IoT in Services for Communities in the Recognition of Plates with

OpenCV-Python

Liliana Enciso

1,2

, Roddy Yaguana

, Renzo Jaramillo

and Elmer Zelaya-Policarpo

Grupo Investigaci

on IoT - Network Communications on Smart Environment, Universidad T

ecnica Particular de Loja,

San Cayetano Alto, Loja, Ecuador

Departamento de Ciencias de la Computaci

on y Electr

onica, Universidad T

ecnica Particular de Loja,

San Cayetano Alto, Loja, Ecuador

Keywords:

Artiﬁcial Neural Networks, IoT, Artiﬁcial Vision, OCR, ANPR, Search for Patterns.

Abstract:

The evolution of the technology day by day advances looking for greater comfort for the users, facilitating

the daily living, allowing to reduce times, costs and resources, the advance of the technology allows greater

learning. The application was designed with the objective of controlling vehicular access to the UTPL by

recognizing the plaque, only the teachers and staff that have access can enter, in this way it will prevent people

outside the university from entering. Periodic control of all income was taken to the university and that way a

statistic can be drawn from the people who really belong to the university.

1 INTRODUCTION

The ﬁeld of image processing is continuously evol-

ving. During recent years there has been a signiﬁ-

cant increase in interest in ﬁelds such as image mor-

phology, artiﬁcial neural networks, color and / or

grayscale image processing, image data compression,

image recognition and analysis systems. knowledge-

based images. Recognition is a basic attribute of the

human being, as well as of other living organisms.

This recognition is usually generated by means of pat-

terns (Peralta, 2009). With the advancement of artiﬁ-

cial vision we could acquire, process, analyze and un-

derstand the images of the real world in order to pro-

duce numerical or symbolic information to be treated

by the computer, just as humans do in order to under-

stand the world we live in surrounds. We will des-

cribe the process of extraction of characteristic pat-

terns of images, through the help of Artiﬁcial Neu-

ral Networks, for the treatment of the images libraries

will be used OpenCV (P. Garc

ıa., 2013).

Computer systems that use computer vision make

use of digital image processing techniques in order

to emulate the vision system and algorithm (Enciso-

Quispe et al., 2018). The research work developed has

its origin in the need to develop a system that is capa-

ble of identifying images quickly and without exces-

sive consumption of resources. Of all this the giants

of the computer science (I. Ortiz, 2016) have realized

and they have launched applications of recognition of

images for mobile devices developing algorithms ba-

sed on the content of the images recognizing forms

and analyzing their resemblance. The idea of this pro-

ject is born from the need to see an object that we do

not recognize and we have not been able to look for

it because we do not know under what name to look

for it, or we have gone down the street and we have

seen a monument and we did not know what it was, or

we have been in a museum and we would have liked

to have more information instantly about the work we

contemplate.

2 RELATED WORKS

Computer vision aims to solve several of the problems

that arise in the daily life of human beings. Every-

thing that humans observe is through our vision, that

is why we have investigated several projects to cap-

ture ideas of how computer vision works.

The recognition of images (P. Garc

ıa., 2013) using

artiﬁcial neural networks describes the process of ex-

tracting characteristic patterns of images, with the

help of Artiﬁcial Neural Networks. The information

of the Neural Network together with additional data

of the images, will be stored in a database and con-

sumed by a web service. A mobile phone with an

Android operating system will consume the informa-

430

Enciso, L., Yaguana, R., Jaramillo, R. and Zelaya-Policarpo, E.

IoT in Services for Communities in the Recognition of Plates with OpenCV-Python.

DOI: 10.5220/0007235004300437

In Proceedings of the 14th International Conference on Web Information Systems and Technologies (WEBIST 2018), pages 430-437

ISBN: 978-989-758-324-7

tion stored in the web service. After making an image

capture with the phone camera, it processes the image

and together with the data consumed by the web ser-

vice will be able to identify which image it is. For

the treatment of the images, OpenCV libraries will be

used, both on the server and on the mobile phone.

The recognition of objects immersed in static ima-

ges using the algorithm HOG and RNA-MLP is a

research project approved by the Technological Uni-

versity of the Mixteca. It is based on presenting an

object recognition system that it uses in the process

of extracting characteristics from the HOG algorithm

and an MLP type RNA in the classiﬁcation process

(Garc

ıa., 2015).

The practical application of artiﬁcial vision for the re-

cognition of faces in an image, using neural networks

and object recognition algorithms of the Opencv li-

brary is a project that focuses on studying the different

methods and algorithms that allow detecting a face in

an image , and develop an application that can be trai-

ned to perform the recognition of a speciﬁc person

within the image (E. Caballero., 2017).

The application of neural networks (Lanzarini and

Giusti., 2014) to the recognition of patterns in medical

images is a project that presents the research in image

analysis and recognition of patterns as a help to medi-

cal diagnosis is widely recognized for its complexity

and importance, in this sense the neural networks re-

sult from It is extremely useful, since not only are they

capable of learning with the help of the expert but they

can themselves generalize the information contained

in the input data, showing relationships that are a pri-

ori complex.

The development (Paucar and Tasiguano, 2016) of

the algorithm for the recognition of vehicle license

plates consists of a digital image algorithm. For this, a

vehicle license plate recognition system has been de-

veloped using the LabView program and its artiﬁcial

vision toolkit and a high resolution IP camera. The

designed software locates the vehicle plate inside the

captured image of the vehicle and recognizes the cha-

racters and numbers that are inside it. The location

of the vehicle plate in the image is made by looking

for patterns. For the recognition of the characters and

numbers contained in the plate, optical character re-

cognition, OCR is used. Once the text on the plate

has been read, the characters and numbers are collated

with thoseof a database that contains all the license

plates of the vehicles that are authorized to enter the

parking lot.

Plaque Recognition through OpenCv - Python: In

our project we describe the development of a software

system oriented to the Recognition of Vehicle Plates

using both hardware and software tools. The system

will locate the vehicle plate inside the image of the

vehicle and will recognize the characters and num-

bers that are inside it and will be compared with those

of the database that contain all the license plates of

the vehicles that are authorized to access the vehicle.

the University on Paris Street, seeking to optimize the

access time.

3 TEORETHICAL FRAMEWORK

Computer vision aims to solve several of the problems

that arise in the daily life of human beings. Every-

thing that humans observe is through our vision, that

is why we have investigated several projects to cap-

ture ideas of how computer vision works.

3.1 Arduino UNO

It is an electronic board based on the ATmega328P

(technical sheet). It has 14 digital input pins, 6 analog

inputs, a 16MHz quartz crystal, a USB connection,

a power connector, an ICSP header and a reset but-

ton. The model of the Arduino with its different ports,

circuits and components, can be programmed with

the Arduino software (IDE). But this module is al-

ready pre-programmed with a boot loader that allows

you to load a new code into the system without the

use of an external hardware programmer (Badamasi,

2014),(Enciso-Quispe et al., 2017a).

Figure 1: Arduino UNO.

3.2 Artiﬁcial Vision

Artiﬁcial computer vision is the ability of the machine

to see the world around it, more precisely to deduce

the structure and properties of the three- dimensio-

nal world from one or more two- dimensional images

IoT in Services for Communities in the Recognition of Plates with OpenCV-Python

431

(Pajares and de la Cruz, 2008). It is deﬁned as a ﬁeld

of ”Artiﬁcial Intelligence”, allows the use of appro-

priate techniques, allow the collection and analysis of

any spatial information obtained through digital ima-

ges. It is composed of a set of processes designed to

perform image analysis. These processes are: captu-

ring images, memorizing information, processing and

interpreting results. With artiﬁcial vision you can:

• Automate repetitive inspection tasks performed

by operators.

• Conduct quality control of products that could not

be veriﬁed by traditional methods.

• Perform inspections of objects without physical

contact.

• Reduce cycle time in automated processes.

3.3 Artiﬁcial Intelligence

Artiﬁcial intelligence (AI) or also known as compu-

tational intelligence is the intelligence shown by ma-

chines, in a broad and somewhat circular deﬁnition,

aims at the study of intelligent behavior in machines.

In turn, intelligent behavior involves perceiving, rea-

soning, learning, communicating and acting in com-

plex environments(Nilsson, 2001).

The machine has to be able to recognize the natural

language in which humans speak. Speech is associa-

ted with a superior intelligence, and for a machine to

be able to recognize it and also to build sentences it

must be able to perform complex morphological ana-

lyzes , syntactic, semantic and contextual of the in-

formation it receives and the phrases it generates. In

order for an artiﬁcial entity to be considered intelli-

gent, the following is considered (Serrano, 2013):

• Natural Language Processing or NLP.

• Automatic reasoning and Machine learning.

3.4 Neuronal Networks

Neural networks are schemes that aim to mimic the

architecture of the brain. It serves to effectively re-

present any function that is very complex in algebraic

terms, as well as pattern classiﬁcation tasks (C. San-

chez, V. Sandonis., 2017). Artiﬁcial neural networks

(ANNs) arelearning systems inspired by the functi-

oning of the human brain. They are mathematical

models constructed based on the functioning of bio-

logical neural networks (H. Vega, A. Cortez, A. M.

Huayna, L. Alarc

on, P. Romero., 2017).

3.5 Detection of Vehicle Plates

The techniques must be chosen according to the needs

of the particular environment in which the system will

operate, so that the methods with the highest perfor-

mance in general will be exposed. In the same way

factors such as performance, execution time and plat-

form should be taken into account, since they will di-

rectly affect the performance of the system.

3.5.1 Processing of Binary Images

This technique is based on combinations of statistical

methods of analysis and mathematical morphology.

(Used to extract useful components of images, for the

representation and description of the shape of regi-

ons)(R.Radha, 2012). The method, is based on the

principle that the change of brightness in the region

of the plate, is more remarkable and stronger than in

any other place of the image.

3.5.2 Grays Level Processing

1. Global Image Processing. For the location of the

plate, several systems use an approach as presented

in (Draghici, 1997), which consists of scanning the

image horizontally, looking for repeated changes

of contrast on a scale of 15 pixels or more. The

assumptions that the contrast between the characters

the bottom of the plate is su ciently good and that

there are at least three to four characters in a plate

whose minimum vertical size is 15 pixels(Draghici,

1997).

2. Partial Analysis of the Image. This method scans

the image of the vehicle partially, making several re-

petitions of the analysis in different sections of the

image taking into account some distance N- the. as a

limit of the section, to then count the existing borders

in said section. If the number of edges is greater than

an average, it can be assumed that the plate was lo-

cated in that area. The execution time of this method

is very fast since it only scans certain images of the

image (J. Cano, 2003).

3.6 Python

Python is a programming language whose philosophy

emphasizes a syntax that favors a readable code, it

is a programming language multiparadigma, since it

supports object orientation, imperative programming

and, to a lesser extent, functional programming. It

has an interpreter by command line in which you can

enter sentences. Each sentence is executed and pro-

duces a visible result, which can help us understand

ITSCO 2018 - Special Session on Internet of Things and Smart Communities

432

Figure 2: Successfull detection of the vehicle plate.

better the language and test the results of the execu-

tion of code proportions quickly ( Python Software

Foundation, 2017).

3.7 OpenCV

OpenCV (Open Source Computer Vision Library) is

released under BSD license and, therefore, is free

for both academic and commercial use. It has C

++, C, Python and Java interfaces and is compatible

with Windows, Linux, Mac OS, iOS and Android.

OpenCV was designed for computational efﬁciency

and with a Strong focus on real-time applications.

Written in optimized C / C ++, the library can take

advantage of the processing of multiple cores. Ena-

bled with OpenCL, you can take advantage of the har-

dware acceleration of the underlying heterogeneous

information platform (OpenCv Team, 2017), (Enciso-

Quispe et al., 2017b).

4 METHODOLOGY

4.1 Analysis

Currently there are different types of problems with

respect to control vehicular in parking areas or areas

of restricted access, added to these problems does not

take a statistical control of the different parameters

that are generated when entering or leaving a vehicle

from the parking lot, that is, vehicles entered, parking

availability, parking demands over time, or for pre-

vention of future problems. This project focuses on

the development of a detection system and extraction

of vehicle plates in the area of access to a parking

lot based on artiﬁcial vision. The system will allow

access to be automated, if the vehicle is in the data-

base and if it is veriﬁed that the vehicle belongs to

someone belonging to the university, it will be given

access, otherwise it will be denied.

4.2 Diagram of the Application

In the Figure 1 and 2, details all the components in-

volved in the use of the application, it shows that the

user approaches the control and is when the IP camera

communicates through the WI-FI with the router the

LPR System complements the next:

• Identiﬁcation of a user using his vehicle license

plate.

• Presentation of information related to a plate.

• User administration that will be identiﬁed later

through vehicular plate.

• Guarantee access to the parking lot.

• It will control a prototype of a front entrance to a

parking lot.

IoT in Services for Communities in the Recognition of Plates with OpenCV-Python

433

Figure 3: Diagram of the application.

Figure 4: Values of the positioning of the letters in the image.

4.3 Software Development

For the development of the system it was decided to

use the good practice SCRUM for this the following

roles are deﬁned:

• Scrum master: person who leads the team to com-

ply with the rules and processes of the selected

methodology.

• The representative of the client who will use the

software. It focuses on the business side. Trans-

fers the vision of the project to the team, formali-

zes the beneﬁts in stories to be incorporated in the

Product Backlog and repriorizes them on a regular

basis.

• Team: consisting of two people with the neces-

sary technical knowledge and who developed the

system along with the stories of each sprint.

ITSCO 2018 - Special Session on Internet of Things and Smart Communities

434

Figure 5: Detect Function.

Figure 6: Return the outline of the plate.

4.4 Criteria of Acceptance

The acceptance criteria are the basis for the appli-

cation which will analyze the requirements, types of

tools that will be used and how they will be develo-

ped, facilitating the functionality of the application,

among them we have:

• Independence of the platform.

• Attractive and functional interface.

• Updated information.

Figure 7: Go through the images on the plates.

5 RESULTS

The system:

• Allows us to capture the plate through the camera

and send it to the system.

• Performs the segmentation of characters by means

of the algorithm.

• Recognizes the characters and very that the dis-

played plate has access.

• Can store the license plate of all the vehicles that

have entered.

• Can store people who have tried to enter without

being part of the university.

6 CONCLUSIONS

• It seeks to automate the access of the staff of the

institution for control of Paris street without pro-

blems and in a faster way minimizing effort and

maximizing its performance.

• Avoid entry to people who are not in the database

of the University.

• Achieve a record of the people who are authorized

to enter the university on Paris Street.

IoT in Services for Communities in the Recognition of Plates with OpenCV-Python

435

Figure 8: Extraction of the number the plate.

REFERENCES

Python Software Foundation (2017). Python. url:https://

www.python.org/.

Badamasi, Y. A. (2014). The working principle of an ar-

duino. In 2014 11th International Conference on Elec-

tronics, Computer and Computation (ICECCO), pa-

ges 1–4.

C. Sanchez, V. Sandonis. (2017). Reconocimiento

Optico

de Caracteres . url:http://www.it.uc3m.es/jvillena/irc/

practicas/08-09/09.pdf.

Draghici, S. (1997). A neural network based artiﬁcial vision

system for licence plate recognition. Signal & Image

Processing, 8(1):113–126.

E. Caballero. (2017). Aplicacion Practica De La Vision Ar-

tiﬁcial Para El Reconocimiento De Rostros En Una

Imagen, Utilizando Redes Neuronales Y Algoritmos

De Reconocimiento De Objetos De La Biblioteca

OpenCv.

Enciso-Quispe, L., Barba-Guaman, L., Sanchez, J., and

Quezada-Sarmiento, P. A. (2018). Simulation of pe-

ople counter for public service buses of loja with iot

concept applying the viola-jones algorithm. In 2018

13th Iberian Conference on Information Systems and

Technologies (CISTI), pages 1–6.

Enciso-Quispe, L., Delgado, J., Vivanco, H., Zelaya-

Policarpo, E., and Quezada-Sarmiento, P. A. (2017a).

Internet of things based on android technology for

people with disabilities. In 2017 12th Iberian Con-

ference on Information Systems and Technologies

(CISTI), pages 1–6.

Enciso-Quispe, L., Quichimbo, J., Luz

on, F., Zelaya-

Policarpo, E., and Quezada-Sarmiento, P. A. (2017b).

Rest architecture in the implementation of a web and

mobile application for vehicular tariff rotating par-

king. In 2017 12th Iberian Conference on Information

Systems and Technologies (CISTI), pages 1–6.

Garc

ıa., A. (2015). Reconocimiento De Objetos Inmersos

En Im

agenes Est

aticas Mediante El Algoritmo Hog

Y Rna-Mlp. Universidad Tecnol

ogica de la Mixteca,

exico (2015).

H. Vega, A. Cortez, A. M. Huayna, L. Alarc

on, P.

Romero. (2017). Reconocimiento de patrones

mediante redes neuronales artiﬁciales . url:http://

sisbib.unmsm.edu.pe/BibVirtual/Publicaciones/risi/

2009 n2/v6n2/a03v6n2.pdf.

I. Ortiz (2016). Google Visi

on Api. url:https://

elandroidelibre.elespanol.com/2016/02/google-

vision-api-reconocimiento-de-fotos.html.

J. Cano, J. C. P. C. (2003). Vehicle license plate segmenta-

tion in natural images. Pattern recognition and image

analysis, pages 142–149.

Lanzarini, L. and Giusti., A. D. (2014). Aplicacion De Re-

des Neuronales Al Reconocimiento De Patrones En

Imagenes Medicas. Universidad Nacional de la Plata,

Argentina (2014).

Nilsson, N. (2001). Inteligencia Artiﬁcial:Una Nueva

ıntesis. Mcgraw-Hill Grupo Editor, Espa

na (2001).

OpenCv Team (2017). OpenCv . url:https://opencv.org/.

P. Garc

ıa. (2013). Reconocimiento De Im

agenes Utilizando

Redes Neuronales Artiﬁciales. .

Pajares, G. and de la Cruz, J. (2008). Visi

on Por Computa-

ITSCO 2018 - Special Session on Internet of Things and Smart Communities

436

dor: Im

agenes Digitales Y Aplicaciones. Alfaomega

Grupo Editor, M

exico (2008).

Paucar, L. and Tasiguano, C. (2016). Desarrollo de algorit-

mos para el reconocimiento de placas de veh

ıculos.

Peralta, A. (2009). Reconocimiento de imagenes a tra-

ves de su contenido. Universidad Pontiﬁcia Comillas,

8(1):202.

R.Radha, C. (2012). A novel approach to extract text from

license plate of vehicles. Signal & Image Processing,

3(4):181–192.

Serrano, A. G. (2013). Inteligencia Artiﬁcial: Fundamen-

tos, pr

actica y aplicaciones. Alfaomega Grupo Editor,

exico (2013).

IoT in Services for Communities in the Recognition of Plates with OpenCV-Python

437