A Novel Method For Face Segmentation And Face Features Extraction Based On Voronoi
Diagram
Paper Number: 01
Abstract:
Recognition of human
faces out of many still images is a research field of fast increasing
interest. This paper presents a novel face detection algorithm for gray
intensity images. At first, face location and extraction has to be
performed in order to get the approximate if not the exact representation of
a given face in an image. Many approaches have been introduced; some deal
with colored images others with binary or gray-scaled ones and some with the
constraint to one face per image while others can process multiple instances
of faces within an image. Some of these approaches gave good results, however
with much price and burden to pay for computation. Thus, a fast and accurate
algorithm is still a field to be explored. Our proposed approach is based on
Voronoi Diagram (VD), which is a well known technique in
computational geometry. We use it to form decision whereby intensity ranges
will be generated based on vertices of the external boundary of
Delaunay triangulation. By performing this; our image
will yield segmented regions. A greedy search algorithm examines for a face
candidate focusing the action around elliptical like regions. VD is used now
in many fields, but mainly researchers focus on its use in
skelatization and generating Euclidean distances, and
here comes our contribution namely in exploiting the Voronoi
and its son (Delaunay) to generate features that
can be used as patterns for face recognition.
back to Accepted Papers Page
Print Protection Using A Diffused Background Watermark
Paper Number: 03
Abstract:
In this paper a new approach
to digital watermarking in printed documents for copyright protection is
described. The process is defined by using analytical techniques and
concepts borrowed from Cryptography. During watermarking the background
is generated using convolution of the watermark image (logo) with a source of
noise in a process termed 'diffusion'. The cover image (text) is
inserted into the foreground of the document using a simple additive process
termed 'confusion'. The watermark is subsequently recovered by removing
the foreground and then correlating with the original noise source. It
is demonstrated that this method is robust to a wide variety of attacks
including geometric attacks, drawing, crumpling and
print/scan attack. Experiments also show that the method is relatively
insensitive to lossy compression, making it well
suited to scanned electronic documents. The details of this method as well as
the experimental results are shown.
back to Accepted Papers Page
Dense Disparity Map : A Multiresolution Approach
Paper Number: 04
Abstract:
Stereo is an important area in computer vision. This paper presents a
efficient stereo matching algorithm based on correlation technique which produces
a dense disparity map by using a pyramid structure. The basic correlation
algorithm consists of computing the correlation scores for every point in the
image by taking a fixed window in the first image and a shifting a window in
the second image. The second window is moved in the second image along the epipolar line. For each point, the correlation score is
computed by using some criteria. Two pixels are said to be matched when a
maximum score is found. The most classic area-based methods used cross
correlation with a fixed window-size but this method didn�t give satisfying
results. In order to decrease a number of invalid points and to increase the
density of sparse disparity map, we propose in this paper a hierarchical
algorithm where the window�s size is fixed in order to perform the matching
at several levels of resolution. Both synthetic and real image tests have
been performed, and good results have been obtained.
back to Accepted Papers Page
A Cell Selection Policy For An Input Buffered Packet Switch
Paper Number: 06
Abstract:
Many Architectures of
Internet routers, ATM and Ethernet switches have been proposed and analyzed in
literature. Theoretically reliable and valid solutions have been developed
but not feasible in practice for commercial and technological reasons. Few
papers develop the implementation and simulation aspects. The objective of
this paper is the design of a packet switch with a minimum cost and minimum
hardware complexity. A Benes network as an
interconnection network combined with a simple cell selection policy
implemented by hardware has been adopted to reduce the hardware cost and
scheduling algorithm complexity. We propose and simulate an input-queuing
architecture using a hardware description language. Performances in terms of
throughput and cell loss are evaluated. The measures confirm that
performances of our switch are quiet similar to those of others switches, but
with less cost and less complexity.
back to Accepted Papers Page
A Realistic Plants� Modeling And Rendering With Lindenmayer Systems And
Constructive Solid Geometry
Paper Number: 10
Abstract:
Nowadays the modeling
of natural phenomena like mountains, snow, plants, and others, is a field in
a constant progress. The research is mainly developed in the modeling and the
visualization of the objects. The objectives of the researchers are the photo
realism of the image, the execution time, and the introduction of more and
more functionalities to the models. In this paper, we�ll present a brief
presentation of a new plants� simulator of the development of plants that
combines the biological modeling of plants with L-systems and the Schematic
modeling based on Constructive Solid Geometry (CSG). The rendering was
realized thanks to RENDERMAN of PIXAR.
back to Accepted Papers Page
Alsat-1 Images and Zooming Enhancement Using Super Resolution Method
Paper Number: 12
Abstract:
In this paper we
investigate the problem of how information contained in multiple, overlapping
Alsat-1 first Algerian microsatellite images of the
same scene may be combined to produce images of superior quality. In each
view, light emitted from the scene is projected onto the sensor array of the
Alsat-1 Imager, producing an irradiance pattern which is temporally
integrated, spatially sampled and quantized to generate an image. In
addition, the image is often blurred due to de-focus and/or motion. We
attempt to reverse these degradations in order to estimate a high resolution
representation of the intensities in the scene. The process of reconstructing
a high resolution image from several different images covering the same
region in the world is called Static Super Resolution. In this paper we
address also the problem of producing an enlarged picture from a given
digital image (zooming). This problem arise frequently whenever
an user wish to zoom in to get a better view of a given
picture.
back to Accepted Papers Page
An Arabic Character Recognition System in Using General Auto-Associative Memory Model
Paper Number: 13
Abstract:
This paper proposes a method
for the recognition of typewritten Arabic characters. Recognition has been
done in using a general auto-associative memory (GAM) based approach, using
improved chain code and invariants moment shape descriptor as the main
features. The role of GAM artificial neural network used as a classifier is
examined, investigated, and its performance is evaluated when trained and
tested with printed Arabic characters using structural features. The feature
vector, which is used to train the GAM recognition model, is the chain code
(direction and length), number of holes and or dots with their position, and
moments of segmented character. By using the new GAM concept called
supporting function, the model shows a storage capacity suitable for storing
all Arabic characters. A simple experiment was conducted to show the
effectiveness of the Sample Learning Algorithm (SLA) in the recognition of
noisy Arabic characters. The system developed has been tested using a set of
data and the result obtained has proved its efficiency and robustness.
back to Accepted Papers Page
A New Approach For Time Series Forecasting Based On Artificial Neural Network
Paper Number: 14
Abstract:
Traditional feedforward neural networks are static structures which
simply map input to output. Motivation from biological consideration, a
method for predicting a new artificial network weighting functions is
devised. Using these weighting functions predicated, time
series are derived. In the applications investigated, forecasting
country population, this method outperformed all standard method of neural
network prediction.
back to Accepted Papers Page
Arabic Phoneme Recognizer Based On Neural Networks
Paper Number: 15
Abstract:
An Arabic phoneme recognizer has been designed and implemented, the
categories of Standard Arabic phonemes were uesed.
According to that 6 sets of Arabic phonemes are defined. One special
Neural network is designed for each set. The
proposed network is a linear hierarchical network
which consists of two subnetworks. The first subnetwork is based on
Kohonen Self-Organizing Map. The second
subnetwork is based on
Learning Vector Quantization.
back to Accepted Papers Page
A New Chaining Blocks Mode For Ciphering
Paper Number: 17
Abstract:
Electronic data and
information exchange (including private and personal information, on line
commercial transaction, etc.) is the most exposed service to harmful attacks,
hence, the important security issue of the service. Security covers the
confidentiality, integrity, authentication and non-repudiation concerns.
Cryptography is known to be a basic tool for internet security. In this paper
we are interested in using cryptographic techniques to secure the electronic
exchanges to protect them from being attacked during their transmission on
the NET. We particularly emphasis on enciphering and our solution offer the
user three security levels: low, medium and high. More ever, the most
important contribution of this paper is that we enforce the enciphering techniques
used by introducing two new chaining blocks modes that allowed us to combine
two different enciphering algorithms at same time to encrypt the message.
back to Accepted Papers Page
Automatic Arabic Phoneme Segmentation
Paper Number: 18
Abstract:
In this paper we propose a new Arabic Phoneme Segmentaion Algorithm
using time domain analysis method. It basically
depends on rules of phonological arabic system. It basicly consists of three rules, one should use the three
rules just stated to begin the process of segmentation in Arabic. These rules
enable one to segment almost any utterance in Arabic correctly and easily.
back to Accepted Papers Page
A Layered Connector Model for A Complex Software System Design
Paper Number: 21
Abstract:
Despite the important role
they play in the definition of software architecture, connectors haven�t
received sufficient consideration across various research works in the field
of software architecture. The concept of connector is one of the most
important aspects that need to be well understood and clarified. Often,
architects put all their skills and focus all their mental effort in defining
the most important parts of the system called components and link them in a
way that satisfy functionality and requirement of the desired system.
Refining process, which takes the system from its high-level Abstract:
ion
state to an implementation state, is often applied to components. In fact,
often an architect is not concerned with connector refinement. At each step
of refinement process, architects exploit an interconnection technology to
link components rather than designing a new one. In such a way, connector
implementation is often scattered among various component, yielding a
non-reusable communication infrastructure. In this paper, we will try to
understand more accurately the concept of connector, which first may be
considered a self sufficient entities like
components, and then we will present various concepts needed to define the
basis for defining a connector model that may be embedded as part of an ADL
or any specification tools. Finally, we propose architecture of a connector
model.
back to Accepted Papers Page
Synchronous Machine
Steady-State Stability Assessment Using An Artificial Neural Network
Paper Number: 24
Abstract:
This paper proposes a neural network approach for steady-state stability assessment for a
single-machine system. The perceptron can forms two
decision regions separated by hyperplane. The
developed neural network is trained in the space of training patterns, this
space can be divided into a stable region and an unstable one by using the
training data. The boundary between both regions is described by a set of
hyperplanes in the pattern space. These
hyperplanes, which can be used to represent the
connection weights between neurons, are known as weight hyperplanes.
It is found that the proposed learning method converges much faster than the
back propagation momentum method.
back to Accepted Papers Page
Computer simulation for the atmosphere effect on the image quality of the thermal camera
Paper Number: 25
Abstract:
Thermal radiation is
attenuated in passage through terrestrial atmospheres mainly by the processes
of absorption and of scattering. Knowledge on the atmospheric transmission
and attenuation of the spectral radiation in the wavelengths between 2-15 mm
at the earth surface is required for many purposes. So, a computer simulation
demonstrates the atmosphere; aerosol and turbulance
effect on the output thermal image quality of the thermal camera has been
developed. The simulation program calculate the detection range that
the system able to detecte
the object as a function of signal to noise ratio (S/N). The results
show that the effect of aerosol effect is much more dominant than the
turbulence in the transmission of the IR radiation in atmosphere. At certain
SNR the range for clear atmosphere is larger than that for dense atmosphere.
back to Accepted Papers Page
Image Compression Using Genetic Algorithm
Paper Number: 31
Abstract:
The principal objective
of this research is an adoption of the Genetic Algorithm (GA) for studying it
firstly, and to stop over the facilities which are introduced from the
genetic algorithm. The candidate field for applying the facilities of the
genetic algorithm is the Image compression field, because researchers took
great interest in this field in the recent years. This research uses the
facilities of the genetic algorithm for the enhancement of the performance of
one of the popular compression method, Vector Quantization (VQ) method is
selected in this work. After studying this method, new proposed algorithm for
mixing the (GA) with this method was constructed and then the required
programs for testing this algorithm was written. A good
enhancement was recorded for the performance of the (VQ) method when mixed
with the (GA). The proposed algorithm was tested by applying it on some image
data files. Some fidelity measures are calculated to evaluate the performance
of the new proposed algorithm. All programs were written by using Visual C++
language (version 6.0) and Visual BASIC language
(version 6.0) and these programs were executed on the Pentium III (500 MHz)
personal computer.
back to Accepted Papers Page
Oil Height Determination From Capacitance Tomography Measurements Using Neural Network
Paper Number: 40
Abstract:
This paper presents a �direct�
method to gas-oil interface level determination using an artificial neural
network approach based on Electrical Capacitance Tomography (ECT)
measurements. �Direct� here means that the gas-oil interface levels are
obtained directly from the ECT measurements without recourse to image
reconstruction. The preliminary work models a separation tank that is
filled with gas and oil. An ECT system, attached around the tank is
used to obtain ECT measurements. Sets of ECT measurements together with
their corresponding oil heights are fed into a Multi-Layer
Perceptron (MLP) neural network system for training
processes. After being trained, the MLP is tested by giving it sets of
independent ECT measurements. The results show that �direct� gas-oil
interface level measurement from ECT data is feasible with the use of a
neural network system.
back to Accepted Papers Page
On the use of Multi
Layer Neural Networks in Automatic Information Retrieval
Paper Number: 41
Abstract:
A new idea is developed
here, which is a particular approach to the use of neural networks in
information retrieval (storing and searching large bibliographic databases).
The idea is theoretical and based on the manipulation of statistical
information derived from document contents in order to build their hidden
layer representations. During the learning process,
the document is entered to the network and processed sentence
after sentence. Each sentence form a tree of
connections through multiple layers to the corresponding document in the
output layer. Each processed document will add vocabulary and connections to
the network and thus, forms a sub-network representation of its content.
During the search process, words included in the query will activate the
corresponding units in the first layer, which in their turns will activate
other units in the upper layers until they will fire one or more output
documents in the last layer. Back-propagation might be used in cases of
firing irrelevant documents.
back to Accepted Papers Page
Client/Server Connection Using Java Language
Paper Number: 42
Abstract:
This Paper presents the
idea of motivation for the client /server paradigm arises from the problem of
rendezvous. The client/server model solves the rendezvous problem by
asserting that in any pair of communicating applications one side must start
execution and wait for the other side to contact it. In this article we
present the natural communication, the communication types, and application
interface on which a server and client contact each other. We implemented it
using java language programming.
back to Accepted Papers Page
Improvement of Land Cover Change Detection Accuracy by Fuzzy Aggregation Operators
Paper Number: 52
Abstract:
Change detection is a principal
task in monitoring and wide area surveillance. Usually, it involves the
comparison of remotely sensed images of the same area acquired on different
dates. The data are handled either by comparative analysis of
independently-produced classifications of data, or by an alternative approach
based on simultaneous analysis using multitemporal
classification of data. since the difference of
performance of these two approaches is well known, we propose the combination
of their outputs to achieve the best possible performance. in
this paper, both approaches utilize the same fuzzy classifier while the
combination is carried out by using various fuzzy aggregation operators.
Experiments on bitemporal SPOT data showed that the
combination can significantly improve the change detection accuracy when
using an appropriate fuzzy aggregation operator.
back to Accepted Papers Page
Time Series Forecasting with Fuzzy Genetic Aigorithms
Paper Number: 53
Abstract:
ln fuis paper we describe a time series forecasting method based on a fuzzy systems.
ln particular til11e series
exhibiting non-linear or chaotic behaviors are selected for forecasting. A
range of l11ethodologies based on a set of fuzzy predictor's permit the
prediction of the sal11e time series, but with a different
fuzzy partitions. Each fuzzy predictor is developed by the means of a
real genetic algorithm. The combination of the fuzzy predictors will reduce
the prediction error. The prediction l11ethod is tested by using Mackay-Glass
til11e series data.
back to Accepted Papers Page
MCLA: A Multi-label Classification Learning Algorithm
Paper Number: 56
Abstract:
Association rule mining
and classification are important tasks in data mining. Using association
rules has proved to be an effective and accurate approach for classification.
In this paper, a new classification based on association rules technique is
presented. It uses a method of quickly finding frequent itemsets.
The proposed technique produces rules with multi-label, and adjusts the
class-label ranking for some of the rules to reflect changes that occur after
evaluating other higher-ranked rules. The results obtained after
experimenting with several different datasets, showed that the presented
method is an accurate and effective classification technique. Furthermore,
the classifiers generated are highly competitive to those produced by
decision trees and other associative classification techniques with regards
to classification accuracy.
back to Accepted Papers Page
Acoustic Characterization Of Heart Valves Activity By Using The Smoothed-Pseudo Wigner-Ville
Distribution
Paper Number: 58
Abstract:
The Phonocardiogram
(PCG) signal, which is the heart sounds temporal tracing, is recorded by using
a Computer�based acquisition system that we had developed. Indeed, the
software part of the acquisition system, written in MATLAB language, drives a
Personal Computer to allow collection of the heart sounds samples in
16 bits format. The acquisition process is carried out through a
microphone at different sampling frequencies. The time�frequency analysis of
the PCG signal is carried out by means if transformed Wigner�Ville
Distributions (WVDs). Hence, the Smoothed�Pseudo Wigner�Ville Distribution (SPWVD) is selected over the
transformed approaches of the WVDs yielding to an
adequate Time�Frequency Representation (TFR). Thus, the SPWVD is applied to
PCG signals of a normal subject acquired from its aortic, tricuspid, mitral and pulmonic
auscultation areas over the chest wall. Therefore, the nature of normal heart
sounds is characterized in systole and diastole phases. The spectral activity
of each heart valve, namely; the mitral, the
tricuspid, the aortic and the pulmonic heart
valves, is characterized in the time�frequency domain. A discussion in
relation to the auscultation area of each PCG signal is then presented
allowing acoustic characterization of each myocardial orifice.
back to Accepted Papers Page
Retrieving Adaptable cases into an associative memory
Paper Number: 60
Abstract:
Case-based reasoning is
a reasoning method resolving new problems by reusing past episodes of resolution
called cases. During the reusing process, solution of old case could be
adapted to fill the constraints of the new problem. The success of adaptation
step depends on the quality of remembered case: �is it easy to adapt?�
However, the recall process is generally guided by similarity. Most similar
cases are not usually the easiest ones to adapt. It will be more accurate to
guide our retrieval by an adaptability criterion. The main idea of this paper
is to perform a retrieval guided by adaptability into an associative memory
called �Case Retrieval Nets� (CRN). This memory model was initially conceived
for retrieving cases by similarity with high performances. We aim to benefit
from these performances to support the adaptability criterion. Which leads to
two advantages: look ahead to adaptation stage without incurring the full
cost of adaptation and secondly transport this knowledge rapidly throughout
the CRN.
back to Accepted Papers Page
A new approach for synchronization in a multimedia scenario
Paper Number: 62
Abstract:
The objective of this
work is to contribute to the domain of edition and presentation of multimedia
scenarios, and particularly in the temporal aspects of scenarios by
considering a p-temporal Petri net for the specification of the multimedia
scenarios. Our objective is to answer to the real user (final users, authors)
requirements. The user should have the possibility to specify temporal
synchronization, interactions, and the possibility to present scenarios. To
achieve these objectives, we define a representation model based on the
following concepts: the temporal and causal relations between multimedia
data, - the user interactions, - the representation of unknown durations, and
the formal specification of scenarios based on p-temporal Petri nets that are
useful to present and simulate the scenarios.
back to Accepted Papers Page
Towards consistency analysis of UML dynamic diagrams
Paper Number: 63
Abstract:
This paper presents a
methodology for verification of consistencies between state and sequence
diagrams of an UML model. Our approach consists first of translating
UML State Diagrams into Interval Timed Petri Nets (ITPN) which time intervals
model well generation and dispatching delays of events. The method deals also
with composite states and most of pseudo-states including the history ones.
Then sequence diagrams are mapped into dispatched events paths adorned with
time tags. The defined semantic consistency requires theses paths to be
subsequences of the reachability graph of the whole
Petri net.
back to Accepted Papers Page
Neural Network In Corner Detection Of Chain Code Series
Paper Number: 65
Abstract:
This paper presents a
Neural Network Classifier to be implemented in corner detection of chain code
series. The classifier directly uses chain code which is derived using
Freeman chain code as training, testing and validation set. The steps of
developing Neural Network Classifier are included in this paper. Comparison
results between Neural Network Classifier corner detection and other
computational corner detection are presented to show the reliability of the
proposed classifier. This paper ends with the discussions on the
implementation of proposed neural network in corner detection of chain code
series. Experimental results have shown that the proposed network has good
robustness and detection performance. This makes this method a great choice
for machine vision.
back to Accepted Papers Page
Metrics For Object Oriented Design (MOOD) To Assess Java Programs
Paper Number: 68
Abstract:
MOOD metrics are well
known metrics used to measure some characteristics of the Object-Oriented
programs. In this research, a system, based on the MOOD, has been developed
to evaluate and grade Java programs. The interval of each MOOD metrics has
been adapted, based on experimental results, to be fit in the evaluation of
Java Programs. Also, a weight factor has been introduced to reflect the
importance of each characteristic. The system has been tested with many
different programs that vary in their complexities and functionalities. Also,
the metrics have been tested in the evaluation of student programs in the
University of
Jordan and grading them. For
all cases, the system shows good results.
back to Accepted Papers Page
Novel Computer Network Approach for Electricity Tariff Management Systems
Paper Number: 72
Abstract:
Conserving electrical
energy has recently become a distinctive element of our everyday life.
Electrical energy consumers in the residential & commercial sectors are
particularly sensitive to price increases. Often, policies for consumption
reduction relying entirely on voluntary response by consumers do not produce
much effect. By contrast, policies that offer financial incentives to
consumers via tariff reduction are highly successful. It is the purpose of
the present paper outlines a novel computer network approach for electricity
tariff management systems. The present system is given the acronym ODETNET:
software Driver for Electricity Tariff management system based on
computer NETworking techniques. ODETNET
methodology is based on using the power grid network itself for computer
networking. ODETNET prototype automatically scans readings from
kilowatt-hour (kWh) meters. These automated meter readings are directly
inputted to a network of computers. This is instead of the normal manual
situation where the human observer notes visually the mechanical registers of
the kWh meter. The innovative contribution of ODETNET results from the fact
that interconnection of computers in ODETNET is via the electrical power grid
network itself. ODETNET prototype experiments were successfully implemented,
with the full cooperation of the Electricity Ministry. Both hardware and
software aspects of ODETNET system prototype are discussed.
back to Accepted Papers Page
Forest Fire prediction. The GIS Simulation Model
Paper Number: 74
Abstract:
The mastery of
strategies of forests fire-fighting passes compulsorily by deep
knowledge of forest fire
phenomenon. One of the efficient means of apprehension
of the forest fire, is to dispose of a tool that is capable to inform
us about the behavior of fire before its apparition according to the given
climatic conditions. In this context, the simulation remains an effective
tool for the prediction of the fire behavior.
It permits to determine with a relative confidence degree, susceptible zones
to be ravaged by the fire during a determined
period. The objective of this work is the
study of mechanisms of forest fire progression by the elaboration of an
automatic tool capable to pattern suitably a fire forest, its parameters, its
propagation and its behavior in a given region. Through out this study, it
will make conspicuous the considerable
property (perhaps unavoidable) the Geographical information Systems
(GIS), in combination with techniques of simulation in the apprehension of a
problematical " fires forest " [DAG 94]. This resides in the power
of the GIS to modelling all phenomenon
presenting a geographical character. The interest that presents such a survey
for operators in charge of the management of a forest fire (Fireman
department, services of forests, local collectivities, etc.) is double. It
permits to define to the long term a homogeneous and coherent politics of
forest fire prevention, whereas the system permits to verify the adequacy of
amenities and presents infrastructures of wrestling against the fire with the
reality of a disaster [DAG 97]. It also allows to determine,
in the perspective of the beginning of a fire, means to put in work for a
coordination of intervention teams and a strategy that remains efficient, of
wrestling against the fire progression.
back to Accepted Papers Page
Speech Denoising using Wavelet Thresholding
Paper Number: 80
Abstract:
During the past decade ,Wavelet Transform (WT) have been applied to
various research areas in statistical signal processing, their application
include speech and image denoising, compression, detection
and pattern recognition. In this paper we analyze the performance of two
thresholding methods for denoising
of speech signal in wavelet domain. The first method uses the Automatic Soft
thresholding criteria for the removal of background noise
form the speech signal and the second method removes the noise using the
Automatic Hard thresholding. Our Practical analysis
results are based on the application of Wavelet based hard and soft
thresholding for the denoising
of speech signal corrupted due to background and scaled white noise. More
specifically we prove that using wavelet based techniques, our signals can
achieve higher signal to noise ratio (SNR), better peak signal to noise ratio
(PSNR),and reduced normalized root mean square error(NRMSE) while the
retained signal energy is 99.9885% and the resulting denoised
signal are generally much smoother.
back to Accepted Papers Page
Tcp Contextualization And Chaos Control In Wired Cum Wireless Environment
Paper Number: 82
Abstract:
Technological advances
in mobile devices has created a new network environment namely the wired cum
wireless environment. Such network has imposed great challenges for network
support. Firstly, while wireless networks provide mobility and convenience to
the mobile users, the quality of service (QoS) of
today�s wireless networks is still lagging compared to wired network thus
degradation the overall network�s performance. Secondly, transport control
protocol (TCP), which is used extensively in wired network, has
extend its usage in wireless network. Though TCP is
optimized for performance on wired networks, its misinterpretation of
wireless errors as network significantly degrade the network performance. In
order to overcome these predicaments we propose a revolutionary control
system that will complement TCP in stabilizing the network and thus increase
network performance. Based on Cynefin framework, we
contextualized TCP control management and focus on the control management at
the border of complex and chaos domain that is at the edge of chaos. We will
apply bifurcation control by delaying the occurrence of bifurcations in the
system. This in return will stabilize the behavioral pattern of TCP traffic
flow in wired cum wireless network environment.
back to Accepted Papers Page
The IP ability to disable the DDoS accessibility
Paper Number: 85
Abstract:
The most popular
methods used in the Distributed Denial of Service (DDoS)
attack and their defensive mechanisms (such as IP traceback
techniques) are discussed in this paper. A new solution is suggested to
prevent flooding Web servers. Readers will find a step by step explanation of
the suggested solution to protect sites from DDoS
attacks.
back to Accepted Papers Page
Construction Method of Audio Signal for Secrete Communication (CASIHO)
Paper Number: 88
Abstract:
However hidden messages
techniques has become more difficult, but methods for detecting information
hiding and analysis have become increasingly more sophisticated for
high-resolution digital images and digital audio carriers, so that the search
for new method that support the hiding information techniques be the only way
to making the analyzer task be as possible be hard. This paper describes a
new approach to construction-hidden messages in digital audio. The approach
uses a linear transform to re-build and change the type of secret information
with new statistical features, than be entered as input that embedded to the
audio carrier by using law bit encoding.
back to Accepted Papers Page
Analysis of Density Effect in Probabilistic Flooding in Mobile Ad Hoc Networks
Paper Number: 89
Abstract:
Broadcasting is a
fundamental and effective data dissemination mechanism, which has several applications
such as route discovery, address resolution, as well as many other network
services. While data broadcasting has many advantages, it induces some
difficulties known as broadcast storm problems, which are natural
consequences of redundant retransmission, collision, and contention. Data
broadcasting has traditionally been based on the flooding protocol, which
floods the network with high number of packet rebroadcasts until the desired
routs are discovered [6]. Density is the number of network nodes per unit
area for a given transmission range. This paper investigates the effect of
density with different speeds on the behaviour of
probabilistic flooding in mobile ad hoc networks (MANETs).
This investigation is conducted through extensive simulation on a network of
size 25 to 100 nodes. The obtained results reveal that the effect of density
with different speeds has a critical impact on the levels of rebroadcast and
reachability achieved by probabilistic flooding.
back to Accepted Papers Page
Lane Following for Marked Road
Paper Number: 90
Abstract:
A low-cost following
lane boundaries algorithm is described in this paper. The algorithm is destined
to painted road with slow curvature. The basic idea proposed in our approach
is that complete processing of each image can be avoided using the knowledge
of the lane markings position in the previous ones. The markings detection is
obtained using Radon transform that exploits the markings brilliance
relatively to the road surface. The experimental tests proved the robustness
of this approach even in shadows presence. The originality of our approach
compared to those using the Hough Transform is that it does not require any
tresholding step and edge detection operator.
Consequently, the proposed method is proved to be much faster.
back to Accepted Papers Page
Optimized Trellis Coded Vector Quantization For Encoding Memoryless Sources
Paper Number: 91
Abstract:
Abstract:
: Trellis coded
quantization, both scalar and vector, improves upon traditional trellis
encoded systems by labelling the trellis branches
with entire subsets rather than with individual reproduction levels. Trellis
Coded Vector Quantization (TCVQ) was introduced as an effective
low-complexity source coding technique which achieves rate-distortion
performance superior to many conventional optimal quantizers.
In this paper, we describe the design of TCVQ encoding system and its
improvement by an optimization procedure which was developed for TCVQ
codebook conception. We will show that, for a fixed rate, TCVQ encoding with
optimized codebook yields better performance than the non optimized encoder.
We will also show the superiority of the TCVQ scheme over conventional
optimal source coding techniques.
back to Accepted Papers Page
An Object Oriented Library Of Parallel Graph Algorithms
Paper Number: 92
Abstract:
Parallel computing is
about 20 years old, since then, parallel computing has solved complex
problems and high-performance applications. Distributed systems consisting of
powerful workstations and high-speed interconnection networks are economical
alternative to special purpose supercomputers. Networks of workstations that
are comprised of independent, general purposes computers are available
everywhere and have much unused computing power to be exploited. Utilizing
these resources for parallel computing is becoming a good idea with the
existence of many developed tools and systems that make these workstations
appears as a single virtual machine. Heterogeneity, high latency
communication, fault tolerance and dynamic load balancing are issues that
should be addressed in these systems to exploit parallelism. Therefore,
developing parallel programs in such systems requires expertise not only in
parallel programming issues but also in distributed computing. Dynamic
distributed parallelism is achieved by developing portable libraries that
provide support for these issues and liberate users from the complexities of
distributed systems and parallel programming, thus the programmer have only
to learn the interface to the library instead of learning parallel
programming. We have chosen to do this by implementing a library of parallel
graph algorithms. Doing so addresses both software design and development
issues, and hardware and performance issues like allocation of computations
to workstations, communication latency, the number
of workstations in the system, granularity of computation, and load
balancing. We can show the efficiency of a library either by comparing the
execution time required for the same problem as a sequential program for the
problem using our library. We have also chosen a second approach and solved
large instances of problems which could not be solved before on a single
computer for reasons of speed or memory constraints.
back to Accepted Papers Page
Towards a Multi-representation Ontology-Based Information Systems Mediation
Paper Number: 97
Abstract:
Recent research shows that ontologies
are a prominent tool for the semantic interoperability of information
sources. Unfortunately, semantic interoperability is not easy to achieve, as
related knowledge is most likely to be described in different terms using
different assumptions and different data structures. In this paper we provide
an interoperability solution based on semantic mediation permitting to define
a system of sharing, retrieving and transparent access to the heterogeneous
information sources. The proposed approach is based on the utilization of a
multi-representation ontology specified in description logics language
extended by stamping mechanism. This solution provides a structured and
shared vocabulary of several domains, serving as support to a semantic
mediator charged to localize the necessary sources to answer a request,
analyze this request for distributed extraction of data and recompose the
results in a user answer.
back to Accepted Papers Page
Concealment of Information in Digital Image
Paper Number: 98
Abstract:
Transmission
secret information provides an interesting problem space for investigating
information hiding. In this paper we present a new hiding system �Single and
Double Hiding�, to embed text file in digital image and all will be hide in
cover image or to hide text file in cover image only. The cover image could
be one of different image file format, such as (BMP, JPEG, GIF, and GIF
animation). The proposed system used cryptography, compression, and image
filtering to recover the weakness in the known methods. The system used three
new invented methods for hiding, which are hiding in edge points, direct
hiding, and hiding with reference images, after hiding process the proposed
system submitted into various kinds of attack to improve its strength.
Finally a comparison between (stego and cover)
image will be presented through using the histogram and the quantify
measures, and the resulting images from hiding didn�t draw any suspicion.
back to Accepted Papers Page
Towards a Methodology for Modelling Combinatorial Search Problems with Constraints
Paper Number: 99
Abstract:
Constraint programming CP is a programming paradigm especially
suited for solving complex combinatorial problems. In this paper, we address
the issue of the methodologies for developing constraint-based programs.
Modeling with Constraints-based Systems has been indirectly tackled through
the improvement of constraints expressiveness. However, the assumption that
constraints are declarative enough to allow a natural approach to modeling
proves itself to be unrealistic. The basic idea of the proposed
constraint-based programming methodology is the identification of
problem/specification families and of efficient corresponding programs. We
show that the synthesis of Constraint programs from specifications can be
automated. In order to enable (or facilitate) automated synthesis, inputs
ought to be formal though it would then be more adequate to say that the
inputs are programs and that synthesis is compilation.
back to Accepted Papers Page
Using Fuzzy AODV Routing Protocol in Ad-hoc Networks
Paper Number: 102
Abstract:
Ad hoc On-Demand Distance Vector (AODV) routing
protocol has been and continue to be a very active and
fruitful research area since it was introduced for the ad hoc wireless
networking. Ever since AODV uses static value for its timeout parameters,
like Active Route Timeout (ART) which states the time that the route can stay
active in the routing table. Timeouts may be more accurately determined
dynamically vi a measurement, instead of using a statically configured value.
To accomplish this, we use fuzzy logic system to obtain dynamic values for
ART depending on situation of transmitter node. The analysis shows that the
proposed design method is quite efficient and superior to conventional design
method with respect to data packet delivery ratio, routing overhead and
average end-to-end delay.
back to Accepted Papers Page
A New Word And Diacritic Segmentation Technique For Arabic Handwriting
Paper Number: 103
Abstract:
Failing to segment a handwritten text correctly will
definitely result in poor recognition, no matter how well the following and
previous stages are designed. We can say that a considerable share of
recognition errors is attributed to the segmentation stage. In this paper a
new algorithm for segmenting handwritten Arabic text-line into words and
diacritics is proposed. In our approach we rely on using fill-in function,
stroke thickness, text line height in segmenting the words/
subwords and diacritics of Arabic handwriting. We have
developed a simple procedure for stroke thickness estimation. In addition to
segmentation, our algorithm performs partial recognition as it can recognize
the Arabic character Alif and some diacritics
during the segmentation process A number of experiments were conducted and
higher accuracy was obtained.
back to Accepted Papers Page
Artificial Neural Network
Model for Prediction Solar Radiation Data: Application for Sizing Stand-alone
Photovoltaic Power System
Paper Number: 104
Abstract:
Abstract:
The prediction of daily global solar radiation
data is very important for many solar applications,
possible application can be found in meteorology, renewable energy and solar
conversion energy. In this paper, we investigate using Radial Basis Function
(RBF) networks in order to find a model for daily global solar radiation data
from sunshine duration and air temperature. This methodology is considered
suitable for prediction time series. Using the database of daily sunshine
duration, air temperature and global solar radiation data corresponding to
Typical Reference Year (TRY). A RBF model has been trained based on 300 known
data from TRY, in this way, the network was trained to accept and even handle
a number of unusual cases. Known data were subsequently used to investigate
the accuracy of prediction. Subsequently, the unknown validation data set
produced very accurate estimation, with the mean relative error (MRE) not
exceed 1.5% between the actual and predicted data, also the correlation
coefficient obtained for the validation data set is 98.9%, these results
indicates that the proposed model can successfully be used for prediction and
modeling of daily global solar radiation data from sunshine duration and air
temperature. An application for sizing of stand-alone PV system has been
presented in this paper in order to shows the importance of this modeling.
back to Accepted Papers Page
Texture Discrimination Using OCON and Multilayer Neural Networks
Paper Number: 105
Abstract:
Texture analysis is an important and useful area of study in
machine vision and image processing. Texture analysis applications have been
utilized in a variety of image processing fields such as: medical image
processing, remote sensing, document processing, etc. One immediate
application of texture analysis in image processing is the recognition of an
image using texture properties (i.e texture
classification). This work aims to investigate the use of OCON
(One-Classes-One-Net) neural networks for the classification of
natural-textured images (from Brodatz album).
Therewithal, attempt to draw a conclusion about the best feature-set
(recommended) that could be used to discriminate texture images and improve
the overall performance. To do this, four different sets of features
(statistical, statistical with radial/angular, spectral) are used to train
OCON based neural networks (Kohonen, and PCA) for
texture classification. Further, the recommended feature-set is used in
training the neural networks presented in this work to classify two
composite-texture -images (one at a time). Then, the
networks performances were tested on the entire composite-texture-image and
their classification abilities were compared visually. During this work,
weight initialization strategy for Kohonen is
developed to improve the convergence time and overcome the isolation problem.
These developed techniques of weight initialization leads to better
clustering operation.
back to Accepted Papers Page
PalmPrint Authentication System (PPAS)
Paper Number: 114
Abstract:
Palmprint authentication is an important complement of Biometric
authentication, this paper explore the method for palmprint
verification depending on the three principle lines that founded in the palm
of the hand. The principle lines of a palmprint
have a remarkable advantage of authentication because these lines have two
important properties (uniqueness, and stability),
from these two important characteristics it became easier to recognize one
person from another depending on the features of each person. It is important
to extract those features from the principle lines of a palm by using one of
features extraction methods, where the moments is a method that was used to
extract those features. Depending on these features an authentication system
depending on a palmprint known as PPAS was
designed, implemented, and tested.
back to Accepted Papers Page
Distributed Computing with Bluetooth
Paper Number: 115
Abstract:
This paper will analyze the capability of Bluetooth
network for distributed computing between its devices, by thoroughly
analyzing the Bluetooth Specification. Researching and comparing the various systems
that Bluetooth runs on. The platform and OS independence claims will
also be checked.
back to Accepted Papers Page
A development Process of Enterprise Portal based on the Web Services
Paper Number: 118
Abstract:
The concept of enterprise portal (EP) has exploded since
the success of the public portals like Yahoo! and MSN. The portal provides an
easy access to internal and external applications using a browser-based interface.
But, before the development of a portal, it is necessary that the enterprise
provides a homogeneous environment offering the interoperability between the
portal and the existing applications. In this context, the Web services are
the best solution to expose the various functionalities of the different
applications. They have a motivate key, which is the interoperability. This
paper presents a development process of enterprise portal based on the Web
services architecture. This process describes how to develop the Web services
by re-using the existing applications and how to build a portal that
integrates and aggregates these services and other external ones from other
portals.
back to Accepted Papers Page
Enhanced Multi-Signature Scheme
Paper Number: 120
Abstract:
An Enhanced Multi-Signature Scheme, EMSS is suggested in
this paper for group-users applications. The algorithm of this method employs
a modified RSA technique, which represents the signature for a classified
data as a separate value taken at the end together with the enciphered
message. It is signified by a short and fixed length signature for a message
regardless of its length, which comes from repeated signing of the message
segments. A serial number is included in the initial stage for bot message encryption and signing, in order to increase
the security and prevent the birthday attack. A major feature of this method
is to put the modulo of all users in order aiming to
solve the re-blocking problem.
back to Accepted Papers Page
Full-wave Analysis of sensitivities of microstrip open-end discontinuity
using the ameliorated moment method
Paper Number: 121
Abstract:
In this paper, we
carry out a dynamic study of analysis of the sensitivities of planar microstrip structures using the ameliorated moment
method. The shape sensitivities are obtained by using a new potential
integral equation for the derivative of the surface currents
with respect to geometrical parameter of
microstrip. This new integral equation is solved together
with the original integral equation with the ameliorated moment method. From
the geometrical derivative of the surface currents, geometrical derivatives
of the S parameters are obtained.
back to Accepted Papers Page
A Swarm Algorithm for
Medical Image Registration Based on the Maximization of Mutual Information information images
Paper Number: 122
Abstract:
In this paper, we propose an algorithm of particle swarm
optimization to solve the problem of medical image registration. This algorithm
combines the robustness of entropy based measures and the search power of the
particle swarm optimization. The main idea is to find the best non linear
transformation that superimposes two medical images by maximizing the mutual
information value through the particle swarm optimization. We show that this
algorithm, besides its simplicity, provides a robust and efficient way to
rigidly register medical images in various situations. This includes for
instance multimodal registration of CT and MR images as well as Scanner and
MR images.
back to Accepted Papers Page
The Intranet and its
impact in the hold of decisions at the level of enterprises
Paper Number: 128
Abstract:
Principal objective of the introduction of a computerized
information system using the new technologies of information and the
communication (NTIC) by the Internet and the intranet are: improvement of the
quality of the hold in charge of problems, assessment of service activities,
restraint of the administrative management, restraint of the continuous
increase of management costs and facilitated the hold of decisions. The
management of the enormous mass of information with the conventional means:
better organization, good administration and qualified and sufficient staff,
drag a difficult and complex management: not of control serious of management
and men, not of transparency in activities and bad administrative management.
Solutions susceptible aim to optimize the use rational of the enormous
material, human and financial resources of enterprise to satisfy our
objectives and to introduce the techniques most modern of treatments of
information by the setting up of a leading diagram of the enterprise.
contributions of the computer tool by the new technologies
of information and communication �NTIC� in the domain of the management are:
better organization of activities, considerable gain in time, minimum of
expense, better exploitation of information in real time and to improve
methods, means of storage, treatments of data and holds of decisions
efficiently. In the setting of this communication we are going to approach
aspects of the Intranet of enterprises in the setting of the decision hole.
back to Accepted Papers Page
Genetic Algorithms Application to the Standard Arabie Vocalic Recognition
Paper Number: 130
Abstract:
The aim ofthis work is to apply
Genetic Aigorithms at Standard Arab vowels Automatic
Recognition. For tha�, wc
have analyzed a corpus composed of several
sent�nces, registering by only one speaker, using Linear
Predicting Coding method. It results a set of discriminate parameters
vectors, which serves as reference for the Parameterization Generic
Modelisation. The adaptation or Fitness function of the
Genetic Model has been dermite. The Genetic
Aigorithms which search to maximize the inverse function
of the Fitness one, allow as to obtain interesting
results.
back to Accepted Papers Page
Arabic Connected
Digits Recognition Using The Probabilistic Approach Of Hidden Markov Models
Paper Number: 138
Abstract:
Although a great deal of effort has gone into studying large
vocabulary speech recognition problems, there remains a number of interesting
problems which do not require the complexity of these large systems. One such
problem is connected digit recognition, which has many practical
applications. Connected digit recognition is also an interesting problem for
another reason, namely that is one in which whole word training patterns are
applicable as the basic speech recognition unit. Thus we can bring all the
fundamental speech recognition technology associated with whole word
recognition to solve this problem. Hidden Markov models HMMs
are probabilistic approach and widely used in most world speech recognition
systems from small to very large vocabularies. Our paper is focused on how to
apply these probabilistic models in a connected arabic
digit (0-9) recognition task using moderate size databases. Results obtained
are satisfactory and would encourage using our system in more sophisticated
tasks.
back to Accepted Papers Page
A New Direction - Forecasting Cryptographic Keys
Paper Number: 141
Abstract:
Internet is the most hazardous environment for data
security. Everybody may have the chance to unveil anybody�s secret messages when
interested if one is given the skill and resources, no matter how strong
security system is used. With the great increase in the number of
applications present on the internet and the increased number of
cyper-customer, it is highly profitable to predict that a
security attack on messages is likely to occur within certain time,
compromising information or encryption keys. This paper presents a new
direction in security that enables internet users to forecast an attack
against a secret key. This technique suggests an algorithm that includes
vital and extremely important factors. In order to perform this task, the
lifetime of the of the used key, the frequency of
key usage, the number of messages encrypted with the same key and the message
routing can be considered. The forecasting algorithm can be equipped with an
alarming tool to safeguard the users. This technique will be useful for any
enterprise in order to forecast the security attacks and in order to change
the key before it can be compromised, so they can budget to defend against
the attacks. This procedure provides insufficient information and will be
considered as useful tool for approaching perfect security according to
Shanon criteria.
back to Accepted Papers Page
Efficient Grid Location Update Scheme for Mobile Ad Hoc Networks
Paper Number: 146
Abstract:
Location update schemes
are vital for the success of position-based routing in mobile ad hoc
networks. However, efficient location update schemes have been focus of
research for scalability of ad hoc networks. This led to development of few
location update schemes. Grid Location Service (GLS) has been one of commonly
referred update scheme in the literature. In this paper, we propose an
efficient grid location update scheme that outperforms the Grid Location
Service scheme. We use the concept of Rings of squares for location updates
and selective query mechanism for destination query. Our simulation results show
that our new grid scheme has better percentage of successful queries with
smaller location update cost to update the location servers. Our new grid
scheme also performs location updates faster than GLS
scheme.
back to Accepted Papers Page
Intelligent Agents and Smart Businesses
Paper Number: 149
Abstract:
Examining potentials of
agents in electronic environment in which companies can proceed
their activities thoroughly, we propose two conceptual models describing
know-how of using agents in strategic decisions. The first model presents a
big picture of using intelligent agents in a competitive environment, where
all the competitors are struggling virtually. The second one
prototypes a typical competitor from the interior
perspective with more details. This work is a beginning and a stand stone to
open a breakthrough toward constructive a smart business media.
back to Accepted Papers Page
A Model for Management of Transactions in the Mobile Databases
Paper Number: 154
Abstract:
In a mobile computing
environment, it is important to develop transaction processing systems that accommodate
the limitations of mobile computing, such as frequent disconnections, low
power, limited bandwidth communication and reduced storage capacity. In this
paper we propose a mobile transaction processing model which
distincts in treatment between LDD: Location Dependent
Data and NLDD: No Location Dependent Data.
back to Accepted Papers Page
Meta-middleware:A Solution Allowing Interoperability between Middlewares
Paper Number: 156
Abstract:
Middlewares such as CORBA, RMI, DotNet or even DCOM allow the interoperability of the
various customers of the company with the server, but their heterogeneity
called into question the interoperability of the servers of different
enterprises (for instance in a B2B context). In order to solve this problem,
we propose a software layer called meta-middleware, which contains
protocolar personalities of the existing
middlewares. Its role consists of translating a request
in the format of the target middleware. In addition to this
meta-middleware layer, it is necessary to extend the existing
middlewares with new components. The latter allow
intercepting the request sent outside the company and extracting
information such as object reference, parameters, name of the invocate
method�.for transmitting them to meta-middleware.
back to Accepted Papers Page
Explanatory dialogical agents for qualitative simulation
Paper Number: 158
Abstract:
Our proposal in this
paper is to explain a qualitative simulation algorithm. Five active software
components, considered as agents, collabrate to
produce this explanation. Communication between them is described using an
Agent Communication Language (ACL). Explanatory text is elaborated
cooperatively with the user. User interaction with explanatory agents is
task-oriented. It is said dialogical interaction. Explanatory agents carrying
out this interaction are called dialogical agents.
back to Accepted Papers Page
Injecting bit flips in VHDL models A comprehensive approach and prelimmary
results
Paper Number: 159
Abstract:
Fault injection in VHDL
models is a well established approach to study the consequences of transient
faults on the behavior of integrated circuits. This paper describes details
of a strategy adopted to reproduce as close as possible real faults
(bit-flips) that may occur in the atmosphere due to neutron radiation.
Benchmark programs were used to get simulated error rates on an 8-bit
microcontroller: the Intel 8051. Obtained results were compared with other
fault injection experiments.
back to Accepted Papers Page
An Ontology Development Process for the Semantic Web
Paper Number: 162
Abstract:
The idea of Semantic
Web is to tend towards a Web where semantics of data would be at the same time
understood both by machines and by human. For this reason, it is necessary to
solve problems involved in the description and the processing of the meaning
contents of the web documents. One of the solutions brought at the present to
this problem of document semantics representation consists of exploiting
ontologies and proposes new languages to represent and
reason on these ontologies. The first languages
which were proposed are RDF and RDFS. These latter, have a very limited
capacity of expressions, which recently led to the development of DAML+OIL
language. it brings to Web languages, the equivalent
of a description logics. The use of ontologies in
this context requires a good design and a good definition. Actually, there is
no a complete process, begining by brut data and
ending at an ontology represented by DAML+OIL language. This
article, is based on the work results realized in
Artificial Intelligence and knowledge engineering, in order to propose a
development process of ontologies defined by DAML+OIL
language.
back to Accepted Papers Page
An Ontological Approach for Building User Profiles
Paper Number: 163
Abstract:
We describe in this
paper how ontologies may be used to model different
user categories. We first look at what aspects need to be described for the
purpose of user model design in the context of a research laboratory. Then we
look at how these aspects are related to each others. In order to be generic
and reusable in different application domains, the user ontology is
formalized with Description Logic and implemented using Semantic Web
technology. Specific characteristics of the users such as physical data,
social background, computer experience level, type of activity, personal
characteristics, preferences, goals, interests and so forth, are integrated
in this ontology.
back to Accepted Papers Page
Objective Evaluation of Multi code books Vector Quantization of LSF Parameters of the CELP Coding
Paper Number: 166
Abstract:
Vector quantization
(VQ) is very used in speech coding. Our goal is the coding of the frequencies
spectral line (LSF) with the less complexity without losing the quality of
speech discounted however. Lot of code books transformation methods of
VQ have been studied until now. Among them the split VQ and multi stage
VQ (SVQ and MSVQ) presenting each of them a disadvantage corrected lately by
the method of Multi codebooks applied for the SVQ giving the multi code books
split vector quantization MC-SVQ thing that is the subject of our study for
an optimal application of this technique on the MSVQ to give the multi code
books multi stage vector quantization MC-MSVQ, and for an acceptable
complexity without loss of information and the quality.
back to Accepted Papers Page
Relationships between attributes to integrate heterogeneous data sources
Paper Number: 168
>
Abstract:
In this paper we
present our manner to build a data warehouse using heterogeneous data
sources. Nowadays, information systems are constructed thanks to the use of
internal data within an organization and
also some of the numerous data
sources available on the web. Thus, the challenge is to integrate all these
kinds of data which may be heterogeneous. Our work is to present a
methodology to construct a set of views (a data warehouse) based on the
extraction of the inter-schema relationships between sources. We define three
types of relationships between attributes: the synonymy, the inclusion and
the disjunction. To illustrate these relationships we consider two relational
and/or object-relational databases. All the different stages proposed in this
work are implemented by the use of a functional prototype.
back to Accepted Papers Page
XML-Based Query Processing in a Data Integration Architecture
Paper Number: 170
Abstract:
Abstract:
. In this paper
we present an XML based query processing for data integration problem. Our
approach is based on two main concepts; ontology and mediator. By providing
local ontologies, which describe the local sources,
we attempt to solve semantic conflict. A global ontology permits to integrate
the local ones using the global concept dictionary and the global mapping
tree. The second concept is used to perform the tasks necessary to the
integration, such as transformations of queries, conversion of local results,
their merging into a global one. In our work, the queries and the results are
represented using XML. The latter provides a strong level of representation
and permits to have coherence in the system. In the implementation phase, we
choose the distributed object paradigm and the web services technology. The
latter, improve the interoperability of the different mediators, especially
in the web environment.
back to Accepted Papers Page
Trinocular Image Matching by Hopfield network
Paper Number: 173
Abstract:
Computer vision has
become in the last years a very important reserach field.
Image matching is one of problems in computer vision known by their
difficulty, its results are used for the purpose of depth
analysis or 3D reconstruction. Matching approaches are grouped generally in
two categories: feature or edge based methods and intensity or region based
methods. We propose a matching method that uses Hopfield neural networks on
images triplet that processes on two stages: the first is based on region
primitives and the second on point frontiers. In the first stage based on regions,
we use region characteristics which are surface, gray level, elongation and
gravity center constraints. Each neuron represents a potential matching of
three regions from three different images. The region relational
characteristics are unicity and adjacency/order
constraints which represent connections between neurons. In the second stage,
we use point primitives extracted from region frontiers to do the matching
process after polygonal approximation of the region frontier. Point
characteristics are correlation similarity at the point neighborhood window,
epipolar constraint and gray level. Point relational
characteristics are orientation and order constraints. As result, results of
matched frontier points can be applied to calculate 3D point coordinates of
each point. Experimental results are shown in our method.
back to Accepted Papers Page
A New Adaptive Zone Filter To Estimate RTT In MANET
Paper Number: 176
Abstract:
The guaranty of the
Quality of Service (QoS) in the Mobile Ad hoc
NETwork (MANET) is a need to generalize their deployment
[1]. Indeed, the performances of wireless network are related to many
parameters: the load of the network, the number of nodes, the state of the
radio link, etc. It is very difficult to quantify exactly all these
parameters. So, we propose to use the Round-Trip Time (RTT) as an estimator
of the network exchange capacity and its available bandwidth. A fine analysis
of this estimator based on the theory of EWMA filters is proposed: the
Adaptive Zone filter. The purpose of the presented work is to improve the
performance of existing algorithms by using a more progressive
behaviour than the traditional Flip-Flop filter. The
performances of the Adaptive Zone filter are analysed
through simulation under OPNET�. The prospects of the RTT estimator
utilization are finally clarified.
back to Accepted Papers Page
Modeling By A Multi Agents System an Educationnal Adaptive Dynamic Hypermedia System
Paper Number: 179
Abstract:
For many years,
hypermedia has been a new research area in the field of computer aided
teaching systems, with its simplicity and its use in the Web which has been the
best technology for distributing information. Three kinds of systems have
successively appeared: classical hypermedia, adaptive hypermedia and
dynamic adaptive hypermedia. This paper presents a modeling of an educational
adaptive dynamic hypermedia system by a multi agent system. This system
generates dynamically the hypermedia pages of the course, these pages adapt
to the model of learner. It use the teaching
activities to present many sights of same concept and to specify the tasks
defining the structure best adapted to understand the concepts to be taught.
This system consists of five agents : learner model agent, that incorporate
and manipulate the learner model, the pedagogical agent of adaptation , which
incorporates the domain model and the pedagogical activities model
defined by the teacher; he specifies the pedagogical structure of
the course and specifies the concepts to be studied before the others
by communicating with the learner model agent to consult the learner model,
then it asks the filtering agent to filter the documents adequate to the
learner model, this last agent applies filters on the multimedia documents
database, then sends the addresses of filtered documents to the
generator of hypermedia page agent, which generates the page and sends it to
the interface agent which visualizes it to the learner. The interface agent
acts as a gateway between the learner and the pedagogical agent for
adaptation, learner model agent and the filtering agent.
back to Accepted Papers Page
Extraction of Road Networks from Satellite Images
Paper Number: 180
Abstract:
The goal of this study
is the definition of a model of extraction of the road network from satellite
images. It is a question of building a model which takes account at the
same time of the radiometric aspect and the space aspect of the road.
We show that a road cannot be characterized satisfactorily by a whole of
methods based on radiometric characteristics like the morphological
operators. On the other hand, it proves that a method based at the same time
on radiometric and space characteristics characterizes a road
satisfactorily. It makes it possible to study the problem of extraction
generally. This method of extraction of the road network is based on an
approach suggested by D. Geman and B
Jedynak [ 1 ] and improved by
us. In this work, we improved the following steps: detection of
homogeneous elements of road, detection of linear components of road and optimization
of a certain cost function with the resolution of the large pixel (area) to
find the best binary configuration representing our road networks. The
experimental study concerning various types of satellite images, more
particularly urban image and semi-urban image, shows that this method
characterizes the road networks satisfactorily, but its major disadvantage is
the arbitrary selection of the parameters used which remain difficult to
calculate in a single way for all the images to treat.
back to Accepted Papers Page
Further Investigations Of Accuracy-Based Fitness Using A Simple Learning Classifier System
Paper Number: 189
Abstract:
After the introduction of
XCS, most current Learning Classifier Systems research involves the use of
accuracy-based fitness, where rule fitness is based on a rule�s ability to
predict the expected payoff from its use. Whilst XCS has been shown to be
extremely effective in a number of domains, its complexity can make it
difficult to establish clear reasons for its behaviour.
In this paper, we investigate drawing on XCS, aspects of the steady-state
genetic algorithm within YCS - previously introduced based on
XCS�s framework - with the aim of further explaining the
effects of operators within accuracy-based fitness. Using the multiplexer
problem and drawing on XCS, results confirm that increased separation in the
fitness function aids performance and that increasing the rate of mutation
and/or degree of crossover has a marked effect
on specificity, but to a lesser extent in the case of the crossover operator.
Also, use of fitness bias in the replacement scheme has shown to increase
specificity to significantly improve performance, particularly for lower
reproductive pressure. Finally, deletion was found to interact with the GA
subsumption process, resulting in improved rule
compression than either deletion or GA subsumption
alone.
back to Accepted Papers Page
Modeling Dynamic Destination Target Tracking
Paper Number: 190
Abstract:
In this paper we present a modeling of dynamic
destination target tracking in public transit network, aiming at minimizing
the travel time associated with the required OD-trip. Public transit network
is viewed as a set of mode lines in which vehicles served passengers at time
points. The proposed strategy accounts for the real time information such as
the number of passengers waiting at station, the forecasting arrival times of
vehicles at stations and the number of passengers boarding vehicle. Delay may
modeled passengers boarding and alighting duration
or modal transfer time from one line to an other. The experimental results
are given in the paper.
back to Accepted Papers Page
Visually Guided Object Grasping Without Calibration of the Stereoscopic System
Paper Number: 197
Abstract:
In this paper we propose
a new method for visual servoing of an autonomous
robot manipulator based on projective geometry. The originality of this work
resides in the fact that we avoid the calibration step that is tedious and
unstable (measurements errors, approximated model). The method proposed is
based only on bi-dimensional data extracted from
the images. To determinate the trajectory of the robot: translational and
rotational movements, we use 3D projective reconstruction. The experimental
results validate the proposed approach, taking several positions of the point
to reach.
back to Accepted Papers Page
VEMMA: A Virtual Electronic Market Place based on Mobile Agents
Paper Number: 199
Abstract:
Several architectures
were proposed in the literature for the modeling of the interactions between
agents. Within the framework of this article, we describe three architectures
based on the multi agents design. In these systems, Buyer and Seller agents interact
in an environment comparable to an electronic market in order to sell and buy
goods. We propose then an electronic architecture of commerce based on
intelligent and mobile agents called VEMMA. We present the formal model of
this architecture as well as an implementation using the Java language and
the RMI technology.
back to Accepted Papers Page
A Text-Dependent Speaker Identification Methods And Experiments On Arabic Database
Paper Number: 201
Abstract:
In this paper we aimed
to apply tow traditional algorithms for speaker recognition on Arabic
database. One using determinist model based dynamic time warping (DTW), other
one using probabilistic model based Gaussian distribution probability. In
both methods, we have study the effect of the order of LPCC coefficients on
the classifiers performance. In particular, we have attempted to conceive a
system able to recognize all speakers based DTW multi-references. In the
second method, we have also study the effect of variance on the rate
identification. Our experiments are evaluated on Arabic database composed of
twenty speakers and 100% identification rate is reached.
back to Accepted Papers Page
Satellite-Derived Surface Radiation Budget over the Algerian area. Estimation of surface albedo
Paper Number: 206
Abstract:
Surface albedo is the main input variable for the algorithms that
solve the surface energy balance. Albedo is here
referred to as the ratio of reflected solar radiation and the total incoming
solar radiation for the wavelength of 0.3 �m to 1.05 �m (visible band).
Absorption and scattering in the atmosphere causes the albedo,
as observed by the satellite in the visible band (planetary
albedo) to differ from the actual surface
albedo. Absorption in the atmosphere is mainly due to
water vapour. Scattering occurs as a result of the
presence of air molecules (e.g.N2,O2)
and aerosols in the atmosphere. After calibration of the visible images, the
planetary albedo is obtained. The atmospheric
correction procedure converts planetary albedo to
surface albedo.
back to Accepted Papers Page
Task-to-Processor Assignment in Heterogeneous Network of Processors (Cluster) Systems
Paper Number: 209
Abstract:
Task to Processor Assignment
(TPA) problem has recently gained a great deal of attention with the ever
increase of using distributed and cluster systems as an
underlying architecture to execute parallel tasks. The challenge is to assign
subtasks of a job such that minimum number of processors is used while
minimizing the finish time and the communication overhead. In this paper, we
propose an algorithm that schedules the subtasks (
processes) on a heterogeneous distributed system, taking into
consideration the processors speeds, processors distances and adjacencies,
communication paths latencies and tasks weights at the same time. Considering
all those elements, the algorithm proposed is successful in obtaining a
schedule with a minimum finish time. The algorithm is validated using an
example and an exhaustive search simulation program.
back to Accepted Papers Page
����� ������� ������� ���������� �������� ������ ����� ���� �� ������ ����������
Paper Number: 211
Abstract:
This study can be summed up as follows: The
embodiment of the recent out comes in the approach of the rudimentary theory
of the neural network and its application in the field of the photovoltaic
pumping water system with centrifugal pump The application of the method of
the back propagation for the error of artificial neural network intelligence,
and the reliance on this intelligence gives an ideal addition to photovoltaic
pumping system. The training of artificial neural network with two inputs and
a hidden layer with four hidden neurons of suitable number examples. That
makes it with strong robustness, good efficiency and simple hierarchy, and an
easy realization.
back to Accepted Papers Page
Fuzzy Identification of Gluconic Acid to Glucose Fermentation Parameters
Paper Number: 213
Abstract:
A construction of
Takagi-Sugeno (TS) fuzzy models using clustering is
presented. The input, which is a sequence of empirical data, is partitioned
into a certain number of clusters. The important property of the
modelling method is the global model accuracy and the
local model interpretability; we use, in our modelling
the global and the local learnings, in order to
improve the interpretability, and obtain a good global approximation. The
model representation is based on the interpolation of a number of local
models, where each local model has a limited range of validity. The method is
illustrated using simulated data from the fermentation of glucose to
gluconic acid by the microorganism Pseudomonas
ovalis in a well-stirred batch reactor.
back to Accepted Papers Page
Methodological Test Integrating the SIG and the Remote Sensing for the survey of the marin pollution in term of hydrocarbons: case of ARZEW harbors (Algerian West).
Paper Number: 214
Abstract:
The problem of marin pollution is probably one of some troubling aspects
of the deterioration of natural middle. this
pollution is a major risk which would hadicap
seriously the biosphere_Geosphere balance. The oil
is the principal pollutant and the Mediterranean sea is concerned by 18% of
the worldwide pollution bye the oils. The harbors of Arzew
are concerned by a big naval traffic; bound to an industrial zone
particularly to risk in matter of pollution. Building an approach which uses
the remote sensing data and the interpretation of the chemical data of the
waters of the taking as well as the integration of these acquaintances in a
Geographical Information System (GIS), constitutes the objectives assigned to
our work
back to Accepted Papers Page
iSWare: An Inter Organizations Web-Based Urban Planning Coordinator
Paper Number: 215
Abstract:
This paper describes a urban planning coordination tool, called �iSWare�. iSWare
means inter Sites groupWare, it is the result of an
analytical study made on a group of independent organizations. Considered
organizations use GIS in their urban projects. iSWare consists of an open component-based system.
It is an online exchange coordinator playing the role of a general
inter-organizations committee (board). It provides an easy internet support
using groupware technologies. iSWare
opts for a specific coordination protocol by applying the theory of
competition and data distribution.
back to Accepted Papers Page
Tools of assistance to the design of image processing applications
Paper Number: 216
Abstract:
This research is part
of an interactive and incremental design
of image processing and analyzing applications framework. Several
tracks are explored, a system of assistance to the design of Image
Processing (IP) applications using an approach based on the case-based
reasoning (CBR) which is still used very little in the field of (IP) is
proposed . This type of reasoning is however well adapted to
badly formalized fields, such as (IP), since it allow the
representation of the exceptions, the use of missing information and
the resolution of a complex problem by combination of solutions of
simpler problems. In this article, the concept of "
case in IP " is defined. Our objective is to
present a vision different from the process of
reasoning based case through the multi-agent paradigm. Previous
work on the use of the case-based reasoning,
presented the case as a passive entity in waiting to be found to offer
an assistance to the decision or the resolution of problem. Here,
the use of active cases organized in
networks. In other words agents in a company which solve
the problems while possibly adapting.
back to Accepted Papers Page
Managing a Pre fetching Scheduling into a SMIL presentation
Paper Number: 219
Abstract:
The SMIL presentation consistency
can not be guaranteed when the required and synchronized medias can not be
delivered in real time for the client player. This is due generally to the
bandwidth fluctuations or to its bad use. In this paper we present an
approach for providing a consistent presentation for SMIL1.0 documents by
scheduling the insertion of pre fetching commands into a SMIL1.0 document.
Contrary to the existing methods, the process is done with managing the
client resources like bandwidth and memory which avoids client player
overflow and crash.
back to Accepted Papers Page
Geographical Information System for Urban Spatial Analysis
Paper Number: 220
Abstract:
Many research works in automated
cartography are carried out to solve problems related to social and urban
human needs. In this paper, an information system for urban spatial analysis
is presented. The proposed system outlines a method for urban spatial
representing, modelling and reasoning that provide
valuable intelligent aids for sorting out different spatial problems such
land use planning, environmental assessment and demo-socio-economic analysis.
Thus, the proposed system would be a major contribution for solving difficulties
frequently encountered by large users in urban agencies.
back to Accepted Papers Page
A Parallel Algorithm for Edge Detection Based on Infinite Impulse Response filter (IIR filter)
Paper Number: 221
Abstract:
In this paper, we
present a parallel algorithm for edge detection based on Infinite Impulse
Response filter (IIR filter). In particular, the Infinite size Symmetric
Exponential Filter (ISEF) which is an optimal IIR filter and computationally
efficient smoothing filter is studied. The proposed algorithm exploits
efficiently all aspects of potential parallelism (spatial parallelism,
temporal parallelism and systolism) inherent in the
considered egde detection algorithms. The designed
concurrent algorithm is expressed in terms of a collection of concurrent
processes communicating and synchronizing in an efficient way in order to
speed up the low-level operations.
back to Accepted Papers Page
Evolutionary Algorithms Based Neural Networks Training for Phonetic Classification
Paper Number: 222
Abstract:
Abstract:
: In this
paper, we propose a biologically inspired method based on evolutionary algorithms
(EA). They are three main aims. First to explain the general mechanism of
evolutionary algorithms. Second to present the manipulation of neural network
by evolutionary algorithms another way determine the population and the
fitness function for each individual. Finally
describe the similarities and differences between genetic algorithms (GA) and
evolution strategies (ES). This new paradigm is applied to
real world problem like speech recognition exactly on TIMIT
database and illustrates the performance of evolution strategies compared to
genetic algorithms.
back to Accepted Papers Page
Infrared (IR) Wireless Remote Control of the iPod Digital Music Player
Paper Number: 224
Abstract:
In this paper, we
propose the use of a multi-threading real-time operating system (RTOS) to
control an infrared (IR) wireless receiver, for use with the
iPod digital music player. Included with the
iPod is a wired remote control, but its usefulness is
questionable at best. A wireless remote control would be very useful when the
iPod is docked and connected to a home
entertainment system. It would be beneficial to have access to the nearly
4,000 songs on the iPod�s 15GB hard disk, through a
high-quality home entertainment system, all available at your fingertips via
a wireless remote control.
back to Accepted Papers Page
A First Step Towards Adding Context to Web Services Standards CWSDL: a Context-Based Web Services
Paper Number: 226
Abstract:
Abstract:
. Our long term
research objective is to add context to Web services, in order to help with better
service discovery and selection in pervasive environments. As a .rst step towards this goal we
propose to add context to the current Web services industry standards namely
the WSDL and UDDI. In this paper we present the CWSDL (Context-Based Web Services
Description Language), we believe that one of the best ways for enhancing
WSDL with context-aware features could be to add context to the WSDL
following the standard schema; instead of creating a completely new language.
It is worth noticing that the work presented in this paper is a preliminary
work and thus needs more enhancements in the future.
back to Accepted Papers Page
Development Of A Vocal Calcula Tor Based On Arabic Speech Recognition System
Paper Number: 233
Abstract:
Technologies vocal and
speech processing currently represent an increasingly significant potential
in Human-machine interface. It is a question of developing a system able to
translate or decode an acoustic signal of ward emitted by a speaker.
ln this article we present basic technologies intervening
in the design of a vocal calculator based on real lime speech recognition
system of isolated Arabic words: the acoustic analysis by treatment of the
signal the use of corpus of training and a statistical modeling by the
continuous Hidden Markov Model, although Several experiments are carried out
for the choice of the optimal parameters of the system. Good results are
obtained using a real lime implementation.
back to Accepted Papers Page
Fuzzy Automatic Classification without prior knowledge applied to brian image segmentation
Paper Number: 235
Abstract:
The objective of this paper
is to propose a blind segmentation method able to localize
aIl relevant objects in medical images by using a fuzzy
classification based mean shift algorithm. To achieve this, we have to build
cartography of attributes consequent upon images characterization. The
objects localization is realized by searching modes frOID
a point sample distribution through mean shift procedure.
ln order to obtain an image automatic segmentation,
the approach is joined at a fuzzy classification based fuzzy c-means (FCM) approach.
The fuzzy insertion here allows to take into account imprecision related to
information extraction which is necessary to region classification.
ActuaIly, the obtained results by our approach are very
encouraging and show an accurate segmentation compared others supervised
techniques.
back to Accepted Papers Page
Design and Implementation of a Voronoi Diagrams Generator using Java
Paper Number: 239
Abstract:
Voronoi Diagrams (VDs) and Delaunay
Triangulations (DTs) are rediscovered mathematical concepts from the 19th
century. These concepts are of great significance in many fields of science.
This comprises of biology, medicine, telecommunication networks, imagery,
geography and countless more. This paper describes the design and the
implementation phases of a system used to generate 2D VDs
and DTs. The paper commences with highlighting these momentous concepts by
introducing their pioneers, over-viewing its functions, related terminologies,
and properties. The incremental algorithm was selected to generate
VDs and DTs respectively. Java was used to develop the
system. The method is briefly described. The intention is to further develop
the system to integrate it with other applications such as generating VD for
a loaded image. The system can be utilized in teaching knowledge-seekers its
usage in real world applications.
back to Accepted Papers Page
The Generalized Service Level Agreement Model and its application to structuring WLANs into Wireless Management Communities
Paper Number: 242
Abstract:
In this work, a
service-driven model for structuring WLANs into
overlay networks of interacting Wireless Management Communities is proposed.
We define the GSLA, a generalized model for the specification of Service
Level Agreements (SLA). The GSLA information model supports multi-party
service relationships through a role-based mechanism. It is intended to catch
up the complex nature of service interactivity in the broader range of SLA modeling of all sorts
of IT business relationships. This model accommodates both granularity and
modularity of behavioral specifications by having each party playing a role
within a service relationship. We then apply the GSLA to structuring
WLANs into Wireless Management Communities (WMC). A WMC
is composed of a set of Parties and is governed by a charter named the
WMC-SLA. A WMC constitutes the basic unit of management upon which will be
installed any form of service interaction between parties belonging to the
wireless community. The models we propose intend to bring a step towards
SLA-driven management within pervasive service environments.
back to Accepted Papers Page