Synergic vision system for motion detection

Motion detection in surveillance and monitoring systems is enhanced by the synergistic combination of different kinds of cameras and their optimal distribution over the area of interest. We developed an optimization model for a synergistic vision system based on integer lineal programming. The objectives consist in the following: on one hand computing the optimal position and orientation of all directional and omnidirectional cameras, in order to maximize the workspace coverage and on the other hand detecting the objects motion in the workspace. To detect efficiently the movement, even under global luminosity changes, is used a background subtraction algorithm, which uses spatial information of texture. The proposed method is evaluated using a representative set of real sceneries and a network of cameras. Outcomes show that our algorithm is able to determine also the minimal cameras network configuration required to cover a given area.