We present a Bayesian framework for learning higherorder transition models in video surveillance networks. Such higher-order models describe object movement between cameras in the...
We propose a novel method for identifying road vehicles between two non-overlapping cameras. The problem is formulated as a same-different classification problem: probability of t...
Abstract. Color constancy is the ability to measure image features independent of the color of the scene illuminant and is an important topic in color and computer vision. As many ...
Triangulation consists in finding a 3D point reprojecting the best as possible onto corresponding image points. It is classical to minimize the reprojection error, which, in the p...
An original concept for computing instantaneous 3D pose and 3D velocity of fast moving objects using a single view is proposed, implemented and validated. It takes advantage of the...
Omar Ait-Aider, Nicolas Andreff, Jean-Marc Lavest,...