_"J
#:~~
~?
••
3
(.~J lkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
'"
X1JlENFIR
lXENAN
INFCRMATICNlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
THE EVENT
CONTACT
In the beginning
DETAILS
Promoter
ABEN - Brazilian Nuclear Energy Association
alternative
to guarantee
the planet,
bringing
a sustainable
together
future for
safe energy supply
Besides being
an energy souree that helps to reduce
- Brazil
greenhouse
benefit
Contads
gas emissions,
Edson Kuramoto - Chair - ELETRONUCLEAR
ekura@eletronuclear.gov.br
Parejo Calvo - Chair
nuclear
areas like medicine,
agriculture.
Wilson Aparecido
nuclear
presents itself as an efficient
with respect for the environment.
Rua Mena Barreto, 161 - Botafogo
Rio de Janeiro/RJ - CEP 22271-100
of the 21 st eentury,
technology
Nuclear
applications
industry
and
energy will have an
important
role to play in the world's
expansion
and teehnologieal
energy
innovations
in the
next deeades.
IPEN-CNEN/SP - wapcalvo@ipen.br
Web: www.inac2009.com.br
In this context, the Brazilian
Phone: + 5521 3797-1751
Fax: + 55213797-1751
Association
Nuclear
Atlantie
Conferenee
will be held at the Windsor
Information
Full Time Assessoria & Eventos - organization
generallogistics
Executive
of ExpolNAC
and
2009.
secretariat
Full Time Assessoria & Eventos
Rua Alberto Veiga, 69 - Marape
Santos/SP - CEP 11070-030
- Brazil
Web: www.inac2009.com.br
Eliete Fraga
KEY
for a Sustainable
pretends
eontemporary
Brazilian
2
nd
,
in Nuclear
Future", the
to diseuss the eontribution
in the nuclear
sedor
can give to
society in order for it to develop
a sustainable
manner.
and foreign
in
To achieve this goal,
specialists will be invited to
involving
on Nuclear
perspectives,
the environment
Meeting
REMEMBER
alongside
the XVI
Reactor Physics and Thermal
(Enfir) and the IX Meeting
Applications
55 13 3252-1290
TO
Inae 2009 will be promoted
Hydraulies
55 13 3877-5027
DATES
"Innovations
conference
Meeting
eliete@fulltime-eventos.com.br
+
2Jfh to Odober
Technology
that innovations
(Inae 2009), whieh
Barra Hotel, in Rio de
and energy security.
lombardi@fulltime-eventos.com.br
+
With the theme
as well as questions
Contacts
Fax:
2009.
from September
Energy
the International
debate and analyze teehnologieal
Vanessa Lombardi
Phone:
Janeiro,
Nuclear
(Aben) is promoting
(Enan). lhe
on Nuclear
on Nuclear
event will also host the I
Industry.
Also a part of the event's official
program
Expolnae, an exhibition
public and
private eompanies,
in which
nuclear
is
and non:nucl.ear,
.
will
•
take part, offering produds and servrces IA
Set-up dates wvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
various areas of applieation.
Expolnae 2009 will
•
26 'h Seplember, 2009 from 08:00 em 1018:00 pm
present tendencies, alternatives, studies, pr?ieds
and accomplishments
Exhibition
opening
hours
•
27 'h Seplember 2009 from 15:00 pm 10 23:00 pm
•
28 'h Seplember 2009 from 08:00 em 10 19:00 pm
•
29 'h Seplember 2009 from 08:00 em 1019:00 pm
•
30 'h Seplember 2009 from 08:00 em 1019:00 pm
•
01 " October 2009 from 08:00 em 10 19:00 pm
nd
02 October 2009 from 08:00 em 10 19:00 pm
•
or indirectly
theme.
laking
engineering,
innovation
•
/ Move-out
03,d Oclober, 2009 from 08:00 em 10 18:00 pm
IA
part in the exhibition
consulting
companies
areas directly
central
will be
and technological
and service providers.
Equipment
and machine
cornponiés
that work with management
manufadurers
research and development,
transmission
Dismantling
developed
linked to theconference's
lubricants,
efficiency
institutions
and technical
participate
fuels and
sanitation
will also be present.
and technological
sof!ware,
energy generatlon,
and distribution,
infrastrudure,
and
and energy
Educational
and scientific,
associations
business
will likewise
in Inac 2009.
Bring your eompany
to this eventl
2
2 0 0 9 1 n te r n a lio n a l
N u c le a r A tla n tic C o n J e r e n c e - I N A C
R io d e J a n e ir o .R l,
B r a z il, S e p le m b e r 2 7 10 O c to b e r
2009
wvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
2. 2 0 0 9
A S S O C IA Ç Ã O B R A S IL E IR A D E E N E R G IA N U C L E A R - A B E N
IS B N :
9 7 8 -8 5 -9 9 1 4 1 -0 3 -8
PATTERN RECOGNITION
OF NUCLEAR TRACES PRODUCED
ALPHA PARTICLES IN SOLID STATE DETECTORS
Ananda C. Trindade
1
,2,\
BY
Paulo E. Cruvíneli, Nivaldo C. da Silva 3 and José H. Salto'lkjihgfedcbaZY
de Computação
I Departamento
Universidade Federal de São Carlos - UFSCar
Rodovia Washington Luis, km 235
13565-905 - São Carlos, SP
anandat@gmail.com,
saito@dc.ufscar.br
2
3
Laboratório
Embrapa Instrumentação Agropecuária
Rua XV de Novembro, 1452
13560-970 São Carlos, SP
cruvinel@cnpdia.embrapa.br
de Poços de Caldas - Comissão Nacional de energia Nuclear,
Rodovia Poços de Caldas - Andradas, km 13
37701-970 Poços de Caldas, MO
ncsi Iva@cnen.gov.br
LAPOC/CNEN
ABSTRACT
A great concem about the spread of radon gas in human living environments requires an improvement of the
detection techniques used nowadays and the development of new technologies that provides greater reliability,
accuracy and agility in the process of analyzing the results generated by emission of radon. Aiming to increase
the analytical capacity of Radon Laboratory, unit of Brazilian Commission ofNuclear Energy, located in Poços
de Caldas, the automation ofthe XY table ofthe optical microscope was proposed and implemented and also a
new technique of nuclear traces analysis generated by alpha particles from radon using digital imaging
processing and artificial neural networks as pattem classifiers.
1. INTRODUCTION
Radon and its decay products contribute to the majority of ionizing radiation received by
worldwide population. Studies carried out a long of the years confirmed the association
between exposure to radon and lung cancer. Radon when inhaled is eliminated almost
immediately, however, its decay products are deposited in the lungs which emits alpha
particles that interact with the lung tissue and may also reach the bloodstream, affecting other
organs like the bone marrow.
According to UNSCEAR (United Nations Scientific Committee on the Effects of Atomic
Radiation), radon contributes approximately 50% of the natural radiation dose received by
the worldwide population annually. Outdoors concentrations ofradon are very low, however,
high concentrations are observed in underground mines, particularly in uranium mines.
Considerable
concentrations
of radon are also found in homes depending on the
characteristics ofthe building materiais, underground and ventilation.
Thus, this work presents an automated technique for obtaining images produced by alpha
particles in solid-state detectors through a digital optical microscope coupled with a digital
camera CCD and motors capable of moving it along the X and Y axes, which guarantees a
complete scan of the detector, increasing agility in the process of capturing images and also
the development of an algorithm to analyze those images, based on techniques of digital
image processing using Hough transform to identify traces and Artificial Neural Networks as
pattems classifiers and indicators that qualify the alpha particle in terms of its angle of
incidence and energy.
2. NUCLEAR TRACES PRODUCED BY NUCLEAR PARTICLES
2.1. Radon
Radon was discovered by Owens and Emest Rutherford in 1899 and it is the product of alpha
decay of 226Ra, result the alpha decay of radioactive series of 238U. It is a colorless, odorless,
tasteless and has half \ife of 3.823 days and for being a noble gas is easily disseminated to
human living environments through building materiais, soil and water.
2.2. Radon detection techniques
The most widely used techniques for radon detection can be divided into two groups: active
detection and passive detection, where each of these techniques can be used to detect 222Rn
and its decay products [1].
Active techniques detection consists of a few liters of air samples which are placed in a
radiation detection system. The main techniques used are the Lucas cells and ionizations
chambers. In the passive techniques detections so\id state nuclear track detectors (SSNTD)
are exposed to the environment for a certain period of time, with a diffusion chamber, which
are permeable only for 222Rn [1].
2.3. Tracks formation and etching process
Tracks formation occurs when a heavy particle (alpha particles, protons, etc.) incises on a
SSNTD causing damage in the material. This damage is named latent track, Figure 1, and
each one has different characteristics according to their energy and angle of incidence baZYXWVUTSRQPON
!
L _.~
- . - r - - - - - - - - - , lkjihgfedcbaZYXWVUT
/
/
/
__ ~..__
..._~
LATENT
TRACK
SSNTD
...._ ._
Figura 1. Latent track on a SSNTD.
To the latent tack be observed, the SSNTD must go through a process of etching, which
consists of a chemical attack that etches its surface, increasing the diameter of the latent
track. This method is called EPC (Chemical Pre Etching). There is also another process of
etching called ECE (Electro Chemical Etching). The difference between then is that the ECE
during the chemical process of etching applies a voltage in SSNTD [2]. The most used
chemical reagents are potassium hydroxide (KOH) and sodium hydroxide (NaOH).
The appearance of an alpha particle incident on a SSNTD after chemical etching process can
be seen in Figure 2, which iIIustrates the surface of CR-39 detector exposed to an
environment with high radon concentrations.
lNAC 2009, Rio de Janeiro, RJ, Brazil.
Figura 2. Surface of an SSNTD exposed to an environment with a high radon
concentration after chemical etching process in NaOH solution.
3. DIGITAL IMAGES PROCESSING
AND ARTIFICIAL
NEURAL NETWORKS
In the last 30 years, considerable advances have been developed in pattems recognition, and
image processing with applications to vision systems. These advances have led to a great
need to develop methods, software and hardware.
From the initial concepts of signal processing and systems theory, image processing depends
mainly on linear filters and convolution masks. Recently, image processing has been
developed mainly in areas of frequency analysis, nonlinear analysis, space-variant filtering
and analysis based on methods that provide better results than the traditional techniques.
Digital image processing can be used in several applications in various levels of knowledge,
as geoprocessing, medicine, X-ray images or MRI, astronomy, recognition of celestial bodies
or applications for facial and retina recognition and also as a form of classification and
pattems recognition.
3.1 Hough transform
Duda and Hart [3] suggested the use of the Hough Transform for straight adapted to circles.
Using the circle equation given by Equation 4, where a and b are its center coordinates and c
its radius, it is possible to use the Hough Transform. The implementation of the Hough
Transform consists to change the image from the Cartesian plane in a parameter space, i.e.
each pixel on the Cartesian plane is converted into a circle in the parameter space. The
intersection of circles in the parameter space will define the center coordinates of a circle and
the total radius ofthe circle [4].
(4)
Figure 6 (a) illustrates a circle with five pixels and radius equal to baZYXWVUTSRQPONMLKJIHGFEDCBA
1 /.fi
in the Cartesian
plane (x, y). Figure 6 (b) illustrates image mapping from the Cartesian plane to a parameter
space.
INAC 2009, Rio de Janeiro, RJ, Brazil.
4,0
4,
3,5
3,5
3,
3,0
3,0
3,5
4,0
5,0
3,03,5 4,0
(a)
Figura 6. (a) CircIe in a Cartesian
5,0
wvutsrqponmlkjihgfedcbaZYXWVUTSRQ
(b)
b.
plane x, y; (b) parameter space a and baZYXWVUTSRQPONMLKJIHGFE
Each pixel, Figure 6 (a) creates a circle in the parameter space, Figure 6 (b). Thus, points a
and b generated in the parameter space are stored in an accumulator array. The intersection of
circles in parameter space indicates the circle center coordinates and its sum indicates the
pixels belonging to the circle. The circle equation is given by Equation 5.
(5)
For this approach, where the radius is known, the size of the accumulator
the dimensions shown in Equation 6.
A C C [X
max
+ 2 .r a io J [ Y max + 2 .r a io ]
array should have
(6)
However, when you want to detect a circle with radius value is unknown is necessary to use a
three dimensional accumulator array with the dimensions shown in equation 7.
A C C (X
max
+ L r a io ).(Y max + L r a io ).(rm ax - rm in
+ 1)
(7)
3.2 Artificial neural networks
An artificial neural network is a mathematical model or computational model that tries to
simulate the structure and functional aspects of biological neural networks. It consists of an
interconnected group of artificial neurons and processes information using a connectionist
approach to computation.
Neuron has a structure represented by Figure 7. Neurons receives input signals, xl, x2 ,..., xm
and are equipped with synapses, which modulate the received signal, represented by the
synaptic weight wj,
INAC 2009. Rio de Janeiro. RJ. Brazil.
Yj
lkjihgfedcbaZYXWVUTSRQPONMLKJIH
Output
Activation
Function
Xm ---f
Synaptic
Weights
Input
Bias
Figura 7. Artificial neuron or artificial neuron network.
The activation function process the stimuli weighted sum by the synaptic weights and is
defined by Equation 8, where Yj is the output generated by each neuron in the layer above.
(8)
With the multilayer perceptron networks was possible to solve non-linear problems, ending
with the Iimitations of single-Iayer perceptron networks. The supervised leaming process
consists to adjust the weights of the network so that they achieve a configuration that is
capable of mapping entries in the desired outputs. Backpropagation
is the most popular
algorithm to make these weight adjustments and to minimize errors.
Backpropagation uses a sequence of two steps, where the first consists in the pattems
presents to the input layer and that cover the entire network until a response is obtained by
the output layer. ln the second step, the output obtained is compared with the desired
response and its error is calculated and then propagated from output layer to input layer and
the weights are adjusted and modified [6].
4. lMAGE ACQUISITION
To image acquisition is used an automatic capture system of images which consists of a
digital optical microscope, Figure 8 (a), which has motors capable of moving it along the X
and Y axes, Figure 8 (b). The motors are controlled by a programmable microcontroller PIC
type 16F877A, Figure 9.
INAC 2009, Rio de Janeiro, RJ, Brazil.
(a)
(b)
Figura 8. (a) Digital optical microscope used to automatic capture images. (b) Digital
optical microscope and its motors capable of moving it along X and Y axes.
Figura 9. Programmable
microcontroller
PIe type 16F877A.
The algorithm developed for the management of the system is able to send instructions to the
microcontrolIer device in the hardware and perform reading of the current situation. This
algorithm is also responsible for the process of capturing images from the microscope to be
analyzed later.
For perfect control of the X and Y axes, were developed two algorithms, which the first one
is installed on a compute r and the other one recorded in a microcontroller. The software
installed on a computer is responsible for positioning the X and Y axes in an initial position
and adjust alI the settings necessary to start the process of image acquisition. After the initial
setup, the system begins to acquire images. When the first image is acquired, the software in
the microcontrolIer sends a control sign indicating a new position for the X and Y axes, and
acquired a new image. This process is repeated until ali images are captured. The software
INAC 2009, Rio de Janeiro, RJ, Brazil.
recorded in the microcontroller receives from serial port commands from computer and send
signals that wil\ put the motors in motion.
5. PATTERNS
To create a database, which consists of images and the bank's standards, were carried out a
series of experimental studies to determinate the behavior of the alpha partic\es incident on
CR-39 due to its energy and angle of incidence. The objective of these studies was to obtain
traces of alpha partic\es of known ener~y and incident angles. The experiments were
performed using a radioactive source of 24 Am, which emits alpha particles with energy of
5.486 Me V, restricting the angle of incidence at 30 0, 45 0, 60 0, 70 lkjihgfedcbaZYXWVUTSRQPONMLKJIHGFED
° and 90 ° with respect to
the normal surface. To restrict the angles were used collimators made of acrylic and with
central holes of 1 mm with pre-determined angles. Figure 10 shows collimators with
incidence angles of90 0, 75 0, 60 0, 45 ° and 30 0.
Figura 10. Collimators with incidence angles of 90°,75°,60°,45°
e 30°.
To avoid the energy loss of alpha particles emitted by radioactive sources, a vacuum camera
was used, Figure 11 Therefore, we can ensure that partic\e's energy wont decrease until
reaches detector surface. For each experiment performed in the vacuum camera, a similar was
performed in the air using the same radioactive source and collimator.
Figura 11. Vacuum camera.
INAC 2009, Rio de Janeiro, RJ, Brazil.
6. ACQUIRE
PROCESSING
AND IMAGE ANAL YSIS
The project under development is based on three steps consisting
digitization), processing and analysis (pattern c1assification).
of: acquisition
(image
The step of image acquisition involves: microscopy, XY table, control signals and acquiring
images. This stage of processing consists on image acquisition from a plastic detecto r using
the automatic control of the XY table and the camera attached to the microscope.
The next step is image processing which includes noise correction
acquired images of the microscope and preparing it to analysis step.
and focus from the
Image analysis, consists to identify patterns in image using the Hough transform, which is
able to identify circular and elliptical patterns, providing the largest and smallest diameters of
each event. Pattern recognition uses an artificial neural network that receives as parameters
data obtained from patterns identification and c1assifies them according to their energy and
angle of incidence.
A bank of images was designed with the purpose of storing images at ali stages of processing,
i.e., the original images, pre-processed images and images with patterns identified.
7. CONCLUSIONS
The automation process based on capturing and analysis of traces in images produced by
alpha particles in solid-state detectors can increase the analytical capacity of the Radon
Laboratory, CNEN unit, installed in Poços de Caldas. Therefore, based on this presented
work it will be possible improve the efficiency to determinate the radon concentration in
indoor environments, reducing errors and risk of false positives comparing with the manual
method used routinely.
REFERENCES
1.
PAULO, S. R. de,baZYXWVUTSRQPONMLKJIHGFEDCBA
" D o s im e tr ia A m b ie n ta ! d e R n - 2 2 2 e filh o s :
a b s o lu ta
do
C R -3 9
le v a n d o
em
c o n ta
o s e fe ito s
d o p la te - o u t
m e d id a
e fa to r e s
de
e fic iê n c ia
a m b ie n ta is " ,
Editora da Unicamp, Campinas, Brazil, 1991. Tese de Doutorado, Unicamp.
2. TOMMASINO, L., "Electrochemical Etching Process for the Detection of Neutrons and
& R a d ia tio n P r o te c tio n , v. 19, n. 1, pp. 12Radon-decay Products", N u c le a r T e c h n o lo g y lkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
19,2004.
3. DUDA, R. O.; HART, P. E.; STORK, D. G. - P a tte r n C la s s ific a tio n , John Wiley and
Sons, 2001.
4. MARTINEZ, A. C. "Um Novo Método para Medidas de Gotas de Chuva com Técnicas
do Processamento
Digital de Imagens", Universidade de São Paulo - Escola de
Engenharia de São Carlos, 2002.
5. HAYKIN, S. " N e u r a l N e tw o r k s : A C o m p r e h e n s iv e F o u n d a tio n " , Prentice Hall, 1999.
6. SUZUMURA, Y. F. "Método para Avaliação da Eficiência de Pulverização Agrícola
Baseado em Processamento de Imagens e Rede Neural, UNITAU - Universidade de
Taubaté,2005.
INAC 2009, Rio de Janeiro, RJ, Brazil.