I am a PhD student in Mathematics and image processing under the joint supervision of Bruno Galerne, maître de conférences (assistant professor) at the MAP5 laboratory, Université Paris Descartes, and Agnès Desolneux, CNRS Senior Research Scientist at the CMLA laboratory, ENS Paris Saclay.
I'm working on the application of Determinantal Point Processes (DPP) to the images framework, in particular to image synthesis.
Determinantal point processes are random sets of points that model repulsion between the points : they assign a higher occurrence probability to "repulsive" sets according to a specific similarity measure (a kernel). It is possible to entirely characterise the repulsion form of the process from this kernel. They can be sampled efficiently and exactly.
We study DPP in the 2D discrete framework of images, where the points are pixels. We want to determine if the use of these processes is consistent to synthesize images.
UFR Mathématiques-Informatique, Université Paris Descartes
45 Rue des Saints-Pères
75006 Paris, France
claire.launay at parisdescartes.fr