Coordination : Cerasela Calugaru et Annamaria Kiss

One of the main issues of quantitative imaging is the correct cellular level segmentation of 3D images of plant tissues, issued from confocal microscopy. The basic segmentation methods we use, are watershed combined with the level set method. The goal of the internship is to optimise and give an efficient implementation of the level set method.

Therefore, the student is invited to

• study the used image analysis methods,
• review the optimization and parallelizing techniques, applicable on the used level set segmentation
• implement the method in C++, using the optimal parallelizing technique
• deliver the implementation as a plugin to existing computing platforms

Atomic-level computer simulation using molecular dynamics require high-performance computing infrastructures, such as the one of the PSMN/ENS de Lyon. We are thus very grateful for the CPU time provided by the PSMN. In recent years the use of GPUs as computing units have become a very attractive alternative to CPUs. The CBP provides us access to several machines with GPU-acceleration.

• Etude du code d’origine (écrit en Python), et une évaluation préliminaire du coût de la réécriture du code en C++, de son optimisation et parallélisation
• Expertise au choix des méthodes numériques, programmation dans le cadre du projet