Automatic Number Plate Recognition (ANPR) System using Machine Learning Techniques

Automatic Number Plate Recognition (ANPR) systems are widely used on a wide range of applications nowadays. The proposed approach has been developed in order to recognise UK number plates from high resolution digital images making use of the latest Computer Vision techniques and Machine Learning met...

Descripción completa

Detalles Bibliográficos
Autor: Fernández Sánchez, Leticia
Tipo de recurso: tesis de maestría
Fecha de publicación:2018
País:España
Institución:Universidad del País Vasco
Repositorio:Addi. Archivo Digital para la Docencia y la Investigación
OAI Identifier:oai:addi.ehu.eus:10810/29790
Acceso en línea:http://hdl.handle.net/10810/29790
Access Level:acceso abierto
Palabra clave:automatic number plate recognition
machine learning
artificial intelligence
computer vision
image processing
support vector machine
K-nearest neighbour
optical character recognition
Descripción
Sumario:Automatic Number Plate Recognition (ANPR) systems are widely used on a wide range of applications nowadays. The proposed approach has been developed in order to recognise UK number plates from high resolution digital images making use of the latest Computer Vision techniques and Machine Learning methods. For this purpose, a comparison among the different existing Computer Vision techniques used in ANPR is carried out and a deep insight on the operation and mode of use of the most commonly used Machine Learning algorithms in ANPR is provided, being these: Support Vector Machines, Artificial Neural Networks and K-Nearest Neighbours. Besides, for the development of an efficient, fast and reliable ANPR application, a huge car images dataset is created from scratch, necessary both for training the Machine Learning algorithms and for evaluating the performance of the developed system. The global results obtained, which are equal to or above 90% of success with a response time of less than 3 seconds, prove that the system is able to compete with other recently developed ANPR systems of similar characteristics.