In a first part, I will illustrate some of my work in computer vision and image processing. Those are dealing with multi-view geometry, the use of geometric reasoning to integrate scene constraints, image registration and segmentation. Without entering into all the details, I will present the principles that characterize those approaches that are now a few years old and I will propose some possible extensions around that work. A second part will address some problems related to electro- and magneto- encephalography (M/EEG), which is my current field of research. I'll focus more particularly on an algorithm that we have proposed for the detection of events of interest for multi-trial data and on some techniques that have been developed for the modeling of the forward M/EEG problem. As in the first part, I will try to propose some some possible developments around this theme.
