Algorithms For Automatic And Robust Registration Of 3D Head Scans

wo methods for registering laser-scans of human heads and transforming them to a new semantically consistent topology defined by a user-provided template mesh are described. Both algorithms are stated within the Iterative Closest Point framework. The first method is based on finding landmark correspondences by iteratively registering the vicinity of a landmark with a re-weighted error function. Thin-plate spline interpolation is then used to deform the template mesh and finally the scan is resampled in the topology of the deformed template. The second algorithm employs a morphable shape model, which can be computed from a database of laser-scans using the first algorithm. It directly optimizes pose and shape of the morphable model. The use of the algorithm with PCA mixture models, where the shape is split up into regions each described by an individual subspace, is addressed. Mixture models require either blending or regularization strategies, both of which are described in detail. For both algorithms, strategies for filling in missing geometry for incomplete laser-scans are described. While an interpolation-based approach can be used to fill in small or smooth regions, the model-driven algorithm is capable of fitting a plausible complete head mesh to arbitrarily small geometry, which is known as "shape completion". The importance of regularization in the case of extreme shape completion is shown.

Constructing a user experience user experience-based mobile learning environment

Mobile learning, in the past defined as learning with mobile devices, now refers to any type of learning-on-the-go or learning that takes advantage of mobile technologies. This new definition shifted its focus from the mobility of technology to the mobility of the learner (O'Malley and Stanton 2002; Sharples, Arnedillo-Sanchez et al. 2009). Placing emphasis on the mobile learner’s perspective requires studying “how the mobility of learners augmented by personal and public technology can contribute to the process of gaining new knowledge, skills, and experience” (Sharples, Arnedillo-Sanchez et al. 2009). The demands of an increasingly knowledge based society and the advances in mobile phone technology are combining to spur the growth of mobile learning. Around the world, mobile learning is predicted to be the future of online learning, and is slowly entering the mainstream education. However, for mobile learning to attain its full potential, it is essential to develop more advanced technologies that are tailored to the needs of this new learning environment. A research field that allows putting the development of such technologies onto a solid basis is user experience design, which addresses how to improve usability and therefore user acceptance of a system. Although there is no consensus definition of user experience, simply stated it focuses on how a person feels about using a product, system or service. It is generally agreed that user experience adds subjective attributes and social aspects to a space that has previously concerned itself mainly with ease-of-use. In addition, it can include users’ perceptions of usability and system efficiency. Recent advances in mobile and ubiquitous computing technologies further underline the importance of human-computer interaction and user experience (feelings, motivations, and values) with a system. Today, there are plenty of reports on the limitations of mobile technologies for learning (e.g., small screen size, slow connection), but there is a lack of research on user experience with mobile technologies. This dissertation will fill in this gap by a new approach in building a user experience-based mobile learning environment. The optimized user experience we suggest integrates three priorities, namely a) content, by improving the quality of delivered learning materials, b) the teaching and learning process, by enabling live and synchronous learning, and c) the learners themselves, by enabling a timely detection of their emotional state during mobile learning. In detail, the contributions of this thesis are as follows: • A video codec optimized for screencast videos which achieves an unprecedented compression rate while maintaining a very high video quality, and a novel UI layout for video lectures, which together enable truly mobile access to live lectures. • A new approach in HTTP-based multimedia delivery that exploits the characteristics of live lectures in a mobile context and enables a significantly improved user experience for mobile live lectures. • A non-invasive affective learning model based on multi-modal emotion detection with very high recognition rates, which enables real-time emotion detection and subsequent adaption of the learning environment on mobile devices. The technology resulting from the research presented in this thesis is in daily use at the School of Continuing Education of Shanghai Jiaotong University (SOCE), a blended-learning institution with 35.000 students.

KrigInv: An efficient and user-friendly implementation of batch-sequential inversion strategies based on kriging

Several strategies relying on kriging have recently been proposed for adaptively estimating contour lines and excursion sets of functions under severely limited evaluation budget. The recently released R package KrigInv 3 is presented and offers a sound implementation of various sampling criteria for those kinds of inverse problems. KrigInv is based on the DiceKriging package, and thus benefits from a number of options concerning the underlying kriging models. Six implemented sampling criteria are detailed in a tutorial and illustrated with graphical examples. Different functionalities of KrigInv are gradually explained. Additionally, two recently proposed criteria for batch-sequential inversion are presented, enabling advanced users to distribute function evaluations in parallel on clusters or clouds of machines. Finally, auxiliary problems are discussed. These include the fine tuning of numerical integration and optimization procedures used within the computation and the optimization of the considered criteria.

An anisotropic elastic-viscoplastic damage model for bone tissue

A new anisotropic elastic-viscoplastic damage constitutive model for bone is proposed using an eccentric elliptical yield criterion and nonlinear isotropic hardening. A micromechanics-based multiscale homogenization scheme proposed by Reisinger et al. is used to obtain the effective elastic properties of lamellar bone. The dissipative process in bone is modeled as viscoplastic deformation coupled to damage. The model is based on an orthotropic ecuntric elliptical criterion in stress space. In order to simplify material identification, an eccentric elliptical isotropic yield surface was defined in strain space, which is transformed to a stress-based criterion by means of the damaged compliance tensor. Viscoplasticity is implemented by means of the continuous Perzyna formulation. Damage is modeled by a scalar function of the accumulated plastic strain D(κ) , reducing all element s of the stiffness matrix. A polynomial flow rule is proposed in order to capture the rate-dependent post-yield behavior of lamellar bone. A numerical algorithm to perform the back projection on the rate-dependent yield surface has been developed and implemented in the commercial finite element solver Abaqus/Standard as a user subroutine UMAT. A consistent tangent operator has been derived and implemented in order to ensure quadratic convergence. Correct implementation of the algorithm, convergence, and accuracy of the tangent operator was tested by means of strain- and stress-based single element tests. A finite element simulation of nano- indentation in lamellar bone was finally performed in order to show the abilities of the newly developed constitutive model.

Oxidative-nitrosative stress and systemic vascular function in highlanders with and without exaggerated hypoxemia

BACKGROUND Acute exposure to high altitude stimulates free radical formation in lowlanders, yet whether this persists during chronic exposure in healthy, well-adapted and maladapted highlanders suffering from chronic mountain sickness (CMS) remains to be established. METHODS Oxidative-nitrosative stress (as determined by the presence of the biomarkers ascorbate radical [A •- ], via electron paramagnetic resonance spectroscopy, and nitrite [NO 2 2 ], via ozone-based chemiluminescence) was assessed in venous blood of 25 male highlanders in Bolivia living at 3,600 m with CMS (n 5 13, CMS 1 ) and without CMS (n 5 12, CMS 2 ). Twelve age- and activity-matched, healthy, male lowlanders were examined at sea level and during acute hypoxia. We also measured fl ow-mediated dilatation (FMD), arterial stiffness defined by augmentation index normalized for a heart rate of 75 beats/min (AIx-75), and carotid intima-media thickness (IMT). RESULTS Compared with normoxic lowlanders, oxidative-nitrosative stress was moderately increased in the CMS 2 group ( P , .05), as indicated by elevated A •- (3,191 457 arbitrary units [AU] vs 2,640 445 AU) and lower NO 2 2 (206 55 nM vs 420 128 nM), whereas vascular function remained preserved. This was comparable to that observed during acute hypoxia in lowlanders in whom vascular dysfunction is typically observed. In contrast, this response was markedly exaggerated in CMS 1 group (A •- , 3,765 429 AU; NO 2 2 , 148 50 nM) compared with both the CMS 2 group and lowlanders ( P , .05). This was associated with systemic vascular dysfunction as indicated by lower ( P , .05 vs CMS 2 ) FMD (4.2% 0.7% vs 7.6% 1.7%) and increased AIx-75 (23% 8% vs 12% 7%) and carotid IMT (714 127 m M vs 588 94 m M). CONCLUSIONS Healthy highlanders display a moderate, sustained elevation in oxidative-nitrosative stress that, unlike the equivalent increase evoked by acute hypoxia in healthy lowlanders, failed to affect vascular function. Its more marked elevation in patients with CMS may contribute to systemic vascular dysfunction.

Function and formation of $\beta$1,4-galactosyltransferase-specific adhesions during growth and differentiation of F9 embryonal carcinoma cells

Galactosyltransferase (GalTase) is localized in the Golgi, where it functions in oligosaccharide synthesis, as well as on the cell surface where it serves as a cell adhesion molecule. GalTase-specific adhesions are functional in a number of important biological events, including F9 embryonal carcinoma (EC) cell adhesions. GalTase-based adhesions are formed by recognition and binding to terminal N-acetylglucosamine (GlcNAc) residues on its glycoprotein counterpart on adjacent cell surfaces. The object of this work has been to investigate the formation and function of GalTase-specific adhesions during F9 cell growth and differentiation. We initially investigated GalTase synthesis during differentiation and found that the increase in GalTase activity was specific for the Golgi compartment; surface GalTase levels remained constant during differentiation. These data indicated that the increase in cell adhesions expected with increased cell-matrix interaction in differentiated F9 cells is not the consequence of increased surface GalTase expression and, more interestingly, that the two pools of GalTase are under differential regulation. Synthesis and recognition of the consociate glycoprotein component was next investigated. Surface GalTase recognized several surface glycoproteins in a pattern that changes with differentiation. Uvomorulin, lysosome-associated membrane protein-1 (LAMP-1), and laminin were recognized by surface GalTase and are, therefore, potential components in GalTase-specific adhesions. Furthermore, these interactions were aberrant in an adhesion-defective F9 cell line that results, at least in part, from abnormal oligosaccharide synthesis. The function played by surface GalTase in growth and induction of differentiation was examined. Inhibition of surface GalTase function by a panel of reagents inhibited F9 cell growth. GalTase expression at both the transcription and protein levels were differentially regulated during the cell cycle, with surface expression greatest in the G1 phase. Disruption of GalTase adhesion by exposure to anti-GalTase antibodies during this period resulted in extension of the G2 phase, a result similar to that seen with agents known to inhibit growth and induce differentiation. Finally, other studies have suggested that a subset of cell adhesion molecules have the capability to induce differentiation in EC cells systems. We have determined in F9 cells that dissociating GalTase adhesion by galactosylation of and release of the consociate glycoproteins induces differentiation, as defined by increased laminin synthesis. The ability to induce differentiation by surface galactosylation was greatest in cells grown in cultures promoting cell-cell adhesions, relative to cultures with minimal cell-cell interactions. ^

Structure-function analyses of {\it PAX6\/} and the molecular bases of aniridia

PAX6 is a transcription activator that regulates eye development in animals ranging from Drosophila to human. The C-terminal region of PAX6 is proline/serine/threonine-rich (PST) and functions as a potent transactivation domain when attached to a heterologous DNA-binding domain of the yeast transcription factor, GAL4. The PST region comprises 152 amino acids encoded by four exons. The transactivation function of the PST region has not been defined and characterized in detail by in vitro mutagenesis. I dissected the PST domain in two independent systems, a heterologous system using a GAL4 DNA-binding site and the native system of PAX6. In both systems, the results show consistently that all four constituent exons of the PST domain are responsible for the transactivation function. The four exon fragments act cooperatively to stimulate transcription, although none of them can function individually as an independent transactivation domain. Combinations of two or more exon fragments can reconstitute substantial transactivation activity when fused to the DNA-binding domain of GAL4, but they surprisingly do not produce much activity in the context of native PAX6 even though the mutant PAX6 proteins are stable and their DNA-binding function remains unaffected. I conclude that the PAX6 protein contains an unusually large transactivation domain that is evolutionarily conserved to a high degree, and that its full transactivation activity relies on the cooperative action of the four exon fragments.^ Most PAX6 mutations detected in patients with aniridia result in truncations of the protein. Some of the truncation mutations occur in the PST region of PAX6, resulting in mutant proteins that retain their DNA-binding ability but have no significant transactivation activity. It is not clear whether such mutants are true loss-of-function or dominant-negative mutants. I show that these mutants are dominant-negative if they are coexpressed with wild-type PAX6 in cultured cells and that the dominant-negative effects result from enhanced DNA-binding ability of these mutants due to removal of the PST domain. These mutants are able to repress the wild-type PAX6 activity not only at target genes with paired domain binding sites but also at target genes with homeodomain binding sites.^ Mutations in the human PAX6 gene produce various phenotypes, including aniridia, Peters' anomaly, autosomal dominant keratitis, and familial foveal dysplasia. The various phenotypes may arise from different mutations in the same gene. To test this theory, I performed a functional analysis of two missense mutations in the paired domain: the R26G mutation reported in a case of Peters' anomaly, and the I87R mutation identified in a patient with aniridia. While both the R26 and the I87 positions are conserved in the paired boxes of all known PAX genes, X-ray crystallography has shown that only R26 makes contact with DNA. I found that the R26G mutant failed to bind a subset of paired domain binding sites but, surprisingly, bound other sites and successfully transactivated promoters containing those sites. In contrast, the I87R mutant had lost the ability to bind DNA at all tested sites and failed to transactivate promoters. My data support the haploinsufficiency hypothesis of aniridia, and the hypothesis that R26G is a hypomorphic allele. ^

Misunderstanding Creativity : User Created Content in Virtual Worlds and its Constraints by Code and Law

Virtual worlds have moved from being a geek topic to one of mainstream academic interest. This transition is contingent not only on the augmented economic, societal and cultural value of these virtual realities and their effect upon real life but also on their convenience as fields for experimentation, for testing models and paradigms. User creation is however not something that has been transplanted from the real to the virtual world but a phenomenon and a dynamic process that happens from within and is defined through complex relationships between commercial and non-commercial, commodified and not commodified, individual and of the community, amateur and professional, art and not art. Accounting for this complex environment, the present paper explores user created content in virtual worlds, its dimensions and value and above all, its constraints by code and law. It puts forward suggestions for better understanding and harnessing this creativity.

Passively Track WiFi Users with an Enhanced Particle Filter using Power-based Ranging

Passive positioning systems produce user location information for third-party providers of positioning services. Since the tracked wireless devices do not participate in the positioning process, passive positioning can only rely on simple, measurable radio signal parameters, such as timing or power information. In this work, we provide a passive tracking system for WiFi signals with an enhanced particle filter using fine-grained power-based ranging. Our proposed particle filter provides an improved likelihood function on observation parameters and is equipped with a modified coordinated turn model to address the challenges in a passive positioning system. The anchor nodes for WiFi signal sniffing and target positioning use software defined radio techniques to extract channel state information to mitigate multipath effects. By combining the enhanced particle filter and a set of enhanced ranging methods, our system can track mobile targets with an accuracy of 1.5m for 50% and 2.3m for 90% in a complex indoor environment. Our proposed particle filter significantly outperforms the typical bootstrap particle filter, extended Kalman filter and trilateration algorithms.

Renal function can be impaired in children with primary hyperoxaluria type 3

BACKGROUND Primary hyperoxaluria type 3 (PH3) is characterized by mutations in the 4-hydroxy-2-oxoglutarate aldolase (HOGA1) gene. PH3 patients are believed to present with a less severe phenotype than those with PH1 and PH2, but the clinical characteristics of PH3 patients have yet to be defined in sufficient detail. The aim of this study was to report our experience with PH3. METHODS Genetic analysis of HOGA1 was performed in patients with a high clinical suspicion of PH after the presence of mutations in the alanine-glyoxylate aminotransferase gene had been ruled out. Clinical, biochemical and genetic data of the seven patients identified with HOGA1 mutations were subsequently retrospectively reviewed. RESULTS Among the seven patients identified with HOGA1 mutations the median onset of clinical symptoms was 1.8 (range 0.4-9.8) years. Five patients initially presented with urolithiasis, and two other patients presented with urinary tract infection. All patients experienced persistent hyperoxaluria. Seven mutations were found in HOGA1, including two previously unreported ones, c.834 + 1G > T and c.3G > A. At last follow-up, two patients had impaired renal function based on estimated glomerular filtration rates (GFRs) of 77 and 83 mL/min per 1.73 m(2), respectively. CONCLUSIONS We found that the GFR was significantly impaired in two of our seven patients with PH3 diagnosed during childhood. This finding is in contrast to the early-impaired renal function in PH1 and PH2 and appears to refute to preliminary reassuring data on renal function in PH3.

Flow and fracture behaviour of high performance alloys

Based on our needs, that is to say, through precise simulation of the impact phenomena that may occur inside a jet engine turbine with an explicit non-linear finite element code, four new material models are postulated. Each one of is calibrated for four high-performance alloys that can be encountered in a modern jet engine. A new uncoupled material model for high strain and ballistic is proposed. Based on a Johnson-Cook type model, the proposed formulation introduces the effect of the third deviatoric invariant by means of three different Lode angle dependent functions. The Lode dependent functions are added to both plasticity and failure models. The postulated model is calibrated for a 6061-T651 aluminium alloy with data taken from the literature. The fracture pattern predictability of the JCX material model is shown performing numerical simulations of various quasi-static and dynamic tests. As an extension of the above-mentioned model, a modification in the thermal softening behaviour due to phase transformation temperatures is developed (JCXt). Additionally, a Lode angle dependent flow stress is defined. Analysing the phase diagram and high temperature tests performed, phase transformation temperatures of the FV535 stainless steel are determined. The postulated material model constants for the FV535 stainless steel are calibrated. A coupled elastoplastic-damage material model for high strain and ballistic applications is presented (JCXd). A Lode angle dependent function is added to the equivalent plastic strain to failure definition of the Johnson-Cook failure criterion. The weakening in the elastic law and in the Johnson-Cook type constitutive relation implicitly introduces the Lode angle dependency in the elastoplastic behaviour. The material model is calibrated for precipitation hardened Inconel 718 nickel-base superalloy. The combination of a Lode angle dependent failure criterion with weakened constitutive equations is proven to predict fracture patterns of the mechanical tests performed and provide reliable results. A transversely isotropic material model for directionally solidified alloys is presented. The proposed yield function is based a single linear transformation of the stress tensor. The linear operator weighs the degree of anisotropy of the yield function. The elastic behaviour, as well as the hardening, are considered isotropic. To model the hardening, a Johnson-Cook type relation is adopted. A material vector is included in the model implementation. The failure is modelled with the Cockroft-Latham failure criterion. The material vector allows orienting the reference orientation in any other that the user may need. The model is calibrated for the MAR-M 247 directionally solidified nickel-base superalloy.

User-definable resource usage bounds analysis for java bytecode

Automatic cost analysis of programs has been traditionally concentrated on a reduced number of resources such as execution steps, time, or memory. However, the increasing relevance of analysis applications such as static debugging and/or certiflcation of user-level properties (including for mobile code) makes it interesting to develop analyses for resource notions that are actually application-dependent. This may include, for example, bytes sent or received by an application, number of files left open, number of SMSs sent or received, number of accesses to a datábase, money spent, energy consumption, etc. We present a fully automated analysis for inferring upper bounds on the usage that a Java bytecode program makes of a set of application programmer-deflnable resources. In our context, a resource is defined by programmer-provided annotations which state the basic consumption that certain program elements make of that resource. From these deflnitions our analysis derives functions which return an upper bound on the usage that the whole program (and individual blocks) make of that resource for any given set of input data sizes. The analysis proposed is independent of the particular resource. We also present some experimental results from a prototype implementation of the approach covering a signiflcant set of interesting resources.

Implementation of wireless communication based on Software Defined Radio

In current communication systems, there are many new challenges like various competitive standards, the scarcity of frequency resource, etc., especially the development of personal wireless communication systems result the new system update faster than ever before, the conventional hardware-based wireless communication system is difficult to adapt to this situation. The emergence of SDR enabled the third revolution of wireless communication which from hardware to software and build a flexible, reliable, upgradable, reusable, reconfigurable and low cost platform. The Universal Software Radio Peripheral (USRP) products are commonly used with the GNU Radio software suite to create complex SDR systems. GNU Radio is a toolkit where digital signal processing blocks are written in C++, and connected to each other with Python. This makes it easy to develop more sophisticated signal processing systems, because many blocks already written by others and you can quickly put them together to create a complete system. Although the main function of GNU Radio is not be a simulator, but if there is no RF hardware components,it supports to researching the signal processing algorithm based on pre-stored and generated data by signal generator. This thesis introduced SDR platform from hardware (USRP) and software(GNU Radio), as well as some basic modulation techniques in wireless communication system. Based on the examples provided by GNU Radio, carried out some related experiments, for example GSM scanning and FM radio station receiving on USRP. And make a certain degree of improvement based on the experience of some investigators to observe OFDM spectrum and simulate real-time video transmission. GNU Radio combine with USRP hardware proved to be a valuable lab platform for implementing complex radio system prototypes in a short time. RESUMEN. Software Defined Radio (SDR) es una tecnología emergente que está creando un impacto revolucionario en la tecnología de radio convencional. Un buen ejemplo de radio software son los sistemas de código abierto llamados GNU Radio que emplean un kit de herramientas de desarrollo de software libre. En este trabajo se ha empleado un kit de desarrollo comercial (Ettus Research) que consiste en un módulo de procesado de señal y un hardaware sencillo. El módulo emplea un software de desarrollo basado en Linux sobre el que se pueden implementar aplicaciones de radio software muy variadas. El hardware de desarrollo consta de un microprocesador de propósito general, un dispositivo programable (FPGA) y un interfaz de radiofrecuencia que cubre de 50 a 2200MHz. Este hardware se conecta al PC por medio de un interfaz USB de 8Mb/s de velocidad. Sobre la plataforma de Ettus se pueden ejecutar aplicaciones GNU radio que utilizan principalmente lenguaje de programación Python para implementarse. Sin embargo, su módulo de procesado de señal está construido en C + + y emplea un microprocesador con aritmética de coma flotante. Por lo tanto, los desarrolladores pueden rápida y fácilmente construir aplicaciones en tiempo real sistemas de comunicación inalámbrica de alta capacidad. Aunque su función principal no es ser un simulador, si no puesto que hay componentes de hardware RF, Radio GNU sirve de apoyo a la investigación del algoritmo de procesado de señales basado en pre-almacenados y generados por los datos del generador de señal. En este trabajo fin de máster se ha evaluado la plataforma de hardware de DEG (USRP) y el software (GNU Radio). Para ello se han empleado algunas técnicas de modulación básicas en el sistema de comunicación inalámbrica. A partir de los ejemplos proporcionados por GNU Radio, hemos realizado algunos experimentos relacionados, por ejemplo, escaneado del espectro, demodulación de señales de FM empleando siempre el hardware de USRP. Una vez evaluadas aplicaciones sencillas se ha pasado a realizar un cierto grado de mejora y optimización de aplicaciones complejas descritas en la literatura. Se han empleado aplicaciones como la que consiste en la generación de un espectro de OFDM y la simulación y transmisión de señales de vídeo en tiempo real. Con estos resultados se está ahora en disposición de abordar la elaboración de aplicaciones complejas.

Diseño de una interfaz de representación gráfica de la base de datos de HRTF

El actual proyecto consiste en la creación de una interfaz gráfica de usuario (GUI) en entorno de MATLAB que realice una representación gráfica de la base de datos de HRTF (Head-Related Transfer Function). La función de transferencia de la cabeza es una herramienta muy útil en el estudio de la capacidad del ser humano para percibir su entorno sonoro, además de la habilidad de éste en la localización de fuentes sonoras en el espacio que le rodea. La HRTF biaural (terminología para referirse al conjunto de HRTF del oído izquierdo y del oído derecho) en sí misma, posee información de especial interés ya que las diferencias entre las HRTF de cada oído, conceden la información que nuestro sistema de audición utiliza en la percepción del campo sonoro. Por ello, la funcionalidad de la interfaz gráfica creada presenta gran provecho dentro del estudio de este campo. Las diferencias interaurales se caracterizan en amplitud y en tiempo, variando en función de la frecuencia. Mediante la transformada inversa de Fourier de la señal HRTF, se obtiene la repuesta al impulso de la cabeza, es decir, la HRIR (Head-Related Impulse Response). La cual, además de tener una gran utilidad en la creación de software o dispositivos de generación de sonido envolvente, se utiliza para obtener las diferencias ITD (Interaural Time Difference) e ILD (Interaural Time Difference), comúnmente denominados “parámetros de localización espacial”. La base de datos de HRTF contiene la información biaural de diferentes puntos de ubicación de la fuente sonora, formando una red de coordenadas esféricas que envuelve la cabeza del sujeto. Dicha red, según las medidas realizadas en la cámara anecoica de la EUITT (Escuela Universitaria de Ingeniería Técnica de Telecomunicación), presenta una precisión en elevación de 10º y en azimut de 5º. Los receptores son dos micrófonos alojados en el maniquí acústico llamado HATS (Hats and Torso Simulator) modelo 4100D de Brüel&Kjaer. Éste posee las características físicas que influyen en la percepción del entorno como son las formas del pabellón auditivo (pinna), de la cabeza, del cuello y del torso humano. Será necesario realizar los cálculos de interpolación para todos aquellos puntos no contenidos en la base de datos HRTF, este proceso es sumamente importante no solo para potenciar la capacidad de la misma sino por su utilidad para la comparación entre otras bases de datos existentes en el estudio de este ámbito. La interfaz gráfica de usuario está concebida para un manejo sencillo, claro y predecible, a la vez que interactivo. Desde el primer boceto del programa se ha tenido clara su filosofía, impuesta por las necesidades de un usuario que busca una herramienta práctica y de manejo intuitivo. Su diseño de una sola ventana reúne tanto los componentes de obtención de datos como los que hacen posible la representación gráfica de las HRTF, las HRIR y los parámetros de localización espacial, ITD e ILD. El usuario podrá ir alternando las representaciones gráficas a la vez que introduce las coordenadas de los puntos que desea visualizar, definidas por phi (elevación) y theta (azimut). Esta faceta de la interfaz es la que le otorga una gran facilidad de acceso y lectura de la información representada en ella. Además, el usuario puede introducir valores incluidos en la base de datos o valores intermedios a estos, de esta manera, se indica a la interfaz la necesidad de realizar la interpolación de los mismos. El método de interpolación escogido es el de la ponderación de la distancia inversa entre puntos. Dependiendo de los valores introducidos por el usuario se realizará una interpolación de dos o cuatro puntos, siendo éstos limítrofes al valor introducido, ya sea de phi o theta. Para añadir versatilidad a la interfaz gráfica de usuario, se ha añadido la opción de generar archivos de salida en forma de imagen de las gráficas representadas, de tal forma que el usuario pueda extraer los datos que le interese para cualquier valor de phi y theta. Se completa el presente proyecto fin de carrera con un trabajo de investigación y estudio comparativo de la función y la aplicación de las bases de datos de HRTF dentro del marco científico y de investigación. Esto ha hecho posible concentrar información relacionada a través de revistas científicas de investigación como la JAES (Journal of the Audio Engineering Society) o la ASA (Acoustical Society of America), además, del IEEE ( Institute of Electrical and Electronics Engineers) o la “Web of knowledge” entre otras. Además de realizar la búsqueda en estas fuentes, se ha optado por vías de información más comunes como Google Académico o el portal de acceso “Ingenio” a los todos los recursos electrónicos contenidos en la base de datos de la universidad. El estudio genera una ampliación en el conocimiento de la labor práctica de las HRTF. La mayoría de los estudios enfocan sus esfuerzos en mejorar la percepción del evento sonoro mediante su simulación en la escucha estéreo o multicanal. A partir de las HRTF, esto es posible mediante el análisis y el cálculo de datos como pueden ser las regresiones, siendo éstas muy útiles en la predicción de una medida basándose en la información de la actual. Otro campo de especial interés es el de la generación de sonido 3D. Mediante la base de datos HRTF es posible la simulación de una señal biaural. Se han diseñado algoritmos que son implementados en dispositivos DSP, de tal manera que por medio de retardos interaurales y de diferencias espectrales es posible llegar a un resultado óptimo de sonido envolvente, sin olvidar la importancia de los efectos de reverberación para conseguir un efecto creíble de sonido envolvente. Debido a la complejidad computacional que esto requiere, gran parte de los estudios coinciden en desarrollar sistemas más eficientes, llegando a objetivos tales como la generación de sonido 3D en tiempo real. ABSTRACT. This project involves the creation of a Graphic User Interface (GUI) in the Matlab environment which creates a graphic representation of the HRTF (Head-Related Transfer Function) database. The head transfer function is a very useful tool in the study of the capacity of human beings to perceive their sound environment, as well as their ability to localise sound sources in the area surrounding them. The binaural HRTF (terminology which refers to the HRTF group of the left and right ear) in itself possesses information of special interest seeing that the differences between the HRTF of each ear admits the information that our system of hearing uses in the perception of each sound field. For this reason, the functionality of the graphic interface created presents great benefits within the study of this field. The interaural differences are characterised in space and in time, varying depending on the frequency. By means of Fourier's transformed inverse of the HRTF signal, the response to the head impulse is obtained, in other words, the HRIR (Head-Related Impulse Response). This, as well as having a great use in the creation of software or surround sound generating devices, is used to obtain ITD differences (Interaural Time Difference) and ILD (Interaural Time Difference), commonly named “spatial localisation parameters”. The HRTF database contains the binaural information of different points of sound source location, forming a network of spherical coordinates which surround the subject's head. This network, according to the measures carried out in the anechoic chamber at the EUITT (School of Telecommunications Engineering) gives a precision in elevation of 10º and in azimuth of 5º. The receivers are two microphones placed on the acoustic mannequin called HATS (Hats and Torso Simulator) Brüel&Kjaer model 4100D. This has the physical characteristics which affect the perception of the surroundings which are the forms of the auricle (pinna), the head, neck and human torso. It will be necessary to make interpolation calculations for all those points which are not contained the HRTF database. This process is extremely important not only to strengthen the database's capacity but also for its usefulness in making comparisons with other databases that exist in the study of this field. The graphic user interface is conceived for a simple, clear and predictable use which is also interactive. Since the first outline of the program, its philosophy has been clear, based on the needs of a user who requires a practical tool with an intuitive use. Its design with only one window unites not only the components which obtain data but also those which make the graphic representation of the HRTFs possible, the hrir and the ITD and ILD spatial location parameters. The user will be able to alternate the graphic representations at the same time as entering the point coordinates that they wish to display, defined by phi (elevation) and theta (azimuth). The facet of the interface is what provides the great ease of access and reading of the information displayed on it. In addition, the user can enter values included in the database or values which are intermediate to these. It is, likewise, indicated to the interface the need to carry out the interpolation of these values. The interpolation method is the deliberation of the inverse distance between points. Depending on the values entered by the user, an interpolation of two or four points will be carried out, with these being adjacent to the entered value, whether that is phi or theta. To add versatility to the graphic user interface, the option of generating output files in the form of an image of the graphics displayed has been added. This is so that the user may extract the information that interests them for any phi and theta value. This final project is completed with a research and comparative study essay on the function and application of HRTF databases within the scientific and research framework. It has been possible to collate related information by means of scientific research magazines such as the JAES (Journal of the Audio Engineering Society), the ASA (Acoustical Society of America) as well as the IEEE (Institute of Electrical and Electronics Engineers) and the “Web of knowledge” amongst others. In addition to carrying out research with these sources, I also opted to use more common sources of information such as Academic Google and the “Ingenio” point of entry to all the electronic resources contained on the university databases. The study generates an expansion in the knowledge of the practical work of the HRTF. The majority of studies focus their efforts on improving the perception of the sound event by means of its simulation in stereo or multichannel listening. With the HRTFs, this is possible by means of analysis and calculation of data as can be the regressions. These are very useful in the prediction of a measure being based on the current information. Another field of special interest is that of the generation of 3D sound. Through HRTF databases it is possible to simulate the binaural signal. Algorithms have been designed which are implemented in DSP devices, in such a way that by means of interaural delays and wavelength differences it is possible to achieve an excellent result of surround sound, without forgetting the importance of the effects of reverberation to achieve a believable effect of surround sound. Due to the computational complexity that this requires, a great many studies agree on the development of more efficient systems which achieve objectives such as the generation of 3D sound in real time.

Higher-order regularization and morphological techniques for image segmentation

La segmentación de imágenes es un campo importante de la visión computacional y una de las áreas de investigación más activas, con aplicaciones en comprensión de imágenes, detección de objetos, reconocimiento facial, vigilancia de vídeo o procesamiento de imagen médica. La segmentación de imágenes es un problema difícil en general, pero especialmente en entornos científicos y biomédicos, donde las técnicas de adquisición imagen proporcionan imágenes ruidosas. Además, en muchos de estos casos se necesita una precisión casi perfecta. En esta tesis, revisamos y comparamos primero algunas de las técnicas ampliamente usadas para la segmentación de imágenes médicas. Estas técnicas usan clasificadores a nivel de pixel e introducen regularización sobre pares de píxeles que es normalmente insuficiente. Estudiamos las dificultades que presentan para capturar la información de alto nivel sobre los objetos a segmentar. Esta deficiencia da lugar a detecciones erróneas, bordes irregulares, configuraciones con topología errónea y formas inválidas. Para solucionar estos problemas, proponemos un nuevo método de regularización de alto nivel que aprende información topológica y de forma a partir de los datos de entrenamiento de una forma no paramétrica usando potenciales de orden superior. Los potenciales de orden superior se están popularizando en visión por computador, pero la representación exacta de un potencial de orden superior definido sobre muchas variables es computacionalmente inviable. Usamos una representación compacta de los potenciales basada en un conjunto finito de patrones aprendidos de los datos de entrenamiento que, a su vez, depende de las observaciones. Gracias a esta representación, los potenciales de orden superior pueden ser convertidos a potenciales de orden 2 con algunas variables auxiliares añadidas. Experimentos con imágenes reales y sintéticas confirman que nuestro modelo soluciona los errores de aproximaciones más débiles. Incluso con una regularización de alto nivel, una precisión exacta es inalcanzable, y se requeire de edición manual de los resultados de la segmentación automática. La edición manual es tediosa y pesada, y cualquier herramienta de ayuda es muy apreciada. Estas herramientas necesitan ser precisas, pero también lo suficientemente rápidas para ser usadas de forma interactiva. Los contornos activos son una buena solución: son buenos para detecciones precisas de fronteras y, en lugar de buscar una solución global, proporcionan un ajuste fino a resultados que ya existían previamente. Sin embargo, requieren una representación implícita que les permita trabajar con cambios topológicos del contorno, y esto da lugar a ecuaciones en derivadas parciales (EDP) que son costosas de resolver computacionalmente y pueden presentar problemas de estabilidad numérica. Presentamos una aproximación morfológica a la evolución de contornos basada en un nuevo operador morfológico de curvatura que es válido para superficies de cualquier dimensión. Aproximamos la solución numérica de la EDP de la evolución de contorno mediante la aplicación sucesiva de un conjunto de operadores morfológicos aplicados sobre una función de conjuntos de nivel. Estos operadores son muy rápidos, no sufren de problemas de estabilidad numérica y no degradan la función de los conjuntos de nivel, de modo que no hay necesidad de reinicializarlo. Además, su implementación es mucho más sencilla que la de las EDP, ya que no requieren usar sofisticados algoritmos numéricos. Desde un punto de vista teórico, profundizamos en las conexiones entre operadores morfológicos y diferenciales, e introducimos nuevos resultados en este área. Validamos nuestra aproximación proporcionando una implementación morfológica de los contornos geodésicos activos, los contornos activos sin bordes, y los turbopíxeles. En los experimentos realizados, las implementaciones morfológicas convergen a soluciones equivalentes a aquéllas logradas mediante soluciones numéricas tradicionales, pero con ganancias significativas en simplicidad, velocidad y estabilidad. ABSTRACT Image segmentation is an important field in computer vision and one of its most active research areas, with applications in image understanding, object detection, face recognition, video surveillance or medical image processing. Image segmentation is a challenging problem in general, but especially in the biological and medical image fields, where the imaging techniques usually produce cluttered and noisy images and near-perfect accuracy is required in many cases. In this thesis we first review and compare some standard techniques widely used for medical image segmentation. These techniques use pixel-wise classifiers and introduce weak pairwise regularization which is insufficient in many cases. We study their difficulties to capture high-level structural information about the objects to segment. This deficiency leads to many erroneous detections, ragged boundaries, incorrect topological configurations and wrong shapes. To deal with these problems, we propose a new regularization method that learns shape and topological information from training data in a nonparametric way using high-order potentials. High-order potentials are becoming increasingly popular in computer vision. However, the exact representation of a general higher order potential defined over many variables is computationally infeasible. We use a compact representation of the potentials based on a finite set of patterns learned fromtraining data that, in turn, depends on the observations. Thanks to this representation, high-order potentials can be converted into pairwise potentials with some added auxiliary variables and minimized with tree-reweighted message passing (TRW) and belief propagation (BP) techniques. Both synthetic and real experiments confirm that our model fixes the errors of weaker approaches. Even with high-level regularization, perfect accuracy is still unattainable, and human editing of the segmentation results is necessary. The manual edition is tedious and cumbersome, and tools that assist the user are greatly appreciated. These tools need to be precise, but also fast enough to be used in real-time. Active contours are a good solution: they are good for precise boundary detection and, instead of finding a global solution, they provide a fine tuning to previously existing results. However, they require an implicit representation to deal with topological changes of the contour, and this leads to PDEs that are computationally costly to solve and may present numerical stability issues. We present a morphological approach to contour evolution based on a new curvature morphological operator valid for surfaces of any dimension. We approximate the numerical solution of the contour evolution PDE by the successive application of a set of morphological operators defined on a binary level-set. These operators are very fast, do not suffer numerical stability issues, and do not degrade the level set function, so there is no need to reinitialize it. Moreover, their implementation is much easier than their PDE counterpart, since they do not require the use of sophisticated numerical algorithms. From a theoretical point of view, we delve into the connections between differential andmorphological operators, and introduce novel results in this area. We validate the approach providing amorphological implementation of the geodesic active contours, the active contours without borders, and turbopixels. In the experiments conducted, the morphological implementations converge to solutions equivalent to those achieved by traditional numerical solutions, but with significant gains in simplicity, speed, and stability.