Um sistema oportunista para detecção de vagas de estacionamento utilizando placas inteligentes acopladas em câmeras de segurança

Abstract

This work aims to present a system for parking spaces detection using opportunistic cameras. The central idea is to use images from cameras that are already installed to identify possible parking spaces. It is important to highlight that legal, privacy and security issues in using cameras for parking detection is out of scope. This work addressed the problem of how to use cameras opportunistically to detect free parking spaces. The idea for the solution is that by means of embedded systems, adapted them to the cameras already installed in the city, we can successfully design the detection of vacancies. The main objective of this work is to present the feasibility of using embedded systems to detect parking spaces, for such, it is proposed three image processing techniques to solve the problem and as main contribution it is presented a distributed architecture for use of cameras opportunistically to obtain the images. The major contributions of this work were the development of three image processing techniques for vehicle detection; a distributed and opportunistic architecture proposal using cameras to detect parking spaces; comparative evaluation of the proposed architecture and two other existing architectures; a routing algorithm for routing requests in the Embedded architecture Initially, it was evaluated three proposed image processing techniques to detect parking spaces, the best technique was the Dilation after Border Detection with a success rate close to 100%, this evaluation was performed to choose the technique used in the rest of this work. After choose the best technique, it was performed an experiment using the proposed architecture and two others that are used for comparison, the result of the experiment demonstrated the feasibility of use the Embedded architecture. To conduct large experiments is a complex task, because of this a simulation was developed using the network simulator Sinalgo to analyze four variables in the system (communication radius, number of auxiliary cameras, amount of simultaneous requests and percentage of failed cameras). The simulation result shows that the factors with the greatest influence on the variation of the average times of the requests are the size of the communication radius and the amount of auxiliary cameras.