A three-dimensional representation method for noisy point clouds based on growing self-organizing maps accelerated on GPUs

The research described in this thesis was motivated by the need of a robust model capable of representing 3D data obtained with 3D sensors, which are inherently noisy. In addition, time constraints have to be considered as these sensors are capable of providing a 3D data stream in real time. This thesis proposed the use of Self-Organizing Maps (SOMs) as a 3D representation model. In particular, we proposed the use of the Growing Neural Gas (GNG) network, which has been successfully used for clustering, pattern recognition and topology representation of multi-dimensional data. Until now, Self-Organizing Maps have been primarily computed offline and their application in 3D data has mainly focused on free noise models, without considering time constraints. It is proposed a hardware implementation leveraging the computing power of modern GPUs, which takes advantage of a new paradigm coined as General-Purpose Computing on Graphics Processing Units (GPGPU). The proposed methods were applied to different problem and applications in the area of computer vision such as the recognition and localization of objects, visual surveillance or 3D reconstruction.

A novel three-dimensional imaging array recording and computational method by means of coded cameras reconstruction

In this paper, we propose a novel three-dimensional imaging method by which the object is captured by a coded cameras array (CCA) and computationally reconstructed as a series of longitudinal layered surface images of the object. The distribution of cameras in array, named code pattern, is crucial for reconstructed images fidelity when the correlation decoding is used. We use DIRECT global optimization algorithm to design the code patterns that possess proper imaging property. We have conducted primary experiments to verify and test the performance of the proposed method with a simple discontinuous object and a small-scale CCA including nine cameras. After certain procedures such as capturing, photograph integrating, computational reconstructing and filtering, etc., we obtain reconstructed longitudinal layered surface images of the object with higher signal-to-noise ratio. The results of experiments show that the proposed method is feasible. It is a promising method to be used in fields such as remote sensing, machine vision, etc. (c) 2006 Elsevier GmbH. All rights reserved.

Determination of viral load and integration status of HPV 16 in normal and LSIL exfoliated cervical cells

L’intégration du génome du virus papilloma humain (VPH) a été reconnu jusqu’`a récemment comme étant un événnement fréquent mais pourtant tardif dans la progression de la maladie du col de l’utérus. La perspective temporelle vient, pourtant, d’être mise au défi par la détection de formes intégrées de VPH dans les tissus normaux et dans les lésions prénéoplasiques. Notre objectif était de déterminer la charge virale de VPH-16 et son état physique dans une série de 220 échantillons provenant de cols uterins normaux et avec des lésions de bas-grade. La technique quantitative de PCR en temps réel, méthode Taqman, nous a permis de quantifier le nombre de copies des gènes E6, E2, et de la B-globine, permettant ainsi l’évaluation de la charge virale et le ratio de E6/E2 pour chaque spécimen. Le ratio E6/E2 de 1.2 ou plus était suggestif d’intégration. Par la suite, le site d’intégration du VPH dans le génome humain a été déterminé par la téchnique de RS-PCR. La charge virale moyenne était de 57.5±324.6 copies d'ADN par cellule et le ratio E6/E2 a évalué neuf échantillons avec des formes d’HPV intégrées. Ces intégrants ont été amplifiés par RS-PCR, suivi de séquençage, et l’homologie des amplicons a été déterminée par le programme BLAST de NCBI afin d’identifier les jonctions virales-humaines. On a réussi `a identifier les jonctions humaines-virales pour le contrôle positif, c'est-à-dire les cellules SiHa, pourtant nous n’avons pas detecté d’intégration par la technique de RS-PCR dans les échantillons de cellules cervicales exfoliées provenant de tissus normaux et de lésions de bas-grade. Le VPH-16 est rarement intégré dans les spécimens de jeunes patientes.

Parallel Monte Carlo approach for integration of the rendering equation

This paper is addressed to the numerical solving of the rendering equation in realistic image creation. The rendering equation is integral equation describing the light propagation in a scene accordingly to a given illumination model. The used illumination model determines the kernel of the equation under consideration. Nowadays, widely used are the Monte Carlo methods for solving the rendering equation in order to create photorealistic images. In this work we consider the Monte Carlo solving of the rendering equation in the context of the parallel sampling scheme for hemisphere. Our aim is to apply this sampling scheme to stratified Monte Carlo integration method for parallel solving of the rendering equation. The domain for integration of the rendering equation is a hemisphere. We divide the hemispherical domain into a number of equal sub-domains of orthogonal spherical triangles. This domain partitioning allows to solve the rendering equation in parallel. It is known that the Neumann series represent the solution of the integral equation as a infinity sum of integrals. We approximate this sum with a desired truncation error (systematic error) receiving the fixed number of iteration. Then the rendering equation is solved iteratively using Monte Carlo approach. At each iteration we solve multi-dimensional integrals using uniform hemisphere partitioning scheme. An estimate of the rate of convergence is obtained using the stratified Monte Carlo method. This domain partitioning allows easy parallel realization and leads to convergence improvement of the Monte Carlo method. The high performance and Grid computing of the corresponding Monte Carlo scheme are discussed.

Efficient calculation of two-dimensional periodic and waveguide acoustic Green’s functions

New representations and efficient calculation methods are derived for the problem of propagation from an infinite regularly spaced array of coherent line sources above a homogeneous impedance plane, and for the Green's function for sound propagation in the canyon formed by two infinitely high, parallel rigid or sound soft walls and an impedance ground surface. The infinite sum of source contributions is replaced by a finite sum and the remainder is expressed as a Laplace-type integral. A pole subtraction technique is used to remove poles in the integrand which lie near the path of integration, obtaining a smooth integrand, more suitable for numerical integration, and a specific numerical integration method is proposed. Numerical experiments show highly accurate results across the frequency spectrum for a range of ground surface types. It is expected that the methods proposed will prove useful in boundary element modeling of noise propagation in canyon streets and in ducts, and for problems of scattering by periodic surfaces.

A Nyström method for a class of integral equations on the real line with applications to scattering by diffraction gratings and rough surfaces

We propose a Nystr¨om/product integration method for a class of second kind integral equations on the real line which arise in problems of two-dimensional scalar and elastic wave scattering by unbounded surfaces. Stability and convergence of the method is established with convergence rates dependent on the smoothness of components of the kernel. The method is applied to the problem of acoustic scattering by a sound soft one-dimensional surface which is the graph of a function f, and superalgebraic convergence is established in the case when f is infinitely smooth. Numerical results are presented illustrating this behavior for the case when f is periodic (the diffraction grating case). The Nystr¨om method for this problem is stable and convergent uniformly with respect to the period of the grating, in contrast to standard integral equation methods for diffraction gratings which fail at a countable set of grating periods.

Computer simulations of two-dimensional colloidal crystals under confinement and shear

In this thesis we are presenting a broadly based computer simulation study of two-dimensional colloidal crystals under different external conditions. In order to fully understand the phenomena which occur when the system is being compressed or when the walls are being sheared, it proved necessary to study also the basic motion of the particles and the diffusion processes which occur in the case without these external forces. In the first part of this thesis we investigate the structural transition in the number of rows which occurs when the crystal is being compressed by placing the structured walls closer together. Previous attempts to locate this transition were impeded by huge hysteresis effects. We were able to determine the transition point with higher precision by applying both the Schmid-Schilling thermodynamic integration method and the phase switch Monte Carlo method in order to determine the free energies. These simulations showed not only that the phase switch method can successfully be applied to systems with a few thousand particles and a soft crystalline structure with a superimposed pattern of defects, but also that this method is way more efficient than a thermodynamic integration when free energy differences are to be calculated. Additionally, the phase switch method enabled us to distinguish between several energetically very similar structures and to determine which one of them was actually stable. Another aspect considered in the first result chapter of this thesis is the ensemble inequivalence which can be observed when the structural transition is studied in the NpT and in the NVT ensemble. The second part of this work deals with the basic motion occurring in colloidal crystals confined by structured walls. Several cases are compared where the walls are placed in different positions, thereby introducing an incommensurability into the crystalline structure. Also the movement of the solitons, which are created in the course of the structural transition, is investigated. Furthermore, we will present results showing that not only the well-known mechanism of vacancies and interstitial particles leads to diffusion in our model system, but that also cooperative ring rotation phenomena occur. In this part and the following we applied Langevin dynamics simulations. In the last chapter of this work we will present results on the effect of shear on the colloidal crystal. The shear was implemented by moving the walls with constant velocity. We have observed shear banding and, depending on the shear velocity, that the inner part of the crystal breaks into several domains with different orientations. At very high shear velocities holes are created in the structure, which originate close to the walls, but also diffuse into the inner part of the crystal.

Comparison of dynamic effects of high-speed traffic load on ballasted track using a simplified two-dimensional and full three-dimensional model.

The vertical dynamic actions transmitted by railway vehicles to the ballasted track infrastructure is evaluated taking into account models with different degree of detail. In particular, we have studied this matter from a two-dimensional (2D) finite element model to a fully coupled three-dimensional (3D) multi-body finite element model. The vehicle and track are coupled via a non-linear Hertz contact mechanism. The method of Lagrange multipliers is used for the contact constraint enforcement between wheel and rail. Distributed elevation irregularities are generated based on power spectral density (PSD) distributions which are taken into account for the interaction. The numerical simulations are performed in the time domain, using a direct integration method for solving the transient problem due to the contact nonlinearities. The results obtained include contact forces, forces transmitted to the infrastructure (sleeper) by railpads and envelopes of relevant results for several track irregularities and speed ranges. The main contribution of this work is to identify and discuss coincidences and differences between discrete 2D models and continuum 3D models, as wheel as assessing the validity of evaluating the dynamic loading on the track with simplified 2D models

BM+-tree: A hyperplane-based index method for high-dimensional metric spaces

In this paper, we propose a novel high-dimensional index method, the BM+-tree, to support efficient processing of similarity search queries in high-dimensional spaces. The main idea of the proposed index is to improve data partitioning efficiency in a high-dimensional space by using a rotary binary hyperplane, which further partitions a subspace and can also take advantage of the twin node concept used in the M+-tree. Compared with the key dimension concept in the M+-tree, the binary hyperplane is more effective in data filtering. High space utilization is achieved by dynamically performing data reallocation between twin nodes. In addition, a post processing step is used after index building to ensure effective filtration. Experimental results using two types of real data sets illustrate a significantly improved filtering efficiency.

Mixed-dimensional consistent coupling by multi-point constraint equations for efficient multi-scale modeling

Finding an appropriate linking method to connect different dimensional element types in a single finite element model is a key issue in the multi-scale modeling. This paper presents a mixed dimensional coupling method using multi-point constraint equations derived by equating the work done on either side of interface connecting beam elements and shell elements for constructing a finite element multiscale model. A typical steel truss frame structure is selected as case example and the reduced scale specimen of this truss section is then studied in the laboratory to measure its dynamic and static behavior in global truss and local welded details while the different analytical models are developed for numerical simulation. Comparison of dynamic and static response of the calculated results among different numerical models as well as the good agreement with those from experimental results indicates that the proposed multi-scale model is efficient and accurate.

An improved computational method in structural dynamics

Purpose – In structural, earthquake and aeronautical engineering and mechanical vibration, the solution of dynamic equations for a structure subjected to dynamic loading leads to a high order system of differential equations. The numerical methods are usually used for integration when either there is dealing with discrete data or there is no analytical solution for the equations. Since the numerical methods with more accuracy and stability give more accurate results in structural responses, there is a need to improve the existing methods or develop new ones. The paper aims to discuss these issues. Design/methodology/approach – In this paper, a new time integration method is proposed mathematically and numerically, which is accordingly applied to single-degree-of-freedom (SDOF) and multi-degree-of-freedom (MDOF) systems. Finally, the results are compared to the existing methods such as Newmark’s method and closed form solution. Findings – It is concluded that, in the proposed method, the data variance of each set of structural responses such as displacement, velocity, or acceleration in different time steps is less than those in Newmark’s method, and the proposed method is more accurate and stable than Newmark’s method and is capable of analyzing the structure at fewer numbers of iteration or computation cycles, hence less time-consuming. Originality/value – A new mathematical and numerical time integration method is proposed for the computation of structural responses with higher accuracy and stability, lower data variance, and fewer numbers of iterations for computational cycles.

GPU accelerated algorithms for computing matrix function vector products with applications to exponential integrators and fractional diffusion

The efficient computation of matrix function vector products has become an important area of research in recent times, driven in particular by two important applications: the numerical solution of fractional partial differential equations and the integration of large systems of ordinary differential equations. In this work we consider a problem that combines these two applications, in the form of a numerical solution algorithm for fractional reaction diffusion equations that after spatial discretisation, is advanced in time using the exponential Euler method. We focus on the efficient implementation of the algorithm on Graphics Processing Units (GPU), as we wish to make use of the increased computational power available with this hardware. We compute the matrix function vector products using the contour integration method in [N. Hale, N. Higham, and L. Trefethen. Computing Aα, log(A), and related matrix functions by contour integrals. SIAM J. Numer. Anal., 46(5):2505–2523, 2008]. Multiple levels of preconditioning are applied to reduce the GPU memory footprint and to further accelerate convergence. We also derive an error bound for the convergence of the contour integral method that allows us to pre-determine the appropriate number of quadrature points. Results are presented that demonstrate the effectiveness of the method for large two-dimensional problems, showing a speedup of more than an order of magnitude compared to a CPU-only implementation.

Two-scale computational modelling of water flow in unsaturated soils containing irregular-shaped inclusions

The focus of this paper is two-dimensional computational modelling of water flow in unsaturated soils consisting of weakly conductive disconnected inclusions embedded in a highly conductive connected matrix. When the inclusions are small, a two-scale Richards’ equation-based model has been proposed in the literature taking the form of an equation with effective parameters governing the macroscopic flow coupled with a microscopic equation, defined at each point in the macroscopic domain, governing the flow in the inclusions. This paper is devoted to a number of advances in the numerical implementation of this model. Namely, by treating the micro-scale as a two-dimensional problem, our solution approach based on a control volume finite element method can be applied to irregular inclusion geometries, and, if necessary, modified to account for additional phenomena (e.g. imposing the macroscopic gradient on the micro-scale via a linear approximation of the macroscopic variable along the microscopic boundary). This is achieved with the help of an exponential integrator for advancing the solution in time. This time integration method completely avoids generation of the Jacobian matrix of the system and hence eases the computation when solving the two-scale model in a completely coupled manner. Numerical simulations are presented for a two-dimensional infiltration problem.

The fully implicit stochastic-alpha method for stiff stochastic differential equations

A fully implicit integration method for stochastic differential equations with significant multiplicative noise and stiffness in both the drift and diffusion coefficients has been constructed, analyzed and illustrated with numerical examples in this work. The method has strong order 1.0 consistency and has user-selectable parameters that allow the user to expand the stability region of the method to cover almost the entire drift-diffusion stability plane. The large stability region enables the method to take computationally efficient time steps. A system of chemical Langevin equations simulated with the method illustrates its computational efficiency.

Negative sequence compensation within fundamental positive sequence reference frame for a stiff micro-grid generation in a wind power system using slip ring induction machine

An isolated wind power generation scheme using slip ring induction machine (SRIM) is proposed. The proposed scheme maintains constant load voltage and frequency irrespective of the wind speed or load variation. The power circuit consists of two back-to-back connected inverters with a common dc link, where one inverter is directly connected to the rotor side of SRIM and the other inverter is connected to the stator side of the SRIM through LC filter. Developing a negative sequence compensation method to ensure that, even under the presence of unbalanced load, the generator experiences almost balanced three-phase current and most of the unbalanced current is directed through the stator side converter is the focus here. The SRIM controller varies the speed of the generator with variation in the wind speed to extract maximum power. The difference of the generated power and the load power is either stored in or extracted from a battery bank, which is interfaced to the common dc link through a multiphase bidirectional fly-back dc-dc converter. The SRIM control scheme, maximum power point extraction algorithm and the fly-back converter topology are incorporated from available literature. The proposed scheme is both simulated and experimentally verified.