connectivity).
Our two main contributions have been i) an efficient method of blue noise resampling of surface meshes, and a reconstruction method that directly generates a semi-regular mesh from stereoscopic images.

One Related Publication:

J.-L. Peyrot, F. Payan, M. Antonini, From stereoscopic images to semi-regular meshes, Signal Processing: Image Communication, Volume 40, p. 97-110, doi: 10.1016/j.image.2015.11.004, January, 2016.


Semi-Regular Remeshing

I worked on the semi-regular remeshing of surfaces during Aymen Kammoun's PhD (2007-2011) and during a collaboration with Basile Sauvage, (Icube, Strasbourg) and Céline Roudet (Le2i, Dijon).
The semi-regular meshes are based on a regular subdivision connectivity. This subdivision connectivity also allows a compact representation, adapted to multiresolution analysis and wavelet compression. Usually, SR meshes are not provided by current acquisition systems or software. As a consequence, if we want a semi-regular mesh, we have to remesh the data.

One Related Publication:

F. Payan, C. Roudet, B. Sauvage, Semi-regular Triangle Remeshing: a Comprehensive Study, Computer Graphics Forum, Blackwell Publishing, Volume 34, Issue 1, pp.86-102, doi: 10.1111/cgf.12461, February, 2015.