This thesis involves restoration, reconstruction, and enhancement of a digitized library of hand-written documents. Imaging systems that perform this digitization often degrade the quality of the original documents. Many techniques exist for reconstructing, restoring, and enhancing digital images; however, many require <i> a priori </i> knowledge of the imaging system. In this study, only partial <i> a priori </i> knowledge is available, and therefore unknown parameters must be estimated before restoration, reconstruction, or enhancement is possible. The imaging system used to digitize the documents library has degraded the images in several ways. First, it has introduced a ringing that is apparent around each stroke. Second, the system has eliminated strokes of narrow widths. To restore these images, the imaging system is modeled by estimating the point spread function from sample impulse responses, and the image noise is estimated in an attempt to apply standard linear restoration techniques. The applicability of these techniques is investigated in the first part of this thesis. Then nonlinear filters, structural techniques, and enhancement techniques are applied to obtain substantial improvements in image quality. / Master of Science

Virtual acoustic techniques can be used to create virtual listening environments for multiple purposes. Using multi-speaker reproduction, a physical environment can take on the acoustical appearance of another environment. Implementation of this environment auralization could change the way customers evaluate speakers in a retail store. The objective of this research is to develop a virtual acoustic showroom using a multi- speaker system. The two main components to the virtual acoustic showroom are simulating living environments using the image source method, and simulating speaker responses using inverse filtering. The image source method is used to simulate realistic living environments by filtering the environment impulse response by frequency-dependant absorption coefficients of typical building materials. Psychoacoustic tests show that listeners can match virtual acoustic cues with appropriate virtual visual cues. Inverse filtering is used to "replace" the frequency response function of one speaker with another, allowing a single set of speakers to represent any number of other speakers. Psychoacoustic tests show that listeners could not distinguish the difference between the original speaker and the reference speaker that was mimicking the original. The two components of this system are shown to be accurate both empirically and psychologically. / Master of Science

In this thesis, we considered the deconvolution problem when the channel is known a priori. LSE and MSE optimum solutions are investigated with deterministic and statistical approaches. We derived closed form LSE expressions and investigated the factors that affect the FIR inverse filters. It turns out that, minimum LSE can be obtained when the system zeros are distributed homogeneously on the z-plane. We proposed partition-based FIR-IIR inverse filters. The selection of FIR and IIR parts is based on partitioning the channel zeros into two regions and using the specified channel zeros to design the best delay FIR and all pole IIR inverse filters. Three methods for partitioning are presented, namely unit circle-based, ring-based and optimum-partitioning. It turns out that ring-based and optimum-partitioning FIR-IIR inverse filter performs better than the best delay FIR inverse filter for the same complexity by about 4-5 dB. For noisy observations, it is shown that, noise should also be considered in the delay selection and partitioning. We extended our results for the design of MSE optimum statistical inverse filters. It is shown that best delay FIR-IIR inverse filters are less sensitive to the estimation errors compared to the IIR Wiener filters and they perform better than the FIR Wiener filters. Furthermore, they are always causal and stable making them suitable for real-time implementations. When the statistical and deterministic filters are compared, it is shown that for low SNR statistical filters perform better by about 1-2 dB, while deterministic filters perform better by about 0.5-1 dB for high SNR

[pt] A grande demanda por visualização através de imagens aliada ao caracter não invasivo das medidas de campo magnético motiva o desenvolvimento de ferramentas para o processamento de sinais em biomagnetismo. A técnica de filtragem inversa no domínio da freqüência, utilizada para restauração de imagens óticas, foi aplicada para se obter, a partir do campo magnético medido, uma imagem de sua fonte. Diversas simulações com distribuições de momentos magnéticos foram realizadas para se avaliar a qualidade da recuperação em presença de ruído. A técnica foi também aplicada em medidas experimentais obtidas com um mangnetômetro supercondutor. / [en] The great demand for images and the non-invasive character of the magnetic field measurements encourage the development of tools for signal processing in biomagnetism. The inverse filtering in the frequency domain technique often used in optical image restoration was applied to obtain an image of the source of magnetic field from its measured magnetic signal. Many simulations with two-dimensional magnetic moment distribution were carried out to determine the quality of the recovery in the presence of noise. The algorithm was also used with experimental data from measurement made with a super conducting magnetometer.

Using a near-field array of acoustic resonators, it is possible to modify a focused pressure field and enforce a spatial frequency corresponding to the resonator array spacing. This higher spatial frequency makes it possible to focus and image with a resolution that is better than if the focusing were in free space. This near-field effect is caused by the phase shifting properties of resonators and, specifically, the delayed phase found in waves with a temporal frequency lower than that of the resonators in the array. Using time reversal, arrays of resonators are explored and the subwavelength focusing is used to describe the ability to image subwavelength features. A one-dimensional equivalent circuit model accurately predicts this interaction of the wave field with an array of resonators and is able to model the aggregate effect of the phononic crystal of resonators while describing the fine spatial details of individual resonators. This model is validated by a series of COMSOL full-wave simulations of the same system. The phase delay caused by a single resonator is explored in a simple experiment as well as in the equivalent circuit model. A series of experiments is conducted with a two-dimensional array of resonators and complex images are produced which indicate the ability to focus complex sources with better resolution.

Neste trabalho, foram utilizadas as vantagens da ferramenta matemática de análise temporal e espectral, a transformada wavelet discreta (DWT), além dos coeficientes de predição linear (LPC) e do algoritmo de inteligência artificial, Least Squares Support Vector Machines (LS-SVM), para aplicações em análise de sinais de voz e classificação de vozes patológicas. Inúmeros trabalhos na literatura têm demonstrado o grande interesse existente por ferramentas auxiliares ao diagnóstico de patologias da laringe. Os componentes da DWT forneceram parâmetros de medida para a análise e classificação das vozes patológicas, principalmente aquelas provenientes de pacientes com edema de Reinke e nódulo nas pregas vocais. O banco de dados com as vozes patológicas foi obtido do Departamento de Otorrinolaringologia e Cirurgia de Cabeça e Pescoço do Hospital das Clínicas da Faculdade de Medicina de Ribeirão Preto (FMRP-USP). Utilizando-se o algoritmo de reconhecimento de padrões, LS-SVM, mostrou-se que a combinação dos componentes da DWT de Daubechies com o filtro LP inverso levou a um classificador de bom desempenho alcançando mais de 90% de acerto na classificação das vozes patológicas. / The main objective of this work was to use the advantages of the time-frequency analysis mathematical tool, discrete wavelet transform (DWT), besides the linear prediction coefficients (LPC) and the artificial intelligence algorithm, Least Squares Support Vector Machines (LS-SVM), for applications in voice signal analysis and classification of pathological voices. A large number of works in the literature has been shown that there is a great interest for auxiliary tools to the diagnosis of laryngeal pathologies. DWT components gave measure parameters for the analysis and classification of pathological voices, mainly that ones from patients with Reinke\'s edema and nodule in the vocal folds. It was used a data bank with pathological voices from the Otolaryngology and the Head and Neck Surgery sector of the Clinical Hospital of the Faculty of Medicine at Ribeirão Preto, University of Sao Paulo (FMRP-USP), Brazil. Using the automatic learning algorithm applied in pattern recognition problems, LS-SVM, results have showed that the combination of Daubechies\' DWT components and inverse LP filter leads to a classifier with good performance reaching more than 90% of accuracy in the classification of the pathological voices.

Lorsque des ondes acoustiques de surface rayonnent dans des fluides, elles provoquent deux effets non linéaires : la pression de radiation et le streaming acoustique. Ces deux effets ont trouvé un grand nombre d’applications pour la microfluidique digitale, la manipulation sans contact et le tri cellulaire. Néanmoins, ces systèmes se heurtent à deux limites. D’une part, chaque application requiert une onde acoustique spécifique : il n’existe pas de dispositif multifonction à ce jour. D’autre part, l’exploration des fonctionnalités offertes par les ondes de surface les plus simples (ondes planes, ondes focalisées) n’a pas permis de réaliser des pinces sélectives permettant de manipuler individuellement des particules ou cellules indépendamment de leurs voisines.Dans une première partie de la thèse, nous développons deux méthodologies pour synthétiser des champs complexes d’ondes de surface. La première méthode utilise un réseau de 32 peignes interdigités contrôlé par la technique du filtre inverse pour générer des champs sur demande. La seconde résout un problème inverse afin de concevoir un transducteur holographique générant spécifiquement le champ demandé. Dans la seconde partie de la thèse, nous utilisons le filtre inverse pour (i) réaliser un laboratoire sur puce multifonction et (ii) étudier le potentiel d’ondes de surface particulières appelées ondes de surface tourbillonnaires. Ces ondes permettent une manipulation sélective et sans contact d’objets microscopiques. Nous terminons la thèse en équipant un microscope d’un transducteur holographique de vortex acoustiques afin de réaliser une manipulation sélective et sans contact de cellules. / When surface acoustic waves radiate in nearby fluids, they trigger two nonlinear effects: acoustic radiation pressure and acoustic streaming. These two effects find numerous applications for digital microfluidics, contactless manipulation and biological cell sorting. Nonetheless, these systems face two limitations. On the one hand, each application requires a specific acoustic wave: there is no multifunction device so far. On the other hand, search for functionalities offered by simple surface acoustic waves (plane and focused waves) has failed to provide a selective tweezers able to manipulate individual particles or cells independently of their neighbors. In the first part of this thesis, we develop two methods to synthesize complex surface acoustic wave fields. The first one employs an array of 32 interdigitated transducers controlled by the inverse filter to generate arbitrary fields on demand. The second method solves an inverse problem to design a holographic transducer to generate a predefined field. In the second part of the thesis, we use the inverse filter to (i) implement a multifunction lab on a chip and (ii) investigate the potentialities of a special type of surface acoustic waves called swirling surface waves. These waves enable a selective and contactless manipulation of microscopic objects. We conclude the thesis by integrating a holographic acoustical vortex transducer on a microscope in order to selectively manipulate biological cells without contact.

In radars, matched filters are used in the receiver of the system. Since the target velocity is not known a priori, degradation occurs due to mismatch of the return signal and the matched filter. The performance of the radar can be improved by using a bank of matched filters. The first topic investigated in this work is optimization of the bank of matched filter structure. Two methods are proposed for the design of the parallel filter structure and computations are performed with both methods. The output signal of a radar receiver filter consists not only of the main peak from the target but also of range sidelobes. In a multi-target radar environment, the sidelobes of one large target may appear as a smaller target at another range, or the integrated sidelobes from targets or clutter may mask all the information of another target. The second part of this thesis discusses the methods for decreasing the sidelobe level of the receiver output. Two methods are studied for this purpose. The first is the classical amplitude weighting and the second is the use of an inverse filter that minimizes total sidelobe energy. Both methods decrease the sidelobe levels while bringing a mismatch loss and main peak broadening. For the inverse filter case it is observed that the effect of inverse filter becomes evident as the filter length is increased beyond some point. Finally, the effects of quantization on video signal and the receiver filter coefficients are evaluated. It is observed that 16 bits quantization is sufficient for all kinds of receiver filters tested.

Ce travail de thèse se décompose en deux parties. Tout d'abord, nous présentons une nouvelle technique de focalisation d'ondes avec un seul émetteur en cavité réverbérante (FIM) en profitant d'un algorithme inspiré du filtre inverse. Via l'étude expérimentale de cavités réverbérantes dans le domaine des ultrasons, nous démontrons la capacité du FIM à optimiser la focalisation quelle que soit le type de cavité (de type ergodique ou non). Dans une deuxième partie, la propagation d'ondes élastiques dans un système formé par un ensemble de tiges d'aluminium collées sur une plaque mince de même nature, est étudiée. Ces tiges (résonateurs quasi-ponctuels) sont arrangées de façon périodique ou aléatoire sur une échelle sub-longueur d'onde. Le métamatériau ainsi constitué révèle la présence de larges bandes de fréquences interdites. De plus, la coexistence de résonances de flexion et de compression dans les résonateurs, ajoutée à la présence d'une conversion d'une partie de l'énergie du mode A0 vers le mode S0 dans la plaque, crée une grande complexité du champ d'onde. C'est ce qui fait de ce type de métamatériau, des objets tout à fait singuliers à l'échelle mésoscopique. / This thesis is divided into two parts. First, we present a new technique for focusing waves with one emitter in reverberant cavity (OCIF) inspired by inverse filter algorithm. Through the experimental study of reverberant cavities in the field of ultrasound, we demonstrate the ability of the OCIF to optimize the focusing no matter what type of cavity (ergodic type or not). In a second part, we investigate the propagation of elastic waves in a system formed by a set of aluminum rods glued to a thin plate of the same material. These rods form a set of quasi-punctual resonators in the propagation plane of waves. It is possible to arrange them periodically or randomly on a subwavelength scale. The metamaterial thus formed shows a complex wave field within it, including the presence of wide prohibited frequency ranges (bandgaps). The experimental and numerical approaches described in this manuscript show the existence of both flexural and compressional resonances in the resonators. Added to the presence of a conversion of a portion of the energy from the $A0$ Lamb mode to the $S0$ one in the plate, such a complexity makes this type of metamaterials, quite unusual objects at the mesoscopic scale.

Neste trabalho, foram utilizadas as vantagens da ferramenta matemática de análise temporal e espectral, a transformada wavelet discreta (DWT), além dos coeficientes de predição linear (LPC) e do algoritmo de inteligência artificial, Least Squares Support Vector Machines (LS-SVM), para aplicações em análise de sinais de voz e classificação de vozes patológicas. Inúmeros trabalhos na literatura têm demonstrado o grande interesse existente por ferramentas auxiliares ao diagnóstico de patologias da laringe. Os componentes da DWT forneceram parâmetros de medida para a análise e classificação das vozes patológicas, principalmente aquelas provenientes de pacientes com edema de Reinke e nódulo nas pregas vocais. O banco de dados com as vozes patológicas foi obtido do Departamento de Otorrinolaringologia e Cirurgia de Cabeça e Pescoço do Hospital das Clínicas da Faculdade de Medicina de Ribeirão Preto (FMRP-USP). Utilizando-se o algoritmo de reconhecimento de padrões, LS-SVM, mostrou-se que a combinação dos componentes da DWT de Daubechies com o filtro LP inverso levou a um classificador de bom desempenho alcançando mais de 90% de acerto na classificação das vozes patológicas. / The main objective of this work was to use the advantages of the time-frequency analysis mathematical tool, discrete wavelet transform (DWT), besides the linear prediction coefficients (LPC) and the artificial intelligence algorithm, Least Squares Support Vector Machines (LS-SVM), for applications in voice signal analysis and classification of pathological voices. A large number of works in the literature has been shown that there is a great interest for auxiliary tools to the diagnosis of laryngeal pathologies. DWT components gave measure parameters for the analysis and classification of pathological voices, mainly that ones from patients with Reinke\'s edema and nodule in the vocal folds. It was used a data bank with pathological voices from the Otolaryngology and the Head and Neck Surgery sector of the Clinical Hospital of the Faculty of Medicine at Ribeirão Preto, University of Sao Paulo (FMRP-USP), Brazil. Using the automatic learning algorithm applied in pattern recognition problems, LS-SVM, results have showed that the combination of Daubechies\' DWT components and inverse LP filter leads to a classifier with good performance reaching more than 90% of accuracy in the classification of the pathological voices.