A Machine-Vision System to Detect Unusual Activities Online at Vehicular Intersections
Sandra Luz Canchola Magdaleno, Joaquín Salas Rodríguez, Hugo Jiménez‐Hernández, José‐Joel González‐Barbosa, Juan B. Hurtado‐Ramos · LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas) · 2009
In this article, we present a real-time machine-vision system to detect vehicles running on red light or performing forbidden turns at crossroads. The system operates during daytime by receiving video streams from two different sources. One of them is a camera viewing the crossroads to detect unusual activity, while a second camera watches the semaphore to keep synchrony with the traffic controller. The system performance and reliability have been tested on a real vehicular intersection during extended periods of time.