arrow drop search cross

I3S: Luca Calatroni, allows UCA to officially join the H2020 RISE NoMADS project!

Upon his arrival at the I3S laboratory, Luca Calatroni, who received funding from the "Complex Systems" academy for his research activities, joined the European project H2020 RISE NoMADS. This project aims to support the creation of a multidisciplinary network of researchers in mathematical analysis, optimization, biomedical imaging and big data analysis. The aim is to fill the current gaps between the theory and the applications of "non-local" methods. This project has received global funding of more than € 1 million from the European Union’s Horizon 2020 research and innovation program.

Publication : 25/06/2020
Partager cet article :
From October 2019, Luca Calatroni joined the Morphème team in the Computer Science, Signals and Systems Laboratory (I3S) as CNRS Researcher. Morphème is a joint project of INRIA, CNRS and Université Côte d´Azur.
This multidisciplinary team specializes in image processing at the interface between computer science and biology.
As a newcomer to UCA, Luca Calatroni has benefited from the Post Office Environment Device (DEP) from the "Complex Systems" academy. This campaign aimed to support recruitment actions in laboratories related to the themes of the academy on permanent positions.
Before being a permanent researcher at the I3S laboratory, Luca Calatroni obtained two post-doctoral grants:
the first funded by the European FP7 Nano2fun Marie-Curie project in 2015, and the second funded by the Jacques Hadamard Foundation in 2017 for the post-doc of excellence Reader Hadamard at CMAP (École Polytechnique).

He briefly tells us about his journey:
"I trained in applied mathematics at the University of Pavia in Italy. In 2015, I did my thesis on mathematical image processing at the Cambridge Center for Analysis of the Department of Applied Mathematics and Theoretical Physics (DAMTP) under the supervision of Carola-Bibiane Schönlieb ".
"In the following years, my work focused on the application of mathematical models for image processing and vision for their application in different disciplines. For example, I recently took care of restoring digital heritage images. The goal was to digitally reconstruct altered images from damaged illuminated original paintings or manuscripts. "

A digital image is made up of elementary units, called pixels, which each represent a portion of the image. An image is thus defined by the number of pixels which compose it in width and in height (which can vary almost to infinity), and by the extent of shades of gray or colors that can take each pixel (we speak of image dynamics).

“The restoration of these digital images is based on fairly recent calculation methods called“ non-local ”. The idea behind these methods is to relate each pixel in an image to all the other pixels in the image. The particularity of this calculation tool is to link all the pixels and not only those which are close to each other. The goal is to come to understand the analogies in terms of common characteristics between the pixel considered and all the others. This process thus makes it possible to process images on which information areas (pixels) are missing using all the information available. If we take the example of the image of a fresco damaged by the effects of time, in certain places colors and shapes are no longer visible. This mathematical method can recover damaged data using similar information found elsewhere in the image. The digital restoration of a painting is therefore very useful for the work of the restorer who, often, cannot perform his physical restoration due to the fragility of the materials used. "

Freshly arrived at UCA, Luca Calatroni decides to associate I3S with the H2020 RISE NoMADS project which had already started in March 2018.

“When I arrived at UCA, I immediately contacted the three leaders of the NoMADS projects: Martin Burger (FAU, Erlangen), Carola-Bibiane Schönlieb (University of Cambridge), and Daniel Tenbrinck (FAU Erlangen) relevance of I3S research activities to the objectives of the NoMADS project. "

NoMADS aims to build an international and multidisciplinary network of universities and companies to fill the current gaps between the theory and the applications of non-local methods.

“Our research project is made up of 16 universities and 7 industrial partners. Our consortium brings together a solid international group of researchers who are experts in mathematics (applied analysis, IT, statistics and optimization), computer vision and data mining. "

The goal of this consortium is to improve the understanding and applicability of non-local methods in a data-flooded world.

With the democratization of digital tools (Internet, mobile telephony, medical imaging.) The amount of data transmitted over communication networks is constantly increasing. Here, the challenge is to succeed in judiciously exploiting these large sets of information for their effective use in several disciplines.

In the context of biomedical imaging, the use of databases capable of codifying a large variability of specific elements (molecules, tissues, organs, etc.) in a compact manner is envisaged. The analysis will thus be optimized, from a quantitative point of view.

"The goal is to use the high availability of data and advanced tools in imaging, signal processing, and optimization to find a way to leverage and if possible, reduce information redundancy. "

Within UCA, the data used will come from images from biology and medicine, which offer a rich field of experience concerning the processing of structured data and have a direct impact on society.

“NoMADS combines two approaches. The first is a finalized theoretical approach to the study of non-local models and their theoretical properties. The second is a finalized digital approach to developing algorithms for their application to big data. As with the restoration of digital images, the aim is to develop calculation methods capable not only of integrating all the data but also of filtering them by similarities even if they are not necessarily close to each other "

"Thereafter, a more concrete approach will be considered by implementing algorithms in products, such as high-quality software for a wide variety of real-world applications. Especially for automatic tools allowing to analyze and classify images in several disciplines. More specifically, for the images concerning the analysis of the biomedical data involved in the project. "

  • More information :