My research domain lies between Image Processing and Low Level
Vision. I aim to address the Vision System as a whole, from the
representation of visual information, algorithms and data structures,
to the parallel implementation on an Embedded Vision System. My
objective is to improve the autonomy of Vision Systems from both
energetic
(computational efficiency), and functional (robustness) points of view.
My contributions are divided in the three following areas:
Image Representation and Processing
Here we deal with different representation models of visual data:
geometric, statistic, discrete...and with the associated data
structures and processing algorithms.
Keywords: Discrete topologies and distances, Scale spaces, Feature spaces, Filtering, Segmentation...
Motion analysis
We seek the most efficient algorithms to extract from a video the
most relevant informations, for videosurveillance or
navigation purposes.
Keywords: Motion Detection, Optical Flow, Target Tracking, Obstacle Detection, Motion Characterisation.
Embedded Vision Systems
We look for an optimal exploitation of the parallel architectures, by adapting the vision algorithms to the available processing and data flow capabilities.
