Graffiti Detection Using Two Views

This paper presents a novel video surveillance approach designed to detect vandal acts occurring on the background of the monitored scene, such as graffiti painting on walls and surfaces, public and private property defacing or etching, unauthorized post sticking. The aim of our approach is to detect this class of events rapidly and robustly. We propose to use two synchronized views to deploy synergically depth and intensity information concerning the monitored scene. Our system can work within unstructured environments and with geometrically unconstrained backgrounds.