1731 research outputs found
Sort by
Crowd Motion Analysis: Segmentation, Anomaly Detection, and Behavior Classification
No full textThe objective of this doctoral study is to develop efficient techniques for flow segmentation, anomaly detection, and behavior classification in crowd scenes. Considering the complexities of occlusion, we focused our study on gathering the motion information at a higher scale, thus not associating it to single objects, but considering the crowd as a single entity. Firstly,we propose methods for flow segmentation based on correlation features, graph cut, Conditional Random Fields (CRF), enthalpy model, and particle mutual influence model. Secondly, methods based on deviant orientation information, Gaussian Mixture Model (GMM), and MLP neural network combined with GoodFeaturesToTrack are proposed to detect two types of anomalies. The first one detects deviant motion of the pedestrians compared to what has been observed beforehand. The second one detects panic situation by adopting the GMM and MLP to learn the behavior of the motion features extracted from a grid of particles and GoodFeaturesToTrack, respectively. Finally, we propose particle-driven and hybrid appraoches to classify the behaviors of crowd in terms of lane, arch/ring, bottleneck, blocking and fountainhead within a region of interest (ROI). For this purpose, the particle-driven approach extracts and fuses spatio-temporal features together. The spatial features represent the density of neighboring particles in the predefined proximity, whereas the temporal features represent the rendering of trajectories traveled by the particles. The hybrid approach exploits a thermal diffusion process combined with an extended variant of the social force model (SFM)
Modeling the interaction of light with photonic structures by direct numerical solution of Maxwell's equations
No full textThe present work analyzes and describes a method for the direct numerical solution of the Maxwell's equations of classical electromagnetism. This is the FDTD (Finite-Difference Time-Domain) method, along with its implementation in an "in-house" computing code for large parallelized simulations. Both are then applied to the modelization of photonic and plasmonic structures interacting with light. These systems are often too complex, either geometrically and materially, in order to be mathematically tractable and an exact analytic solution in closed form, or as a series expansion, cannot be obtained. The only way to gain insight on their physical behavior is thus to try to get a numerical approximated, although convergent, solution. This is a current trend in modern physics because, apart from perturbative methods and asymptotic analysis, which represent, where applicable, the typical instruments to deal with complex physico-mathematical problems, the only general way to approach such problems is based on the direct approximated numerical solution of the governing equations. Today this last choice is made possible through the enormous and widespread computational capabilities offered by modern computers, in particular High Performance Computing (HPC) done using parallel machines with a large number of CPUs working concurrently. Computer simulations are now a sort of virtual laboratories, which can be rapidly and costless setup to investigate various physical phenomena. Thus computational physics has become a sort of third way between the experimental and theoretical branches. The plasmonics application of the present work concerns the scattering and absorption analysis from single and arrayed metal nanoparticles, when surface plasmons are excited by an impinging beam of light, to study the radiation distribution inside a silicon substrate behind them. This has potential applications in improving the eciency of photovoltaic cells.
The photonics application of the present work concerns the analysis of the optical reflectance and transmittance properties of an opal crystal. This is a regular and ordered lattice of macroscopic particles which can stops light propagation in certain wavelenght bands, and whose study has potential applications in the realization of low threshold laser, optical waveguides and sensors. For these latters, in fact, the crystal response is tuned to its structure parameters and symmetry and varies by varying them. The present work about the FDTD method represents an enhacement of a previous one made for my MSc Degree Thesis in Physics, which has also now geared toward the visible and neighboring parts of the electromagnetic spectrum. It is organized in the following fashion. Part I provides an exposition of the basic concepts of electromagnetism which constitute the minimum, although partial, theoretical background useful to formulate the physics of the systems here analyzed or to be analyzed in possible further developments of the work. It summarizes Maxwell's equations in matter and the time domain description of temporally dispersive media. It addresses also the plane wave representation of an electromagnetic field distribution, mainly the far field one. The Kirchhoff formula is described and deduced, to calculate the angular radiation distribution around a scatterer. Gaussian beams in the paraxial approximation are also slightly treated, along with their focalization by means of an approximated diraction formula useful for their numericall FDTD representation. Finally, a thorough description of planarly multilayered media is included, which can play an important ancillary role in the homogenization procedure of a photonic crystal, as described in Part III, but also in other optical analyses. Part II properly concerns the FDTD numerical method description and implementation. Various aspects of the method are treated which globally contribute to a working and robust overall algorithm. Particular emphasis is given to those arguments representing an enhancement of previous work.These are: the analysis from existing literature of a new class of absorbing boundary conditions, the so called Convolutional-Perfectly Matched Layer,
and their implementation; the analysis from existing literature and implementation of the Auxiliary Differential Equation Method for the inclusion of frequency dependent electric permittivity media, according to various and
general polarization models; the description and implementation of a "plane wave injector" for representing impinging beam of lights propagating in an arbitrary direction, and which can be used to represent, by superposition, focalized beams; the parallelization of the FDTD numerical method by means of the Message Passing Interface (MPI) which, by using the here proposed,
suitable, user dened MPI data structures, results in a robust and scalable code, running on massively parallel High Performance Computing Machines like the IBM/BlueGeneQ with a core number of order 2X10^5. Finally, Part III gives the details of the specific plasmonics and photonics applications made with the "in-house" developed FDTD algorithm, to demonstrate its effectiveness. After Chapter 10, devoted to the validation of the FDTD code implementation against a known solution, Chapter 11 is about plasmonics, with the analytical and numerical study of single and arrayed metal nanoparticles of different shapes and sizes, when surface plasmon are excited on them by a light beam. The presence of a passivating embedding silica layer and a silicon substrate are also included. The next Chapter 12 is about the FDTD modelization of a face-cubic centered (FCC) opal photonic crystal sample, with a comparison between the numerical and experimental transmittance/reflectance behavior. An homogenization procedure is suggested of the lattice discontinuous crystal structure, by means of an averaging procedure and a planarly multilayered media analysis, through which better understand the reflecting characteristic of the crystal sample.
Finally, a procedure for the numerical reconstruction of the crystal dispersion banded omega-k curve inside the first Brillouin zone is proposed. Three Appendices providing details about specific arguments dealt with during the exposition conclude the work
Affective Analysis of Abstract Paintings Using Statistical Analysis and Art Theory
No full textThis research thesis aims to provide a novel approach to Emotion Recognition of Images: based on empirical studies, we employ the state-of-the-art computer vision techniques in order to understand what makes an abstract artwork emotional. We identify and quantify the emotional regions of abstract paintings. We also investigate the contributions of the main aspects present on abstract artworks (i.e., colour, shape and texture) to automatically predict emotional valence of them. By using eye-tracking recordings we investigate the link between the detected emotional content and the way people look at abstract paintings. We apply a bottom-up saliency model to compare with eye-tracking in order to predict the emotional salient regions of abstract paintings. Finally, we use the metadata associated to the paintings (e.g., title, description and/or artist statement) and correlate it with the emotional responses of the paintings. This research opens opportunity to understand why an abstract painting is perceived as emotional from global and local scales. Moreover, this work provides to art historians and art researches with a new perspective on the analysis of abstract paintings
Socially-Aware Interfaces for Supporting Co-located Interactions
No full textIn everyday life we engage in social interactions and use our body to communicate social actions and intentions in a seamless way. Our body, our behaviour and the behaviour of the people around us tell a lot about the social situations in which we are involved and of the social context in which we act. For example, the way people direct their attention while engaged in a conversation or the spatial distribution of a group during an activity can reveal information about the social context that characterises that particular social situation. Inspired by these observations, several researchers have investigated ways to automatically interpret social signals, using sensors and algorithms for detecting and modelling behavioural cues related to social interactions. With the increasing availability of sensors and technology in our environment, we believe that this research direction is highly relevant nowadays and presents an opportunity for further investigation of social signals in the HCI domain. In this dissertation, I discuss the design and evaluation of socially-aware systems, technologies that are sensitive to the social context in which they are deployed. The focus of this work is to explore interfaces that support engagement in co-located multi-user interactions, taking into account users’ nonverbal behaviour, including gaze, facial expressions, and body movements. In particular, the research goals are twofold: to understand which nonverbal cues and social signals reflect engagement in co-located group activities and to design systems that can utilise this information.
The thesis presents an in-depth analysis of the state of art in designing socially-aware systems. Moreover, it presents empirical studies that describe the design, implementation and assessment (both in real-world and laboratory settings) of such systems. This research has involved the development and evaluation of prototypes in their real context of use. Specifically, the thesis focuses on socially-aware systems in the form of two multi-user technologies meant to be used in different social contexts: a responsive display deployed in a public setting, and an ambient display for supporting conversations in small group discussions. Finally, based on the findings and insights gained through these studies, the thesis provides generalised characteristics for socially-aware systems and presents MOSAIC, a theoretical framework for framing the complexity of the design space of such technology. The final aim of this work is to support HCI researcher and practitioners in exploring the opportunities and limitations of technologies that react and respond to the social context
Causation in private enforcement of competition law: a comparative analysis of divergent national approaches
No full textCompetition law damages actions are often characterised by the uncertainty of the causal connection between the infringement and the harm. The anticompetitive damage consists in a pure economic loss caused by a market distortion. Here, the complexity of the market structures, combined with the interdependence of individuals’ assets, fuel this causal uncertainty. The recently adopted Damages Directive, in line with the decisions of the Court of Justice of the EU, established a regime based on the compensatory principle, consequently giving central importance to the definition and to the assessment of the causal connection between the infringement of competition law and the harm suffered. Surprisingly, the Directive provides no guidance to the assessment of the causal connection. By consequence, national courts apply domestic principles of causations, which are deeply rooted in their legal traditions. This thesis first addresses the concept of causation in competition law damages actions, discussing the main and more relevant approaches in tort law theory. Therefore, it describes the different solutions for the assessment of causation in competition law adopted by national courts and critically analyses the approach laid down by the European Union law and Courts. In order to examine in depth the reasons of the causal uncertainty in competition law damages actions, it delves into the analysis of the proof of causation and of the use of econometrics. The underlying comparative analysis serves, from a practical perspective, to observe how judges engage the problem of causal uncertainty in competition damages actions, while, from a theoretical standpoint, it addresses and complements the research of appropriate approaches
Antisense-mediated splicing correction approaches for retinal dystrophies and dysfunctions
No full textRetinal dystrophies are a large and extremely variegate group of diseases causing visual impairments and eventually blindness. As of today, no cures are available for such conditions, and the most advanced clinical approaches are based on classical gene transfer techniques. The recent development of genetic tools designed to manipulate splicing offers a unique opportunity to target several of these diseases, potentially reaching a highly specific and controlled effect in contrast to the more investigated gene supplementation therapies. These molecular tools are mainly part of two classes: antisense oligonucleotides (AONs) and chimeric or adapted small nuclear RNAs (snRNAs). They function by altering how the splicing machinery recognize a target sequence, usually an exon, allowing to increase its presence (exon inclusion) or decrease it (exon skipping) in mature mRNA. The first aim of this work has been the identification of genes causing retinal dystrophies having mutations potentially targetable by splicing-correction approaches. Following screening of mutation databases we identified three genes, each of them having mutations with characteristic interesting for a possible application of such approaches: CACNA2D4, RPGR, and USH2A. The CACNA2D4 gene encodes for an accessory subunit of high-voltage-activated (HVA) calcium channels and, when mutated, causes retinal cone dystrophy 4 (RCD4). It was selected because of the presence of a mouse model of retinal dystrophy (Wycisk et al., 2006a), allowing an in vivo test of the designed strategy. Since the approach consisted in inducing skipping of the exon hosting the mutation, we analysed the splicing pattern of the target exon in vivo and in vitro; the functionality of the rescued protein resulting from therapeutic exon skipping. The analysis revealed on one side that the rescued protein was not functional, therefore showing how the approach was not feasible for CACNA2D4. Interestingly, we found evidence of two newly identified splicing isoforms of the gene, one of which mimics the effect of the mutation. A mutation in RPGR intron 9 has been shown to cause retinitis pigmentosa by increasing the inclusion in mature mRNA of an alternatively spliced exon (E9a) (Neidhardt et al., 2007). Using splicing reporter minigene assays, we were able to design efficient chimeric U1snRNAs able to mediate E9a skipping thus correcting the genetic defect. In the USH2A gene we instead identified an interesting region hosting 5% of known mutation which could be approachable with exon skipping. The implementation of assays able to assess functionality of USH2A and RPGR will be fundamental for the development of new splicing-correction approaches for these genes
La scultura a Piacenza in età sforzesca
No full textLa ricerca si propone di studiare le sculture lapidee e fittili realizzate tra il 1450 e il 1521 circa per la città di Piacenza e il suo territorio. Dalla catalogazione delle opere e dalla ricerca d'archivio emergono, oltre a una panoramica sui principali cantieri ecclesiastici e nobiliari della città, le figure di numerosi scultori e botteghe impegnati nella realizzazione di statue, rilievi ed elementi architettonici
Laser cladding with metallic powders
No full textthe influence of the powder material and the main processing parameters on geometrical features and dilution of the clads is investigated and discussed. Physical and analytical model that allow the explanation of the process and the prediction of the clad geometry and dilution is discussed. Using these models, useful tools for cladding operators and engineers are proposed. The energetic balance of the process is presented. Energetic redistribution in laser cladding process is analysed in detail, and quantification of process efficiency and energy losses is given. The influence of the processing parameters and the chemical/physical
properties of the materials is considered throughout the various experiments performed. Some selected properties of the coating produced in different processing conditions are analysed. In specific, the variation of the chemical composition of the clad due to substrate dilution is considered, and its effect on the characteristics of the coatings is discussed
The Development of socio-moral Evaluations in human Infants and Preschoolers
No full textI will be presenting some empirical findings on the development of socio-moral evaluations in infants and preschoolers, this subject has had a long trajectory in the field of psychology in the last decades. For this reason, it is important that I describe some of the classical theories and recent literature that explains the development of morality. More specifically, I studied next to my collaborators: the emerging capacity of ‘the sense of fairness’, ‘harm’, ‘prosocial helping behaviors’, and the ‘logic of merit’. Our research supports a nativist viewpoint where the origins of justice, fairness, and equity are of an innate foundation rather than an acquired one
Modeling and Compostion of Environment-as-a-Service
No full textWireless-enabled electronic devices are becoming cheaper, more powerful and thus more popular. They include sensors, actuators, smartphones, tablets, wearable devices, and other complex devices. They can carry out complex tasks, cooperating with their ``neighbors''. However, it is difficult to develop mobile applications to exploit the full power of available resources because the computational capabilities on devices are not homogeneous, and their connectivity changes with physical movement. We propose a mobile environment model to describe the connected devices and study the structural and behavioral characteristics of the environments. Based on the model, we design the routing protocols and a language to support the composition of environments. We propose a framework to provide a unified, flexible and scalable service for task/process deployment and execution on top of the heterogeneous and dynamic mobile environments. We compare different architectures, and discuss the optimization of resources discovery and routing algorithm. A proof-of-concept framework is implemented and shows the feasibility of our Environment-as-a-Service approach. Finally, we explore the theoretical principles and practical techniques for performance optimization, including a data prefetching technique and a dynamic process/task allocation algorithm