Desarrollo de un Sistema de Acceso para Parqueaderos y Cobro Tarifario Utilizando Reconocimiento de Patrones.

Abstract

This research is a development of an accessing system for parking and charging through recognizing prompts, It aims to automatize and better the conditions of service of parking, decreasing register time and charging the correct price in the Faculty of Informatics and Electronics at Superior Polytechnic of Chimborazo. Using documental analysis neuronal networks of supervising learning were assessed and the most appropriate neural network of Retro-propagation was chosen for work. The system took into consideration the operative system Linux-Ubuntu by using QT Creator as compiler and for artificial vision the bookstores of OpenCV, programming language was C++. Neural network was coded and samples were chosen for its entertainment. The system of segmentation and classification was designed. A scale model of parking was build including two cameras IP on line with the computer. The results were positive in 99.256% classifying license plates. Processing time was relatively low with a medium value of 156.21ms. It is also taken into account that retro-propagation network has a high development classifying characteristics and management of big data volumes. Entering and leaving parking time have been reduced because of the new system. It is recommended to researches of the Faculty of Informatics and Electronics to consider this work to keep studying neuronal artificial networks.