Percepción Visual

From Robotica

Tridimensional Tracking

right|300px We are working on a particle filter to localise and track objects in 3D. We have adapted the CONDESATION algorithm, that is a Sequential Monte Carlo technique, to track objects in 3D without doing any triangulation. No matching is needed, as the particle filter fuse information from different cameras and performs an implicit matching. The algorithm scales to more than two cameras and does not filter the entire image, using active vision principles. Several test have been conducted using two, three and four cameras covering spaces from 8 to 120 cubic metres.

Surveillance Applications

left|300px Non intrusive surveillance application needs to localise in 3D any moving object, in order to detect its proximity to "protected" areas. Our applications uses four Firewire cameras and two different techniques for 3D estimation: a particle filter and a fly algorithm, both based on colour and motion image information.