1359 research outputs found
Sort by
Picture processing for enhancement and recognition
Get PDFRecent years have been characterized by an incredible growth in computing power and storage capabilities, communication speed and bandwidth availability, either for desktop platform or mobile device. The combination of these factors have led to a new era of multimedia applications:
browsing of huge image archives, consultation of online video databases, location based services and many other. Multimedia is almost everywhere and requires high quality data, easy retrieval of multimedia contents, increase in network access capacity and bandwidth per user. To meet all the mentioned requirements many efforts have to be made in various research areas, ranging from signal processing, image and video analysis, communication protocols, etc.
The research activity developed during these three years concerns the field of multimedia signal processing, with particular attention to image and video analysis and processing. Two main topics have been faced: the first is relating to image and video reconstruction/restoration (using super resolution techniques) in web based application for multimedia contents' fruition; the second is
relating to image analysis for location based systems in indoor scenario.
The first topic is relating to image and video processing, in particular the focus has been put on the development of algorithm for super resolution reconstruction of image and video sequences in order to make easier the fruition of multimedia data over the web. On one hand, latest years have been characterized by an incredible proliferation and surprising success of user generated multimedia
contents, and also distributed and collaborative multimedia database over the web. This brought to serious issues related to their management and maintenance: bandwidth limitation and service costs are important factors when dealing with mobile multimedia contents’ fruition. On the other hand, the current multimedia consumer market has been characterized by the advent of cheap but rather high-quality high definition displays. However, this trend is only partially supported by the deployment of high-resolution multimedia services, thus the resulting disparity between content and display formats have to be addressed and older productions need to be either re-mastered or postprocessed in order to be broadcasted for HD exploitation. In the presented scenario, superresolution
reconstruction represents a major solution. Image or video super resolution techniques allow restoring the original spatial resolution from low-resolution compressed data. In this way, both content and service providers, not to tell the final users, are relieved from the burden of
providing and supporting large multimedia data transfer. The second topic addressed during my Phd
research activity is related to the implementation of an image based positioning system for an indoor navigator. As modern mobile device become faster, classical signal processing is suggested to be used for new applications, such location based service. The exponential growth of wearable devices, such as smartphone and PDA in general, equipped with embedded motion (accelerometers)
and rotation (gyroscopes) sensors, Internet connection and high-resolution cameras makes it ideal for INS (Inertial Navigation System) applications aiming to support the localization/navigation of objects and/or users in an indoor environment where common localization systems, such as GPS (Global Positioning System), fail. Thus the need to use alternative positioning techniques.
A series of intensive tests have been carried out, showing how modern signal processing techniques can be successfully applied in different scenarios, from image and video enhancement up to image recognition for localization purpose, providing low costs solutions and ensuring real-time performance
Le riscritture di Great Expectations. Sei letture del classico dickensiano
Get PDFThis thesis originated from the discovery of a rich group of texts published as an “answer” to the novel Great Expectations by Charles Dickens.
After an introductory section offering a panoramic view of the great fortune of this Dickensian classic, my research deals with the main key issues regarding the debate around intertextuality.
Afterwards it focuses on the definition of rewriting, necessary for the choice of a homogeneous group of texts.
The novels that this thesis analyses were published between 1905 and 2006 and the choice of a comparative approach has brought to light the particular interpretation of the Dickensian novel given by these rewritings
Sintomi non motori e disautonomia cardiovascolare in pazienti sardi affetti da malattia di parkinson con e senza mutazioni del gene LRRK2
Get PDFBackground: Although Parkinson’s disease (PD) is commonly described as a motor disorder, non-motor symptoms and cardiovascular dysautonomia are important factors of disability and impaired quality of life in PD patients. To date, the definite correlation of cardiovascular symptoms with motor and other non-motor symptoms of PD are still largely unclear. Besides, the discovery of genetic mutations related to PD, such as LRRK2 mutations, provides better possibility to identify specific phenotypes both for motor and non-motor symptoms.
Obiectives: The main objective of our study was directed to investigate the presence and the correlation of cardiovascular dysautonomia with other non-motor symptoms in PD patients, with with or without LRRK2 mutations. Finally, we performed a study of heart rate variability (HRV) and exercise stress echocardiography with the attempt to investigate dysautonomia in these groups of patients.
Methods: Forty fifteen Sardinian PD patients (240 men and 165 women) were included in the study. PD patients were screened for the presence of LRRK2 mutations. Motor impairment and disability were assessed using the Hoehn & Yahr staging and the Unified PD Rating Scale (UPDRS) part-III. Cardiovascular symptoms and other non-motor symptoms were assessed with the Non-Motor Symptoms Scale (NMSS). Correlations between cardiovascular symptoms and other PD features were studied. Heart rate variability (HRV) and exercise stress echocardiography were performed in two subgroups of patients (with or without LRRK2 mutations) and in a control group, matched for age and gender.
Results: Cardiovascular symptoms were among the five non-motor symptoms more reported and were correlated with multiple non motor symptoms, especially with fatigue and loss of weight. PD patients with LRRK2 mutations reported less severity of cardiovascular symptoms. HRV revealed reduced LF/HF ratio both in PD patients with or without LRKK2 mutations, with enhancement of HF in LRRK2 patients. Cardiovascular symptoms detected at NMSS correlated with LF/HF ratio. Echocardiography show a Δ Strain rate reduced in parkinsonian patients. Δ Strain rate was also correlated with the LF/HF ratio in the LRRK2 group.
Discussion: The better assessment of non-motor and cardiovascular symptoms in PD patients with and without LRRK2 mutation, both with the systematic administration of symptomatic scales, such as the NMSS, and with a combined approach with HRV and exercise stress echocardiography, might lead to the detection of better monitoring systems and to the possible improvement of current strategies used in the treatment of these disabling conditions
A framework for feature selection in high-dimensional domains
Get PDFThe introduction of DNA microarray technology has lead to enormous impact in cancer research, allowing researchers to analyze expression of thousands of genes in concert and relate gene expression patterns to clinical phenotypes. At the same time, machine learning methods have become one of the dominant approaches in an effort to identify cancer gene signatures, which could increase the accuracy of cancer diagnosis and prognosis. The central challenges is to identify the group of features (i.e. the biomarker) which take part in the same biological process or are regulated by the same mechanism, while minimizing the biomarker size, as it is known that few gene expression signatures are most accurate for phenotype discrimination.
To account for these competing concerns, previous studies have proposed different methods for selecting a single subset of features that can be used as an accurate biomarker, capable of differentiating cancer from normal tissues, predicting outcome, detecting recurrence, and monitoring response to cancer treatment. The aim of this thesis is to propose a novel approach that pursues the concept of finding many potential predictive biomarkers. It is motivated from the biological assumption that, given the large numbers of different relationships which are possible between genes, it is highly possible to combine genes in many ways to produce signatures with similar predictive power. An intriguing advantage of our approach is that it increases the statistical power to capture
more reliable and consistent biomarkers while a single predictor may not necessarily provide important clues as to biological differences of interest.
Specifically, this thesis presents a framework for feature selection that is based upon a genetic algorithm, a well known approach recently proposed for feature selection. To mitigate the high computationally cost usually required by this algorithm, the framework structures the feature selection process into a multi-step approach which combines different categories of data mining methods. Starting from a ranking process performed at the first step, the following steps detail a wrapper approach where a genetic algorithm is coupled with a classifier to explore different feature subspaces looking for optimal biomarkers. The thesis presents in detail the framework and its validation on popular datasets which are usually considered as benchmark by the research community. The competitive classification power of the framework has been carefully evaluated and empirically confirms the benefits of its adoption. As well, experimental results obtained by the proposed framework are comparable to those obtained by analogous literature proposals. Finally, the thesis contributes with additional experiments which confirm the framework applicability to the categorization of the subject matter of documents
Il diritto amministrativo e l’amministrazione digitale: l’evoluzione del decreto legislativo 7 marzo 2005 n. 82 dalla teoria alla pratica
Get PDFSince its first introduction, the Digital Administration Code has become target of criticism by the academic studies and the Council of State. Object of the critic was the fact that the Code appeared as a purely theoretical text, too far off the administrative reality so its dispositions appeared hardly applicable. The aim of this work is to verify if the Legislator has adopted to the suggestions given by the doctrine and the Council of State, in order to apply the DAC and fulfill its purpose (and expectations), with a critical approach to the reforms that this Act has undergone
Low dimension polymer-based nanostructures for photovoltaics
Get PDFIn this thesis it is discussed the effect of the low dimensionality on several physical properties of hybrid
nanostructures for new generation photovoltaics. We investigated several hybrid and organic model
systems, consisting of an organic polymer playing the role of electron donor and a low-dimension
nanostructure (1D or 0D) which acts as electron acceptor and transporter. Model potential molecular
dynamics has been used to characterize the polymer morphology onto the nanostructured substrates. In
particular, the wrapping phenomena on one-dimensional structures (ZnO nanoneedles and carbon
nanotubes) have been analyzed as a function of several physical variables such as temperature, substrate
crystallography, polymer chain length and density. It has been thus possible to observe that wrapped
configurations are only metastable on carbon nanotubes at room temperature and in absence of solvents.
Nevertheless, wrapped configurations induced by the solvent can be frozen due to the interactions among
neighboring polymer chains. According to this study, it is possible to enhance the polymer-nanotube
alignment (and thus improving the polymer transport properties) through a suitable tuning of the
synthesis parameters. Conversely, wrapped geometries are stable on ZnO nanoneedles, due to the small
polymer mobility on the ZnO surface. The results obtained on the morphology of polymer-ZnO hybrids
have then been used as a starting point to evaluate the electronic structure and the optical absorption
properties. Hybrid models consisting in a 120-atoms ZnO nanoparticle and a set of oligothiophenes have
been studied through the density functional theory, and the energy-level alignment has been obtained by
using the Δ-self-consistent-field method. 120-atoms ZnO nanoparticles have been synthesized and found
to be particularly stable. They therefore not only represent a useful model for computational studies, but
are also of potential technological interest. An important result thus obtained is to demonstrate that the
interaction between the two organic/inorganic moieties shifts the energy levels, giving rise to a nonstaggered
junction. This phenomenon is not present in planar ZnO substrates, but it is rather induced by
the nanostructuration of the hybrid polymer/metal oxide system. From the methodological standpoint, a
simplified model to predict the energy-level alignment at the interface has been developed, allowing to
spare computational resources. Finally, since the atomic configuration of the 120-atoms ZnO nanoparticles
is unknown, we calculated the optical absorption spectra in the near ultra-violet of a set of (ZnO)60
isomers: this information can be compared to experimental spectroscopic data and can thus be used to
elucidate the most abundant structure of this cluster.
The results of the present work suggest that the use of nanostructures, although opening interesting
technological possibilities such as increasing the donor/acceptor interface, also requires a critical
readdressing of our understanding of morphologies and electronic level alignment in low-dimension
systems
Analysis and synthesis techniques of nonlinear dynamical systems with applications to diagnostic of controlled thermonuclear fusion reactors
Get PDFNonlinear dynamical systems are of wide interest to engineers, physicists and mathematicians, and this is due to the fact that most of physical systems in nature are inherently non-linear. The nonlinearity of these systems has consequences on their time-evolution, which in some cases can be completely unpredictable, apparently random, although fundamentally deterministic. Chaotic systems are striking examples of this.
In most cases, there are no hard and fast rules to analyse these systems. Often, their solutions cannot be obtained in closed form, and it is necessary to resort to numerical integration techniques, which, in case of high sensitivity to initial conditions, lead to ill-conditioning problems and high computational costs.
The dynamical system theory, the branch of mathematics used to describe the behaviour of these systems, focuses not on finding exact solutions to the equations describing the dynamical system, but rather on knowing if the system stabilises to a steady state in the long term, and what are the possible attractors, e.g. a quasi-periodic or chaotic attractors.
Regarding the synthesis, from both a practical and a theoretical standpoint, it is very desirable to develop methods of synthesizing these systems. Although extensive theory has been developed for linear systems, no complete formulation for nonlinear systems synthesis is present today.
The main topic of this thesis is the solution of engineering problems related to the analysis and synthesis of nonlinear and chaotic systems.
In particular, a new algorithm which optimizes Lyapunov exponents estimation in piecewise linear systems has been applied to PWL and polynomial chaotic systems. In the field of complex systems synthesis, a systematic method to project systems of order 2n characterized by two positive Lyapunov exponents, has been proposed. This procedure couples nth-order chaotic systems with a suitable nonlinear coupling function. Furthermore, a method for the fault detection has been developed. In the field of time series analysis, a new denoising method, based on the wavelet transform of the noisy signal, has been described. The method implements a variable thresholding, whose optimal value is determined by analysing the cross-correlation between the denoised signal and the residuals and by applying different criteria depending on the particular decomposition level. Finally, a study of dynamical behaviour of Type I ELMs has been performed for a future modelization of the phenomenon. In this context, a statistical analysis of time intervals between successive Type I ELMs has been proposed.---------------------------------- Il tema principale di questa tesi è la soluzione di problemi ingegneristici legati all’analisi e alla sintesi di sistemi dinamici non lineari. I sistemi dinamici non lineari sono di largo interesse per ingegneri, fisici e matematici, e questo è dovuto al fatto che la maggior parte dei sistemi fisici in natura è intrinsecamente non lineare.
La non linearità di questi sistemi ha conseguenze sulla loro evoluzione temporale, che in certi casi può rivelarsi del tutto imprevedibile, apparentemente casuale, seppure fondamentalmente deterministica. I sistemi caotici sono un esempio lampante di questo comportamento. Nella maggior parte dei casi non esistono delle regole standard per l’analisi di questi sistemi. Spesso, le soluzioni non possono essere ottenute in forma chiusa, ed è necessario ricorrere a tecniche di integrazione numerica, che, in caso di elevata sensibilità alle condizioni iniziali, portano a problemi di mal condizionamento e di elevato costo computazionale.
La teoria dei sistemi dinamici, la branca della matematica usata per descrivere il comportamento di questi sistemi, non si concentra sulla ricerca di soluzioni esatte per le equazioni che descrivono il sistema dinamico, ma piuttosto sull’analisi del comportamento a lungo termine del sistema, per sapere se questo si stabilizzi in uno stato stabile e per sapere quali siano i possibili attrattori, ad esempio, attrattori quasi-periodici o caotici.
Per quanto riguarda la sintesi, sia da un punto di vista pratico che teorico, è molto importante lo sviluppo di metodi in grado di sintetizzare questi sistemi. Sebbene per i sistemi lineari sia stata sviluppata una teoria ampia e esaustiva, al momento non esiste alcuna formulazione completa per la sintesi di sistemi non lineari.
In questa tesi saranno affrontati problemi di caratterizzazione, analisi e sintesi, legati allo studio di sistemi non lineari e caotici.
La caratterizzazione dinamica di un sistema non lineare permette di individuarne il comportamento qualitativo a lungo termine. Gli esponenti di Lyapunov sono degli strumenti che permettono di determinare il comportamento asintotico di un sistema dinamico. Essi danno informazioni circa il tasso di divergenza di traiettorie vicine, caratteristica chiave delle dinamiche caotiche. Le tecniche esistenti per il calcolo degli esponenti di Lyapunov sono computazionalmente costose, e questo fatto ha in qualche modo precluso l’uso estensivo di questi strumenti in problemi di grandi dimensioni. Inoltre, durante il calcolo degli esponenti sorgono dei problemi di tipo numerico, per ciò il calcolo deve essere affrontato con cautela. L’implementazione di algoritmi veloci e accurati per il calcolo degli esponenti di Lyapunov è un problema di interesse attuale.
In molti casi pratici il vettore di stato del sistema non è disponibile, e una serie temporale rappresenta l’unica informazione a disposizione. L’analisi di serie storiche è un metodo di analisi dei dati provenienti da serie temporali che ha lo scopo di estrarre delle statistiche significative e altre caratteristiche dei dati, e di ottenere una comprensione della struttura e dei fattori fondamentali che hanno prodotto i dati osservati. Per esempio, un problema dei reattori a fusione termonucleare controllata è l’analisi di serie storiche della radiazione Dα, caratteristica del fenomeno chiamato Edge Localized Modes (ELMs). La comprensione e il
16
controllo degli ELMs sono problemi cruciali per il funzionamento di ITER, in cui il type-I ELMy H-mode è stato scelto come scenario di funzionamento standard. Determinare se la dinamica degli ELM sia caotica o casuale è cruciale per la corretta descrizione dell’ELM cycle. La caratterizzazione dinamica effettuata sulle serie temporali ricorrendo al cosiddetto spazio di embedding, può essere utilizzata per distinguere serie random da serie caotiche.
Uno dei problemi più frequenti che si incontra nell’analisi di serie storiche sperimentali è la presenza di rumore, che in alcuni casi può raggiungere anche il 10% o il 20% del segnale. È quindi essenziale , prima di ogni analisi, sviluppare una tecnica appropriata e robusta per il denosing.
Quando il modello del sistema è noto, l’analisi di serie storiche può essere applicata al rilevamento di guasti. Questo problema può essere formalizzato come un problema di identificazione dei parametri. In questi casi, la teorie dell’algebra differenziale fornisce utili informazioni circa la natura dei rapporti fra l’osservabile scalare, le variabili di stato e gli altri parametri del sistema.
La sintesi di sistemi caotici è un problema fondamentale e interessante. Questi sistemi non implicano soltanto un metodo di realizzazione di modelli matematici esistenti ma anche di importanti sistemi fisici reali. La maggior parte dei metodi presentati in letteratura dimostra numericamente la presenza di dinamiche caotiche, per mezzo del calcolo degli esponenti di Lyapunov. In particolare, le dinamiche ipercaotiche sono identificate dalla presenza di due esponenti di Lyapunov positivi
Flexural models of reinforced concrete beams under blast load
Get PDFNowadays, the issue of structural safety under blast loading has become a dramatic problem. The tragic news
of the terrorist attacks of recent years (9/11/2001, New York; 7/7/2005, London; 7/23/2005, Sharm El Sheik;
1/24/2011, Moscow; etc), raise important, urgent questions regarding the real safety and reliability of our buildings. Extreme loads such as impacts, explosions, etc., can occur in everyday life with unexpectedly high frequency. Actually, the problem of terrorist attacks, so important for strategic and military building design, has been linked to residential and industrial building explosion accidents. The present thesis deals with the flexural failure of Reinforced Concrete beams under blast loads. The main aim is firstly to develop dynamic and energy models capable of evaluating the response of R.C. under explosive load. Then a sensitivity analysis is obtained by means of the above mentioned models in order to determine what are the key
parameters in the beam response. In this way it is possible to attain simple predictive polynomial formulations and, finally, simple table for early structural assessment of beams under blast load during the design procedure.
The thesis is composed of three sections. Various dynamic models are developed in Section 1, taking into
account the strain-rate sensitivity of both steel and concrete, as well as other nonlinearities in structural
behaviour. Models with different levels of complexity are presented, from the simplest Single Degree Of
Freedom (SDOF) system to Continuous Beam and Finite Element models. The characteristics, advantages and disadvantages of each approach are stated and discussed. The author compares his models with some experimental findings available in the scientific literature. The principal innovation that can be inferred from this section is the hypothesis of distributed plasticity along the beam (in the continuous model), represented by a non-linear smooth relationship between bending moment and curvature. This relationship, in addition to the Euler-Bernoulli beam equation, yields a non-linear Partial Differential Equation solved by means of the Finite Difference Method. In Section 2, the same problem is solved in a different way: starting from the principle of energy conservation, the innovative procedure, developed by the author, can calculate the displacement field of a R.C. beam under blast load. This procedure, less accurate than the previous one, produces very good results regarding midspan displacement, especially as it requires less computation time.
In order to better understand the phenomenon, with the aim of identifying the key parameters in structural response, a sensitivity analysis is developed in Section 3. To this end, the author has performed a numerical investigation referring to the SDOF model presented in Section 1. Several numerical simulations are performed, with random variation of beam and load characteristics. Results are expressed in terms of maximum deflection and maximum velocity for each case. Then a least-squares interpolation has produced
various polynomial curves and surfaces representing both a simplified tool to estimate structural response and a sensitivity analysis of the key parameters involved. One of the possible developments of this useful work is represented by simple tables that provide the response of the beam under blast load for early assessment in design procedures
Light vector meson production at forward rapidity in pp collisions at the LHC with the ALICE detector
Get PDFLow-mass vector mesons (ρ,ω and Φ) give us information on the hot and dense state of strongly interacting matter produced in heavy ion collisions. In fact, in the QGP an enhancement of the strange particle production should be present (strangeness enhancement), which means a Nø/Nρ+ω ratio increasing in A-A collisions with respect to pp colli-
sions; pp collisions provide a reference for these studies. Vector meson production in pp collisions is also useful by itself, to study soft Quantum ChromoDynamics (QCD) in the LHC energy range, since calculations in this regime are based on QCD inspired phenomenological models that
must be tuned to the data. The ALICE experiment at the LHC can access vector mesons produced at forward rapidity through their decays in muon pairs. I show the ratio
Nø/Nρ+ω vs pT , the ratios σρ/σω and σΦ/σω, and the Φ and ω pT and rapidity dierential cross sections for the data sample at 7 TeV. For the data sample at 2.76 TeV, because of the limited statistics, I have show only the Nø/Nρ+ω ratio and Φ and ω integrated cross sections
Models and frameworks for studying social behaviors
Get PDFStudies on social systems and human behavior are typically considered domain of humanities and psychology. However, it appears that recently these issues have attracted a strong interest also from the scienti�c community belonging to the hard sciences {in particular from physics, computer science
and mathematics. The network theory o�ers powerful tools to study social systems and human behavior. In particular, complex networks have gained a lot of prestige as general framework for representing and analyze real systems.
From an historical perspective, complex networks are rooted in graph theory {which in turn is dated back to 1736, when Leonhard Euler wrote the paper on the seven bridges of K�onigsberg. After Euler's work, di�erent mathematicians (e.g. Cayley) focused their research on graphs {opening the
possibility of applying their results to deal with theoretical and real problems. As a result, complex networks emerged as multidisciplinary approach
for studying complex systems. From a computational perspective, models based on complex networks allows to extract information on complex systems composed by a great number of interacting elements. A variety of systems
can be modelled as a complex network (e.g. social networks, the World Wide Web, internet, biological systems, and ecological systems). To summarize, any such system should give the possibility of viewing its elements as
simple (at some degree of abstraction), while assuming the existence of nonlinear interactions, the absence of a central control, and emergent behavior. Nowadays, scientists belonging to di�erent communities use complex networks as a framework for dealing with their preferred research issues, from a theoretical and/or pratical perspective. This work is aimed at illustrating
some models, based on complex networks, deemed useful to represent social behaviors like competitive dynamics, groups formation, and emergence of linguistics phenomena