Analog-Resistive Networks for Motion Detection
Résumé
In this paper we present two analog networks that implement an algorithm for motion detection in image sequences. The time consuming stochastic or suboptimal deterministic relaxation algorithms used by the initial Markov Random Field based model are replaced by the energy minimization of the analog network to its state of minimal power dissipation. We describe the elementary cell derived from the model equation and the network architectures. Possible application are motion remote control and motion coding at video rate.