Performance evaluation of a video surveillance system using stochastic petri nets for license plate detection on highways

Traffic crimes are a recurring problem in several countries around the world. Crimes involving car, motorcycle, and other vehicle theft occasionally appear in newspaper headlines and news programs. Currently, it is possible to use monitoring cameras to detect and track illegal vehicles. However, implementing a system of this nature in the real world requires abundant physical and digital resources. This paper proposes a performance model in stochastic Petri nets to evaluate a video surveillance system by detecting license plates on highways. Using stochastic models, such as Petri nets, enables a quick planning process with low or no cost. The use of Petri nets, in particular, generates greater accuracy between the model and reality, as it allows complex behaviors, such as competition and parallelism, to be captured. Results show a tendency to have a greater impact on metrics when the capacity of the system’s two initial modules and the number of cameras used are varied. The general results presented in this paper allow us to understand how each parameter influences the system’s response time individually and collectively. The proposed case studies are a practical guide for system administrators to use the model.