Journal of the Faculty of Electrical Engineering
Topics: computers science, control systems, electronics,
electrotechnics, drive systems
AIMS AND SCOPE
Advances in Electrical and Computer Engineering, the Electrical Section is the scientific official journal of the Faculty of Electrical Engineering.
The publication policy is to publish high quality original scientific articles and reviews with permission from the Editorial Board. The authors must translate themselves their articles into a foreign language accepted by the UE. It should not have been previously publish and should not be under consideration for publication elsewhere.
The scope of the journal covers: computers science, artificial
intelligence, pattern recognition, knowledge engineering, process control
theory and applications, distributed systems and computer networks, microcontroller
- based industrial networks, electronics, telecommunications, data transmission,
software engineering, energy converters, electrotechnics, electric machines
modeling and design, control of electric drive systems, non-conventional
energy conversion, optimization techniques in electrical power systems,
sensors, transducers modeling and design.
Copyright © 2005 Faculty of Electrical Engineering
Stefan cel Mare University of Suceava, Romania
All rights reserved. No part of this publication
may be reproduced, stored in a retrieval system, photocopied, recorded
or other wise, without written permission from the editor.
When authors submit their papers for publication, they
agree that the copyright for their article be transferred to the Faculty
of Electrical Engineering, the Stefan cel Mare University of Suceava,
Romania, if and only if the articles are accepted for publication. The
copyright covers the exclusive rights to reproduce and distribute the article,
including reprints and translations.
Permission for other use. The copyright owners
consent does not extend to copying for general distribution, for promotion,
for creating new works, or for resale. Specific written permission must
be obtained from the Editor for such copying.
Disclaimer. Whilst every effort is made by the
publishers and editorial board to see that no inaccurate or misleading
data, opinion or statement appear in this journal, they wish to make it
clear that the data and opinions appearing in the articles, as well as
linguistic accuracy, are the sole responsibility of the author.
The papers may be requested by e-mail directly to their authors with respect of the above copyright conditions !
[TABLE OF CONTENTS] [Issues on Internet] [DISTRIBUTION NETWORK] [EDITORIAL BOARD] [(e)MAIL]
Advances in Electrical
and Computer Engineering
ISSN 1582-7445 - No 1 / 2005
EDITOR |
Adrian GRAUR - Faculty of Electrical Engineering
ASSOCIATE EDITORS |
Ștefan Gh. PENTIUC
- Faculty of Electrical Engineering
Leon MANDICI - Faculty of Electrical Engineering
Dorel CERNOMAZU - Faculty
of Electrical Engineering
EDITORIAL BOARD |
Tudor AMBROS
- Technical University of Chișinău - MOLDAVIA
Dumitru N. ALEXANDRU
- Technical University Iasi
Oleg BREKHOV - Moscow Aviation Institute
- RUSSIA
Mihai CERNAT - "Transilvania" University Brasov
Cristophe CHAILLOU - Universite
des Sciences et Technologies de Lille - FRANCE
Mihai CRETU
- Technical University Iași
Luc DE BACKER - Katholieke Hogeschool
Sint-Lieven Gent - BELGIUM
Timothy HALL - University of Limerick
- IRELAND
Ileana HAMBURG -Institute for Technology
Gelsenkirchen - GERMANY
Stefan HOLBAN - "Politehnica" University
Timisoara
Maria G. IOANNIDES - National Technical
University of Athens - GREECE
Ioan JURCA - "Politehnica" University
Timisoara
Vladimir KATIC - University of Novi Sad - SERBIA AND MONTENEGRO
Eleftherios A. KAYAFAS - National
Technical University of Athens - GREECE
Mykola KIRICHENKO - Institute of
Cybernetics Kiev - UKRAINE
Valeriu KRAPIVIN - Moscow Institute
of Radioengineering and ELectronics - RUSSIA
Ioan LETIA - Technical University
Cluj-Napoca
Volodymyr MESYURA - The State Technical
University of Vinnitsa - UKRAINE
John MILNER - City University London
- GREAT BRITAIN
Radu MUNTEANU - Technical University Cluj-Napoca
Jose OLIVEIRA e SA - Instituto Politecnico
do Porto - PORTUGAL
Costică NITU - "Politehnica" University București
Adrian PETRESCU - "Politehnica" University București
Mircea PETRESCU - "Politehnica" University București
Nicola PITRONE - Universita degli Studi
Catania - ITALIA
Radu Emil PRECUP - "Politehnica" University Timisoara
Jean Paul SIX - Université des Sciences
et Technologies de Lille - FRANCE
Fedir SOPRONIUC - University of Cernăuți - UKRAINE
Mircea STRATULAT - "Politehnica" University Timisoara
Gavril TODEREAN
- Technical University Cluj-Napoca
Petru TODOS - Technical University
of Chisinau - MOLDAVIA
Bernard TOURSEL - Université des
Sciences et Technologies de Lille - FRANCE
Advances
in Electrical and Computer Engineering
ISSN 1582-7445 - No 1 / 2005
TABLE OF CONTENTS
FEE INFO
16 21 May 2005
The Faculty of Electrical Engineering organized the 12th Annual International Students Computers Contest Hard & Soft. The theme of the contest was a Lift Control System based on remote instrumentation, analysis / control and feedback.
The contestants were required to design and build an electronic system linked to a PC where data manipulation had to lead to system feedback and user display. Some elements of the system could be remote. Three implementations were demanded: hardware, software, hardware-software.
Participants: teams from Beijing (China), Brasov, Bucuresti, Cernautzi (Ukraine), Chisinau (Moldavia), Craiova, Iasi, Suceava, Timisoara.
President of the jury: Professor Timothy HALL from University of Limerick, the Republic of Ireland. His team included Cathal McHUGO, Kevin JOHNSON from the same university, and Claudine LECOCQ from Ecole Politechnique Universitaire de Lille, France.
The prizes were awarded to:
the Politehnica University of Timisoara - 1st prize
the University of Technology in Beijing - 2nd prize
the Technical University of Moldavia in Chisinau - 3rd prizeAn International Robotics Seminar was held the same week. It brought together scientists, engineers, manufacturers, graduate students and postdoctoral scholars in order to provide and exchange information on recent developments and future trends in Robotics Theory, Artificial Intelligence Focused on Robotics, Electrical Drives Control, Pattern Recognition.
Professor Adrian GRAUR
HYBRID NEURAL APPLICATION IN MANURA ROBOT LOCALIZATION
Ioan Alfred LETIA
Technical University of Cluj-Napoca
Baritiu 28, RO-3400 Cluj-Napoca, Romania
letia@cs.utcluj.ro
Oliviu MATEI
Technical University of Cluj-Napoca
Baritiu 28, RO-3400 Cluj-Napoca, Romania
letia@cs.utcluj.ro
Abstract. The robot Manura has been designed for cleaning
barns and equipped with short-range sensors. The localization of this operational
robot is accomplished by a neural network. We focus on the learning techniques
applied in this project. The training is improved by adding some self-generated
exemplars
Keywords: robotics, localization, hybrid neural
approach regression, classification, prediction, Manura robot
RESERVOIR INFLOW FORECAST USING NEURAL NETWORKS: A
CASE STUDY OF WANGCHU RIVER OF BHUTAN
Jigme SINGYE1 , Katsumi MASUGATA2, Murai TADAKUNI
Department of Electrical and Electronic System Engineering,
Faculty of Engineering, Toyama University,
3190 Gofuku, Toyama, 930-8555, Japan
1 jsingye@ecs.toyama-u.ac.jp
2 masugata@eng.toyama-u.ac.jp
Abstract. Efficient river inflow forecast is essential
for various purposes such as for flood control, distribution of water for
irrigation purposes, etc. It is also a highly useful tool for the efficient
operation and maintenance of hydro power plants as the economics of the
hydro electricity depend largely on the reservoir height as well as the
inflow rate into the dam. Through effective inflow prediction mechanism,
proper maintenance and operation schedule of the generating machines can
be planned, thereby reducing the forced outages and increasing the generation
output.
In this paper, a neural network based approach is presented to forecast
the daily river inflow for a Wangchu River of Bhutan since on this river
basin lie the two biggest hydro plants, Chukha Hydropower Corporation (CHPC)
and Tala Hydro Project Authority (THPA), with generating capacities of
336 MW and 1020 MW, respectively. For a small Himalayan kingdom where half
the total national revenue comes from the sale of hydro power alone, with
almost around 80 percent exported to India, it is essential to have a proper
mechanism to accurately forecast the inflow as it would help implement
the optimal utilization of water resources as well as plan efficient load
scheduling. The latter is particularly important for power export as prior
electric generation anticipation based on river inflow is critical in transacting
the power sale in advance, thereby significantly improving the revenue
earned.
In this paper, two types of neural networks are designed and their
performances compared in predicting the next day inflow. Both the networks
are extensively tested using the inflow and other weather related data
from the year 1999 to 2003.
Keywords: reservoir inflow, neural network, hydro power.
BUS COMMUNICATION IN ROBOTIC LEADING SYSTEMS
George MAHALU1, Adrian GRAUR2
"Ștefan cel Mare" University of Suceava, Str.Universității nr.9, RO-720225
Suceava
1)mahalu@eed.usv.ro,
2)adriang@eed.usv.ro
Abstract. In this paper work are proposing some original
solutions particular to the data transfer mean and protocols corresponding.
The bus imagined has some specific features that conferring originally
behaviors. The protocol involved is thinking in this application to working
in logical mean with the hardware-designed structure. In this sense is
performed entire hardware in relation with the signal type generated by
the various kinds of sensors. Was be given attention to the adapting electronic
signal structures like as an interface between sensors involved, from one
the side, and the physical bus system (with all its conflict problems which
can appear). The protocol will be considering these features and will be
thinking in so manner to increasing the capabilities of the system.
The authors contributions are located both electronic hardware part
(through the physical line transmitting and the adapter electronic modules)
and software part (through the implemented protocol routine system). The
electronic hardware part is performing with an 8-bit PCB80C553 Philips
microcontroller board. The software part consists in three different routines
type: a program to communication with the PC system (involved in the loading
procedure), an interfacing program between the microcontroller system and
the user, and a protocol communication with the designed bus program. For
the PC communication is used the C language and for the user interface
and the protocol communication on the bus is used the microcontroller's
assembler language.
For PC/PCB communication can be used a wireless channel data transfer
and so is increased the handling system capabilities.
Keywords: I2C bus, communication, self-arbitration, bus
adapting, data signal. clock signal.
CONVERTERS EMC - CAPACITIVE COUPLING AND PARASITIC CAPACITANCES
Irena KOVÁÈOVÁ
Department of Electrical Drives and Mechatronics, Technical University
of Koice
Letná 9, 042 00 Koice, Slovak Republic
Irena.Kovacova@tuke.sk
Dobroslav KOVÁÈ
Department of Theoretical Electrical Engineering and Electrical Measurement,
Technical University of Koice
Park Komenského 3, 042 00 Koice, Slovak Republic
Dobroslav.Kovac@tuke.sk
Abstract. The paper is dealing with some electromagnetic
compatibility (EMC) problems of converters and inverters, which are utilized
for feeding of electric machines and its analyzing by PSPICE program. The
main attention is focused on implications of parasitic capacitance and
capacitive coupling existence.
Keywords: EMC, converter, PSPICE simulation, parasitic
capacitances, capacitive coupling.
DIAGNOSABILITY OF FLEXIBLE ROBOTIC MANIPULATORS FOR
RAILWAYS SYSTEMS
Calin CIUFUDEAN
"Stefan cel Mare" University of Suceava
str.Universitatii nr.9, RO-720225 Suceava
calin@eed.usv.ro
Alexandru Bogdan LARIONESCU
"Stefan cel Mare" University of Suceava
str.Universitatii nr.9, RO-720225 Suceava
lari@eed.usv.ro
Daniel POPESCU
Technical University of Constructions,
66 P.Protopopescu Bvd. 73232, Bucharest
dpopescu@instal.utcb.ro
Abstract. Fault detection is a crucial and challenging
task in the automatic control of complex system, e.g. in flexible manufacturing
systems, and robotic systems as representative classes of discrete event
systems.
In this paper, we consider the discrete event systems (DES) modeled
via a state space representation. The control objective is the avoidance
of a given set of states. We achieve this goal by verifying that certain
predicates, specified in terms of states, are always false. We model the
DES involved in physical systems with a class of controlled state machines,
respectively the formalism of the controlled Petri nets. A model of a railway
system illustrates the theoretical approach. Possible extensions of the
work are also discussed.
Keywords: Controlled Petri nets, discrete event systems,
railway systems.
IMPROVEMENT OF FSM SYNTHESIS USING MSI AND LSI CIRCUITS
1 Alexandru VALACHI
2 Radu SILION
3 Mihai TIMIȘ
Technical University Gh.AsachiIasi
Automatic Control and Computer Science Faculty
Computer Science & Engineering Department
1 avalachi@cs.tuiasi.ro
2 rsilion@cs.ruiasi.ro
3 mtimis@cs.tuiasi.ro
Abstract. Synthesis and implementation of FSM machines
will be more easy with complex circuits, like MSI and LSI circuits. As
the realized functions by these circuits are more complex, the combinational
logic network around the finite automata kernel is too simple,
[1],[2]. Usually, heuristics methods can be used for synthesis of
the additional combinational logic network, [3],[4]. It will be deducted
the transition matrix.
Keywords: Finite State Machine, Transition Matrix, Counter,
Logic Primitive, Medium Scale Integration, Lower Scale
Integration.
USING A MULTIAGENT SYSTEM FOR CONTROLLING A MANUFACTURING
CELL WITH A ROBOT
Leonard IURESCU
"Stefan cel Mare" University of Suceava
str.Universitatii nr.9, RO-720225 Suceava
iurescu@eed.usv.ro
Abstract. The aim of this paper is to present a solution
based on a Multiagent System for a flexible manufacturing cell. The system
is composed of three agents controlling a conveyor band, a stoking area,
and a robot arm that takes pieces from the conveyor and put them in the
stocking area. The manufacturing cell is supervised by video cameras connected
to a computer.
Keywords: intelligent agents, multiagent system, flexible
manufacturing cell, control algorithms
FAULT DIAGNOSIS OF WALL HANG BOLIERS USING A NEURO-FUZZY
MODEL
Alexandru BÂRLEANU
Technical University Gh. Asachi
Department of Computer Engineering
Blvd. D. Mangeron No. 53A, Iasi 7050, Romania
alexb@cs.tuiasi.ro
Dragos AROTARITEI
University of Medicine and Pharmacy Gr. T. Popa
Faculty of Biomedical Engineering
Faculty of Biomedical Engineering, Str. Universitatii No. 16, Iasi,
Romania
dragos_aro@yahoo.com
Abstract. A fault diagnosis system based on neuro-fuzzy
model is proposed. The wall hang boiler is modeled as a second order fuzzy
model in a dynamic configuration. The identification model methos is used
as basic tool for estimation of the parameters of neuro-fuzzy system. The
signals used in fault diagnosis need various preprocessing phases due to
many component tolerances, noises and environment disturbances. The experimental
results along with theoretical support are presented.
Keywords: neuro-fuzzy model, fault identification, wall
hang boilers, model based fault detection.
OVERVIEW ON MULTIMODAL BIOMETRIC SYSTEMS
Cătălin LUPU
"Stefan cel Mare" University of Suceava
str.Universitatii nr.9, RO-720225 Suceava
catalinlupu@seap.usv.ro
Abstract. In computer systems, there is an urgent need
for accurate authentication techniques to prevent unauthorized access.
Authentication is the process of confirming the correctness of the claimed
identity Many computers that store critical information are vulnerable
to unauthorized access because of weak authentication. In some cases, the
safety of the public can be at risk, such as in the case of a multi-billion
dollar passenger-screening system defenceless against terrorists with forged
security badges. Traditional authentication techniques such as the ubiquitous
username / password method are inadequate for personal identity since they
can only provide proof of possession and/or proof of knowledge. Only biometrics,
the authentication of individuals using biological identifiers, can offer
true proof of identity. Current research suggests that multimodal biometric
systems, those that use more than one biological identifier, can improve
the accuracy of biometric systems. This improvement in accuracy depends
on critical factors in design, implementation and security. This paper
explains each of these critical factors so the increase in accuracy observed
in current research can be achieved in real-world applications.
Keywords: biometric, identification, verification, recognition,
image acquisition, iris, fingerprint, face detection
KNOWLEDGE REPRESENTATION AND PROCESSING WITHIN AN INTELIGENT
SYSTEM FOR ASSISTED DOCUMENTATION AND RESEARCH REGARDING VEGETAL GENETICS
RESOURCES
Nicolae MORARIU
"Stefan cel Mare" University of Suceava
str.Universitatii nr.9, RO-720225 Suceava
nmorariu@eed.usv.ro
Abstract. The paper presents some aspects regarding the representation and processing of the data and knowledge within an information system for vegetal genetics, dedicated to the educational and research institutions in this field. The data and knowledge that are to be stored and processed in a documentation and assisted research system for vegetal genetics, refers resources of the vegetal genetics representing all the vegetal life forms: wild plants, varieties and local populations, lines, hybrids, weeds, improved forms, etc., all of these being subject to the genetic erosion phenomena, pathogenic elements aggression (phytopatology) and environment factors. The plant description (the phenotype), the genetic hereditary heritage (the genotype) as well as the results of the experiments and researches performed in the vegetal field are fed into the systems database within a predefined evolutionary self developing structure. Among the specific research problems in this field one can mention: population recognition and classification, plant disease diagnostication. The basis concept was inspired by the activity of Suceava Genebank - Romania.
Keywords: assisted research,, vegetal genetics, phenotype,
genotype, data and knowledge base, evolutionary structure, artificial
intelligence, pattern recognition.
USING NEUROSHELL NEURAL NETWORKS SIMULATOR FOR PREDICTION
PROBLEMS SOLVING
Sorin VLAD, Nicolae MORARIU
Ștefan cel Mare University of Suceava,
str. Universității, No 9, RO-720225, Romania
nmorariu@eed.usv.ro, sorinv@seap.usv.ro
Abstract. NeuroShell is one of the many neural networks
simulators available. In order to build a correct model to solve a specific
problem, some steps must be taken into considerations. The paper presents
an overview of the NeuroShell 2 program and some practical results obtained
for the chosen problems.
Keywords: neural networks, NeuroShell, simulation, prediction.
EFFICIENT ALGORITHMS FOR THE PLANNING OF SOME INTELLIGENT
VEHICLES USED FOR THE DETECTION AND EXTINCTION OF FIRE
Valeriu LUPU
"Stefan cel Mare" University of Suceava
str.Universitatii nr.9, RO-720225 Suceava
valeriul@seap.usv.ro
Abstract. This article presents efficient algorithms for
the planning of some vehicles that have the mission to detect and extinct
a fire. The machines work in a varied land, with hills and valleys, with
holes, cliffs, mined land and with other obstacles they may run into their
way (this obstacles are from the beginning in the studied land or they
can appear instantly), and whom the machine must avoid. The machines can
depart all at the same moment or they can leave one at a moment. The machines
can meet each other on their way but they must avoid the collisions by
communicating with the server. The machines are equipped with complex sensorial
systems like video cameras, transducers, proximity sensors (thermics, smoke,
gases, etc.), and neuro-fuzzy logic. There are used rapid sequential algorithms
of the Strassen and Winograd for the matrices multiplication (using parallel
calculus) and the Dijkstras algorithm to compute the optimum way between
two points using parallel programming.
Keywords: fire, extinction, intelligent machines, computer
vision, parallel programming.
DESIGN OF AN APPLICATION FOR CONTROL-ACCESS USING IRIS
RECOGNITION
Cătălin LUPU
"Stefan cel Mare" University of Suceava
str.Universitatii nr.9, RO-720225 Suceava
catalinlupu@seap.usv.ro
Abstract. Algorithms developed by the professor John Daugman from Cambridge University for recognizing persons by their iris patterns have now been tested in six field and laboratory trials, producing no false matches in several million comparison tests. The recognition principle is the failure of a test of statistical independence on iris phase structure encoded by multi-scale quadrature wavelets. The combinatorial complexity of this phase information across different persons spans about 249 degrees of freedom and generates a discrimination entropy of about 3.2 bits/mm2 over the iris, enabling real-time decisions about personal identity with extremely high confidence. The high confidence levels are important because they allow very large databases to be searched exhaustively (one-to-many .identification mode.) without making false matches, despite so many chances. Biometrics that lack this property can only survive one-to-one (.verification.) or few comparisons. This paper presents an efficient software product that recognizes a person by its iris pattern by searching in a large iris database.
Keywords: software product, iris, image, recognition.
NEURO-FUZZY LOGIC AND COMPLEX SENSORIAL SYSTEMS USED FOR THE PLANNING OF SOME INTELLIGENT VEHICLES
Valeriu LUPU
"Stefan cel Mare" University of Suceava
str.Universitatii nr.9, RO-720225 Suceava
valeriul@seap.usv.ro
Abstract. This article presents neuro-fuzzy logic and
complex sensorial systems used for the planning of some intelligent vehicles
that have the mission to transport persons with handicaps inside of a building
(or in a room), transport of the person with disabilities inside their
home without companion ,baggage transport in airports (without driver),
container transport in harbours, marrow places (without crew), transport
of dangerous waste products, transport of persons in winter conditions,
cleaning of waters, mountains (cleaning of the waters from oil waste),
detection and extinction of fire. The machines work in a varied land, with
hills and valleys, with holes, cliffs, mined land and with other obstacles
they may run into their way (this obstacles are from the beginning in the
studied land or they can appear instantly), and whom the machine must avoid.
The cars can leave all at one moment or individually. They can meet on
their way, but they must avoid the collisions by communicating between
them through the server. The machines are equipped with complex sensorial
systems like video cameras, transducers, proximity sensors, neuro-fuzzy
logic. Sequential rapid algorithms like Strassens and Winograds are
used for the matrices multiplication (by using parallel calculus) and Dijkstras
algorithm for the calculus of the optimum road between two points by using
parallel programming.
Keywords: transportation, neuro-fuzzy logic, parallel
programming, intelligent machines
MATHEMATICAL MODELS FOR A ROBOT ARM CONTROL SYSTEM
Ștefan-Gheorghe PENTIUC
Radu Daniel VATAVU
Remus Cătălin PRODAN
Leonard IURESCU
Tudor Ioan CERLINCĂ
Cristian IVĂNCESCU
"Stefan cel Mare" University of Suceava
str.Universitatii nr.9, RO-720225 Suceava
pentiuc@eed.usv.ro
Abstract. The control system for a robot arm is an ongoing
image processing project for which many enhancements have been done during
the last years. The paper introduces a new control system for a robot arm,
developed in the Electrical Engineering Department. The control is represented
by the robots movements of grabbing and moving objects located on an operational
table to user selected destination points. The enhancements involve robot
visual-based automatic calibration with respect to the working field axes
and scale, fast and robust object detection. A self-calibration algorithm,
based on video motion and color detection, is used in order to provide
a good conversion of the objects coordinates from the image space into
the working area coordinates system. The algorithm proves to be general
enough to be employed in other similar image processing based scenarios
where self-calibration might be useful
Keywords: robot control, image processing, automatic
self calibration, video motion detection, edge detection
AUTONOMOUS SYSTEM FOR SEARCHING AND ELIMINATING FIRE SOURCES
Cristian Andy TĂNASE
"Stefan cel Mare" University of Suceava
str.Universitatii nr.9, RO-720229 Suceava
andy_t@eed.usv.ro
Adrian GRAUR
"Stefan cel Mare" University of Suceava
str.Universitatii nr.9, RO-720229 Suceava
adriang@eed.usv.ro
Andrei LEONTE
ALCATEL SEL AG, Stuttgart, Germany
coriolan.leonte@alcatel.ro
Dorin PALANCIUC
IPDevel, Bd. Vasile Milea 1D, Bucuresti, Sect. 6.
dorin.palanciuc@ipdevel.ro
George HOLCA
SC. DotNet S.R.L. Str. Mărășești, Nr. 44, Bl. T1, Suceava
george@dotnet.ro
Abstract. This paper introduces the concept of an
autonomous machine designed to search a candle inside a maze. The shape
and dimensions of the maze, as well as the height of the candle are given.
The robot has three proximity sensors: left, right and front. Using these
sensors, the robot adjusts its position relative to the walls. The electric
compass and the sensors for reading the white tape delimiting the room
exits are used to determine the starting room. Once this phase is completed,
the robot follows a pre-established pattern and searches the candle in
every room. Once the candle is found the robot switches on an extinguishing
fan.
Keywords: robot, maze, candle, proximity, sensors.
ADAPTIVE SKIN COLOR DETECTION IN UNCONSTRAINED ENVIRONMENTS
USING 2D HISTOGRAM PARTITIONING
Radu Daniel VATAVU1 2, Laurent GRISONI2, Samuel
DEGRANDE2,
Christophe CHAILLOU2, Ștefan Gheorghe PENTIUC1
1"Stefan cel Mare" University of Suceava
str.Universitatii nr.9, RO-720225 Suceava
raduvro@yahoo.com, pentiuc@eed.usv.ro
2Laboratoire dInformatique Fondamentale de Lille
Universite des Sciences et Technologies de Lille
Laurent.Grisoni@lifl.fr,
Samuel.Degrande@lifl.fr,
Christophe.Chaillou@lifl.fr
Abstract. An adaptive approach for skin color detection
is presented based on image filtering with dynamically computed thresholds.
Two dimensional hue/saturation histograms are analysed and partitioned
into rectangular blocks. For each block, simple parameters are computed
such as: 1st order moments, standard deviations and simple ratios. A multilayered
perceptron is trained using a set of images containing both skin and non-skin
blocks. For each frame in a video sequence, each block is assigned a probability
of being skin and the parameters of the winning block, if any, are used
for image thresholding.
Keywords: skin color, two dimensional histogram, thresholding,
multilayered perceptron.
Advances in Electrical and Computer
Engineering
ISSN 1582-7445 - No 1 / 2005
DISTRIBUTION NETWORK
The issues of the journal are distributed in 30 University
Libraries from Belgium, France, Germany, Great Britain, Greece, Ireland,
Italy, Moldavia, Portugal, Romania, Russia, Ukraine, USA and Yugoslavia
Universitatea Tehnica
a Moldovei
Biblioteca universitara Bd. Stefan cel Mare 168 277012 CHISINAU Republica Moldova |
|
Laboratoire d'Informatique
Fondamentale de Lille
Bibliotheque Batiment M4 Cite Scientifique 59655 Villeneuve dAscq Cedex FRANCE |
University of Cernautzi
University Library Ul. Kocjubinskogo 2 274012 CERNOVCY UKRAINE |
L2EP E.C. Lille
Batiment P2 Cite Scientifique BP 48 59655 Villeneuve dAscq Cedex FRANCE |
The State Technical
University of Vinnitsa
University Library 95, Khmelnitska shore 286021 VINITSA UKRAINE |
KAHO Katholieke
Hogeschool Sint-Lieven Gent
Gebroeders Desmetstraat 1 B-9000 GENT BELGIUM |
Moscow Aviation
Institute
Library Volo Kolamskoye chosse 4 125 871 MOSCOW RUSSIA |
Universita degli
Studii di Catania
Dipartimento Elettrico Elettronico e Sistemistico V.le A.Doria, 6 95125 CATANIA ITALY |
Moscow Institute
of Radioengineering and Electronics
Institute Library Pr. Vernadskogo 78 117 454 MOSCOW V454 RUSSIA |
Instituto Superior
de Engenharia do Porto (ISEP)
Biblioteca (Library) Rua S. Tome 4200 PORTO PORTUGAL |
CITY UNIVERSITY
University Library Northampton Square LONDON EC1V 0HB GREAT BRITAIN |
Universite de Franche
Comte
Bibliotheque Universitaire 30, Avenue de lObservatoire 25030 BESANCON Cedex Belfort FRANCE |
University of Colorado
University Library COLORADO SPRING 1420 Austin Bluffs Pkwy CO 80907 USA |
Texas Cristian University
University Library BOX 32886, FORT WORTH 76 129 Texas USA |
National Technical
University of Athens
University Library ZOGRAPHON ATHENS 15773 GREECE |
University of Limerick
Limerick IRELAND |
Institut Arbeit und Technik
Gelsenkirchen GERMANY |
University of Novi Sad
Novi Sad YUGOSLAVIA |
Institute for Cybernetics
Kiev UKRAINE |
Universitatea "Politehnica" Bucuresti
Biblioteca universitara BUCURESTI |
Biblioteca Centrala Universitara
BUCURESTI |
Universitatea Tehnica
"Gh. Asachi" Iasi
Biblioteca universitara 6600 IASI |
Universitatea Tehnica
Cluj-Napoca
Biblioteca universitara 26, G.Baritiu RO-3400 CLUJ- NAPOCA |
Universitatea Tehnica
din Timisoara
Biblioteca universitara TIMISOARA |
Biblioteca Centrala
Universitara
IASI |
Universitatea din Oradea
Biblioteca universitara ORADEA |
Universitatea din Galati
Biblioteca universitara GALATI |
Universitatea din
Craiova
Biblioteca universitara CRAIOVA |
Biblioteca Centrala
Universitara
CLUJ - NAPOCA |
Advances
in Electrical and Computer Engineering
ISSN 1582-7445 - No 1 / 2005
Subscription and Paper Submission
Mail
prof. Adrian GRAUR
Faculty of Electrical Engineering
The "Stefan cel Mare" University of Suceava,
str.Universitatii nr.1
RO-5800 SUCEAVA
Romania
e-mail: adriang@eed.usv.ro