The Singular Value Decomposition (SVD) has many applications in image processing. The SVD can be used to restore a corrupted image by separating significant information from the noise in the image data set. This thesis outlines broad applications that address current problems in digital image processing. In conjunction with SVD filtering, image compression using the SVD is discussed, including the process of reconstructing or estimating a rank reduced matrix representing the compressed image. Numerical plots and error measurement calculations are used to compare results of the two SVD image restoration techniques, as well as SVD image compression. The filtering methods assume that the images have been degraded by the application of a blurring function and the addition of noise. Finally, we present numerical experiments for the SVD restoration and compression to evaluate our computation.

The Kalman Filter has many applications in control and signal processing but may also be used to reconstruct a higher resolution image from a sequence of lower resolution images (or frames). If the sequence of low resolution frames is recorded by a moving camera or sensor, where the motion can be accurately modeled, then the Kalman filter may be used to update pixels within a higher resolution frame to achieve a more detailed result. This thesis outlines current methods of implementing this algorithm on a scene of interest and introduces possible improvements for the speed and efficiency of this method by use of block operations on the low resolution frames. The effects of noise on camera motion and various blur models are examined using experimental data to illustrate the differences between the methods discussed.

In the past two decades, Fluorescence microscopy has imparted tremendous impact in Biology and Imaging. Several super-resolution Fluorescence imaging techniques (e.g. PALM, STED, STORM, 4Pi and structured illumination) have enabled diff raction-unlimited imaging. But high resolution is limited to a depth of few tens of microns. Thus, deep tissue imaging and simultaneous volume imaging have become a highly sought after feature in Fluorescence microscopy. The research work in this thesis address these issues by using spatial filtering techniques to tailor the point spread function (PSF) which uniquely characterizes the optical sys-tem. The advantage of this approach lies in the fact that intricate details about the focal region can be computed and designed with the help of well established theory and experimentation. In particular, this technique was applied to both spherical and cylindrical lenses. The former was used to generate Bessel-like, non-diffracting beams which demonstrated the ability to penetrate deep inside tissue-like media and thereby yielded an imaging depth of nearly 650μm as compared to about 200μm for a state-of-the-art confocal microscope. The latter gave rise to light-sheet and it's extended version that is ideal for planar imaging at large penetration depths. Another development is the generation of multiple light-sheet illumination pattern that can simultaneously illuminate several planes of the specimen. The proposed multiple light-sheet illumination microscopy (MLSIM) technique may enable volume imaging in Fluorescence microscopy. The first two chapters of this thesis are introductory in nature and provides a general overview of the principles of Fluorescence microscopy and three state-of-the-art Fluorescence imaging techniques; namely confocal, multi-photon and light-sheet based microscopy. Confocal microscopes are widely considered as a standard tool for biologists and this discussion shows that even though they have made signi ficant contributions in the fields of biophysics, biophotonics and nanoscale imaging, their inability to achieve better penetration depth has prevented their use in thick, scattering samples such as biological tissue. The system PSF of a confocal microscope broadens as it goes deeper in-side a scattering sample resulting in poor-resolution thereby destroying the very concept of high resolution, noise-free imaging. Additionally, confocal microscopy suffers from in-creased photo-bleaching due to o -layer (above and below the focal plane) excitation and low temporal resolution since it requires point-by-point scanning mechanism. On the other hand, multi-photon microscopy offers several advantages over confocal microscopy such as reduced photo-bleaching and inherent optical sectioning ability, however, it still lacks in providing high temporal resolution. Light-sheet based microscopy have gained popularity in recent years and promises to deliver high spatio-temporal resolution with minimized photo-bleaching. Recently, a considerable amount of research has been dedicated to further develop this promising technique for a variety of applications. The ability to look deeper inside a biological specimen has profound implications. How-ever, at depths of hundreds of microns, several effects (such as scattering, PSF distortion and noise) deteriorates the image quality and prohibits detailed study of key biological phenomenon. Chapter 3 of this thesis describes the original research work which experimentally addresses to this issue. Here, Bessel-like beam is employed in conjugation with an orthogonal detection scheme to achieve imaging at large penetration depth. Bessel beams are penetrative, non-di ffracting and have self-reconstruction properties making them a natural choice for imaging scattering prone specimens which are otherwise inaccessible by other microscopy imaging techniques such as, Widefield, CLSM, 4PI, Structural illumination microscopy and others. In this case such a Bessel-like beam is generated by masking the back-aperture of the excitation objective with a ring-like spatial filter. The proposed excitation scheme allow continuous scanning by simply translating the detection optics. Additionally, only a pencil-like region of the specimen can be illuminated at a given instance thereby reducing premature photobleaching of neighboring regions. This illumination scheme coupled with orthogonal detection shows the ability of selective imaging from a desired plane deep inside the specimen. In such a configuration, the lateral resolution of the illumination arm determines the axial resolution of the overall imaging system. Such an imaging system is a boon for obtaining depth information from any desired specimen layer that includes nano-particle tracking in thick tissue. Experiments performed by imaging the Fluorescent polymer tagged-CaCO3 particles and yeast cell in a tissue-like gel-matrix demonstrates penetration depth that extends up to 650 m. This will advance the field of fluorescence imaging microscopy and imaging. Similar to the ability to observe deep inside a sample, simultaneous 3D monitoring of whole specimens play a vital role in understanding many developmental process in Biology. At present, light-sheet based microscopy is the prime candidate amongst the various microscopy techniques, that is capable of providing high signal-to-background-ratio as far as planar imaging is concerned. Since spatial filtering technique was found to successfully give rise to novel features (such as large penetration depth) in a fluorescence microscope setup, a logical extension would be to implement a similar approach with a light-sheet based microscope setup. These implementations are discussed in Chapter 4 of this thesis where spatial filtering is employed with cylindrical lenses. For facilitating computational and experimental studies, a vectorial formalism was derived to give an explicit computable integral solution of the electric field generated at the focal region of a cylindrical lens. This representation is based on vectorial diffraction theory and further enables the computation of the point spread function of a cylindrical lens. Commonly used assumptions are made in the derivation such as no back-scattering and negligible contribution from evanescent fields. Stationary phase approximation along with the Fresnel transmission coefficients are employed for evaluating the polarization dependent electric field components. Computational studies were carried out to determine the polarization effects and calculate the system resolution. Experimental comparison of light-sheet intensity pro les show good agreement with the theoretical calculations and hence validate the model. This formalism was derived as a first step since it gives the essential understanding of tightly focused E-fields of a high N.A. cylindrical lens systems and thereby helps in further understanding the effect of spatial filtering. As the next step, generation of extended light-sheet for fluorescence microscopy is pro-posed by introducing a specially designed double-window spatial filter at the back-aperture of a cylindrical lens. The filter allows the light to pass through the periphery and center of a cylindrical lens. When illuminated with a plane wave, the proposed filter results in an extended depth-of-focus along with side-lobes which are due to other interferences in the transverse focal plane. Computational studies show a maximum extension of light-sheet by 3:38 times for single photon excitation, and 3:68 times for multi-photon excitation as compared to state-of-art single plane illumination microscopy (SPIM) system and essentially implies a larger field of view. Finally, generation of multiple light-sheet pattern is proposed and demonstrated using a different spatial filter placed at the back aperture of a cylindrical lens. A complete imaging setup consisting of multiple light-sheets for illumination and an orthogonal detection arm, is implemented for volume imaging in fluorescence microscopy. This proposed scheme is a single shot technique that enables whole volume imaging by simultaneously exciting multiple specimen layers. Experimental results confirm the generation of multiple light-sheets of thickness 6:6 m with an inter-sheet spacing of 13:4 m. Imaging of 3 5 m sized fluorescently coated Yeast cells (encaged in Agarose gel-matrix) is per-formed and conclusively demonstrates the usefulness and potential of multiple light-sheet illumination microscopy (MLSIM) for volume imaging. As part of the future scope of the research work presented in this thesis, the Bessel-beam based improved depth microscopy technique may attract applications in particle tracking deep inside tissues and optical injection apart from fluorescence imaging applications. The vectorial formalism derived for cylindrical lens can be used to predict other, complex optical setups involving cylindrical lenses. Extended light-sheet generation proposed in this work by using appropriate spatial filtering with a cylindrical lens, complements the existing and popular selective plane illumination microscopy technique and may facilitate the study of large biological specimens (such as, full-grown Zebra sh and tissue) with high spatial resolution and reduced photobleaching. Finally, the MLSIM technique presented in this thesis may accelerate the field of developmental biology, cell biology, fluorescence imaging and 3D optical data storage.

Esta tese propõe um método de avaliação de sistemas de obtenção de imagem radiológica através das Funções de Transferência simuladas para qualquer região do campo de radiação. Esse processo de simulação reflete as mudanças sofridas pelo ponto focal e, portanto, pela Função de Espalhamento de Ponto (FEP) ao longo do campo. A avaliação utiliza informações obtidas de Funções de Transferência ótica bi-dimensionais calculadas através da aplicação de Transformadas de Fourier sobre as Funções de Espalhamento de Ponto simuladas. O método permite realizar um mapeamento da resposta em freqüências espaciais dos sistemas radio1ógicos para todo o campo de radiação, comparar a nitidez da imagem de sistemas distintos com relação a regiões equivalentes e estudar uma região especifica para prever as distorções que serão inseridas na imagem. / In this work a method of evaluation for the radiological imaging systems performance by simulated transfer functions for any X-ray field region is proposed. The simulation procedures provide information of the changes of the focal spot sizes and, therefore, of the Point Spread Function (PSF) along the field. The evaluation yields the bi-dimensional optical Transfer Functions calculated from Fourier Transformation of the simulated PSFs. This method provides: the spatial frequency response for the entire radiation field of the radiological systems; comparisons of the image sharpness relative to equivalent field regions for different systems; and the study of a singular region in order to predict image distortions.

Light allows to non-invasively study the complex and dynamic biological phenomenon undergoing within cells and tissues in their native state. The development of super-resolution microscopes in recent years has helped to overcome the fundamental limitation imposed by Abbe’s diffraction limit, thereby revolutionizing the field of molecular and cellular biology. With the advancement of various super-resolution techniques (like STED, PALM, and 4Pi) it is now possible to visualize the nanometeric cellular structures and their dynamics in real time. The limitations of existing fluorescence microscopy techniques are: poor axial resolution when compared to their lateral counterpart, and their inability to produce high resolution images of dynamic samples. This thesis covers two broadly connected areas of fluorescence imaging techniques while addressing these limitations. First, the PSF engineering and spatial filtering technique for axial super-resolution microscopy and second, the integration of light sheet illumination PSF with microfluidic cytometry for imaging cells on-the-go. The first chapter gives an explicit description on the fundamentals of fluorescence imaging. This introductory chapter includes a variety of optical microscopes, PSF engineering, the resolution limit imposed by the wave nature of light, the photochemistry of the fluorescent dyes, and their proper selection for fluorescence experiments. In addition to the state-of-art imaging techniques, namely Laser Scanning Confocal Microscopy and Light Sheet Microscopy, this chapter also gives a brief explanation on the evolution of imaging cytometry techniques. Their high speed analytic capability (i.e sorting and counting) makes this technique an important tool in health care diagnosis and other various biomedical applications. The chapter ends with a discussion on the operating principle of the flow cytometers and their limitations. The second chapter in this thesis describes the spatial filtering technique for engineering the PSF to eliminate the side-lobes in the system PSF of the 4Pi Confocal Microscopes. Employing an amplitude mask with binary light transmission windows (also called binary filters), the incident light is structured to minimize the secondary lobes. These lobes are responsible for exciting the off-focal planes in the specimen, hence provide incorrect map of the fluorophore distribution in the object. The elimination of the side-lobes is essential for the artifact-free axial super-resolution microscopy. This second chapter describes the spatial filtering technique in details (its mathematical formulation, application in fluorescence microscopy for generation of desired PSF including Bessellike beam). Specifically, spatial filtering technique is employed in 4Pi type-C Confocal Microscope. The spatial mask used results in the reduction of the side-lobes in 1PE case while they are nearly eliminated in 2PE variant of the proposed technique. The side-lobes are reduced by 46% and 76% for 1PE and 2PE when compared to the existing 4Pi type-C Confocal Microscope system. Moreover, OTF of the proposed system confirms the presence of higher frequencies in the Fourier domain indicating high resolution imaging capability. Apart from the resolution in lateral and axial dimension, achieving high resolution while imaging dynamic samples is another challenge that is limiting the field of fluorescence microscopy to flourish. The third and fourth chapters are entirely dedicated towards the work that was carried out to develop imaging techniques on a microfluidic platform for imaging dynamic samples. The fusion of microscopy and flow cytometry has given rise to the celebrated field of imaging flow cytometry. In recent years, the focus has shifted towards miniaturized cytometry devices. Apart from the reduced cost of the sample reagents and the assays, portability and easy handling make the microfluidic devices more relevant to developing countries. The commercially available cytometers are bulky and quite costly. In addition to these practical concerns, they are complex in operation and limited in performance. Most of the existing cytometers use different inlets for sheath and sample flow to achieve the hydrodynamic focusing of the sample assays in a narrow and confined region. The laser beam in the illumination arm interrogates with the flowing samples at this region and the response is captured by the detection optics. The same principle is extensively used in most of the microfluidic based flow cytometers reported till date. Apart from the hydrodynamic force other effects like electro-osmotic, acoustic, and dielectrophoresis have also been exploited to achieve flow focusing in the microfluidic channel. Despite omitting the necessity of external syringe pump as required in pressure driven based cytometers, they all rely upon point-source based excitation scheme and thereby can not interrogate the cells flowing through the entire microfluidic channel. The third chapter describes the integration of light sheet illumination PSF with microfluidic flow cytometry for simultaneous counting and imaging cells on-the-go. The chapter starts with the description on photolithography procedure for preparing SU8 master and PDMS casting procedure adopted to prepare dedicated microfluidic chips for the developed imaging system. The research work reported here demonstrates the proof-ofprinciple of light sheet based imaging flow cytometer. A light sheet fills the entire microfluidic channel and thus omits the necessity of flow focusing and point-scanning based technology. Another advantage lies in the orthogonal detection geometry that totally cuts-off the incident light, thereby substantially reducing the background in the acquired images. Compared to the existing state-of-the-art techniques, the proposed technique shows marked improvement. Using fluorescently coated Saccharomyces cerevisiae cells, cell counting with throughput as high as 2090 cells/min was recorded. Overall the proposed system is cost-effective and simple in channel geometry. Apart from achieving efficient counting in operational regime of low flow rate, high contrast images of the dynamic samples are also acquired using the proposed cytometry technique. Further, visualization of intra-cellular organelles is achieved during flow in light sheet based high-throughput cytometry system. The fourth chapter demonstrates the proof of concept of light-sheet-based microfluidic cytometer in conjugation with 2π/3 detection system for high-throughput interrogation and high resolution imaging. This system interrogates the flow channel using a sheet of light rather than the existing point-scanning based techniques. This ensures single-shot scanning of specimens flowing through the microfluidic flow channel at variable flow rates. In addition to high throughput counting at low flow rate, visualization of the intra-cellular organelle (mitochondrial network in human cancerous cells) during flow is achieved with sub-cellular resolution. Using mitochondrial network tagged HeLa cells, a maximum count of 2400 cells/min at the optimized flow rate of 700 nl/min was recorded. The 2π/3 detection system ensures efficient photon collection and minimal background caused by scattered illumination light. The other advantage of this kind of detection system which includes 8f detection optics, is the capability to produce variable magnification using the same high NA objective. This thesis opens up in vivo imaging of sub-cellular structures and simultaneous cell counting in a miniaturized flow cytometry system. The developed imaging cytometry technique may find immediate applications in the diverse field of healthcare diagnostics, lab-on-chip technology, and fluorescence microscopy. The concluding chapter summarizes the results with a brief discussion on the future aspects of this field (e.g., live-cell imaging of infectious RBC in microfluidic device and 3D optical sectioning of flowing cells). The field of imaging flow cytometry has immense applications in the overlapping areas of physics and biology. The hydrodynamic forces which are used to achieve flow focusing of the sample assays can have an adverse effect in the cell morphology, thereby altering the cellular functions. Light sheet based cytometry system lifts off the requirement of flow focusing and ensures a single shot scanning of entire samples flowing through the microfluidic channel. The similar concept can be used to study the developmental biology of an entire organism, such as C. elegans. This enables the direct observation of developmental and physiological changes in the entire body. Such an organism can be kept alive for a longer duration in microfluidic chambers, and the neural development and mating behaviors can be extensively studied.

L'un des challenges majeurs de la cosmologie moderne réside en la nature même de la matière et de l'énergie noire. La matière noire peut être directement tracée à travers son effet gravitationnel sur les formes des galaxies. La mission Euclid de l'Agence Spatiale Européenne fournira précisément des données à cette fin. L'exploitation de telles données requiert une modélisation précise de la Fonction d'Étalement du Point (FEP) de l'instrument d'observation, ce qui constitue l'objectif de cette thèse.Nous avons développé des méthodes non-paramétriques permettant d'estimer de manière fiable la FEP sur l'ensemble du champ de vue d'un instrument, à partir d'images non résolues d'étoiles, ceci en tenant compte du bruit, d'un possible sous-échantillonnage des observations et de la variabilité spatiale de la FEP. Ce travail tire avantage d'outils et concepts mathématiques modernes parmi lesquelles la parcimonie. Une extension importante de ce travail serait de prendre en compte la dépendance en longueur d'onde de la FEP. / One of the biggest challenges of modern cosmology is to gain a more precise knowledge of the dark energy and the dark matter nature. Fortunately, the dark matter can be traced directly through its gravitational effect on galaxies shapes. The European Spatial Agency Euclid mission will precisely provide data for such a purpose. A critical step is analyzing these data will be to accurately model the instrument Point Spread Function (PSF), which the focus of this thesis.We developed non parametric methods to reliably estimate the PSFs across an instrument field-of-view, based on unresolved stars images and accounting for noise, undersampling and PSFs spatial variability. At the core of these contributions, modern mathematical tools and concepts such as sparsity. An important extension of this work will be to account for the PSFs wavelength dependency.

This thesis describes an innovative method for topographic measurement of diffuse surfaces. Tested surface is measured indirectly using nanoparticles distributed across the studied area. An image of every particle is captured by CCD camera as a double helix point spread function whose angular rotation corresponds to local surface height. Used point spread function is the result of an interference of non-diffracting vortex beams that are formed by a spiral phase mask from light originating from a nanoparticle. Diploma thesis presents an overview of current techniques for surface topography measurement. Next, working principle of proposed method is described and its experimental application is discussed. An influence of signal-to-noise ratio and image sampling on reconstruction precision is studied using numerical simulations and, as a result, optimal experimental parameters are proposed. Practical potential of the method is demonstrated by 3D reconstruction of planar and spherical surfaces in the depth range of up to 9 times the depth of focus of used microscope objective.

No description available.

生活中在使用相機時，由於機器晃動或物體移動所造成的模糊影像時常可見。當影像模糊的成因是影像曝光時間內相機與拍攝物體相對線性移動時，則我們稱為線性動態模糊。理論上，模糊影像可以表示成原始影像與點擴散函數的旋積，本文的研究重點為點擴散函數中模糊參數的估計，雷登轉換將被運用在此問題上。我們首先介紹兩個現有方法，我們將探討這些方法中用來消除雜訊的步驟之適用性及必要性。另一方面，在模糊參數的估計過程中，我們在雷登轉換加入圓限制以及採用移動平均法。我們透過實驗證實，本篇提出的方法可以獲得更準確的估計結果以及更好的模糊影像還原效果。 / Nowadays, collecting a digital image becomes convenient and low-cost due to rapid progress in digital camera technology. Blurred images frequently appear because of camera shake or moving objects. There are several different types of blur. When the blur is caused by the linear motion between the object and the camera during the light exposure, it’s called a linear motion blur. Mathematically, a blurred image is expressed as a convolution of a point spread function and the original image. Our study considers Radon transform for the estimation of the point spread function. To improve the existing methods, a circle restriction and the moving average method are applied in the estimating procedure. Through intensive experiments, the proposed method is found enable to produce more accurate estimation and better performance in image restoration.

Apresenta-se o desenho, construção e caracterização de uma plataforma experimental para a obtenção de imagens magnéticas bidimensionais (2D) geradas pela distribuição não uniforme em gel de vaselina de micro-partículas magnéticas (magnetita- Fe3O4), acomodadas em fantomas magnéticos de geometrias irregulares. A instrumentação é basicamente formada por um arranjo multicanal de 12-sensores magnetorresistivos de última geração (modelo HMC 1001/1002 da Honeywell), os quais convertem os sinais magnéticos, a serem medidas, em voltagens diferenciais, que posteriormente passam-se pela etapa de condicionamento analógico multisinais, e adquiridos por uma placa de aquisição PCI de 16 canais simples, e geradas pelas fontes magnéticas (fantomas) as quais eram posicionadas acima de uma tabua porta-fantoma a qual era acionada por um sistema de posicionamento x-y, utilizando-se dois motores de passo controlados via porta paralela. A obtenção e processamento das imagens de forma automática foi levado acabo por médio da ferramenta computacional SmaGimFM v1.0 (grupo de scripts escritos pelo autor, em LABVIEW v8.1 e Matlab v7.3). A montagem experimental foi desenhada para realizar o scan numa área de ate (20x18) cm2. O sistema consegue medir campos na ordem de poucos nano-teslas (10-9 T). Foi demostrado experimentalmente que: a detectibilidade do sistema está na ordem de 100 pT/?Hz; a resolução, o menor valor da indução magnética detectada e a resolução espacial dos sensores foi aproximadamente de (3±1) nT e (3.0± 0.1) mm, respectivamente, este último obtido para uma distancia sensor-fonte média de (6.0± 0.1) mm. O nível de ruído ambiental médio foi corroborado experimentalmente no valor de 10 nT. O fator de Calibração para todos os sensores alimentados com 8V, foi aproximadamente de 10-6 T/V, confirmando o valor da sensibilidade nominal oferecida pelo vendedor no data-sheet dos sensores. Os multisinais sempre foram pré-processadas para a remoção dos offset, e posteriormente era realizadas uma interpolação bi-cúbica, para gerar imagens magnéticas com uma alta resolução espacial da ordem de (256x256) pixels. A funções de transferência da modulação e espalhamento pontual do sistema foram estudados e os sensores foram espaçados e fixados de acordo com os resultados destes estudos. Nesta tese todas as imagens cruas foram geradas pelo mapeamento da resposta do sistema multicanal de magnetômetros a pequenas distancias e geradas pela presença de micropartículas de magnetita (Fe3O4) não tratada termicamente e dispersada em oitos fantomas planares com geometrias complexas e chamados como: PhMão; PhNum; PhLines; PhCinco; PhTrês; PhCircle; PhQuadSmall e PhQuadBig. As imagens magnéticas de cada um destes fantomas é apresentada. A cada experimento, estes fantomas eram magnetizados pela ação de um pulso magnético uniforme no volume dos fantomas, com um valor aproximadamente de 81.6 mT, e produzido por um sistema de bobinas par de Helmholtz. Para fazer o registro experimental das imagens magnéticas, os fantomas foram posicionados a uma altura fixa em relação aos sensores, e movidos numa direção de scan, assim nos detectores observávamos as voltagens gerados pela variação no campo remanente devido às diferentes concentrações de micro-partículas magnéticas magnetizadas foram medidos e controlados por um computador pessoal. Usando as imagens cruas (imagens ruidosas e borradas) e outras informações a priori, foram obtidas as imagens reconstruídas das fontes do campo magnético, tais como, a distribuição de partículas ferrimagnéticas no interior dos fantomas, a qual é relacionada com a susceptibilidade magnética das amostras. Encontrar as imagens das fontes magnéticas, é resolver o problema magnético associado, e nosso trabalho estas restaurações foram realizadas usando-se os seguintes algoritmos numéricos de deconvolução, filtragem espacial de Wiener e Fourier, o filtragem Pseudo-inversa, o método do gradiente conjugado e os procedimentos de regularização de Tikhonov e Decomposição de Valores singulares truncados, dentre outros. Estes procedimentos foram implementados e testados. As imagens reconstruídas das fontes magnéticas de quatro fantomas são apresentadas. Estas técnicas foram programadas computacionalmente por médio de um conjunto de scripts chamados de SmaGimFM v1.0, estes foram escritos nos linguagens computacionais MATLAB® desde a MathWorks Inc.; e LABVIEW desde a National Instruments Inc. Estes resultados preliminares mostram que o sistema de imagens apresenta potencial para ser aplicada em estudos na área da Física Médica, onde imagens com moderada para alta resolução espacial e baixa amplitude da indução magnética são exigidas. Contudo, podemos afirmar que à distância sensor-fonte é crítica e afeta a resolução das imagens. O sistema é capaz de registrar imagens na ordem de 10-9 T, e sua elevada resolução espacial indica que pode ser testada como uma nova técnica biomagnética para gerar imagens em 2D de partículas magnéticas dentro de objetos, na região do campo próximo, para futuras aplicações médicas / We have designed and build a magnetic imaging system for obtaining experimental noisy and blurred magnetic images from distribution of ferromagnetic tracers (magnetite Fe3O4). The main part of the magnetic imaging system was formed by a linear array composed of 12-magnetoresistive sensors from Honeywell Inc. (HMC 1001). These sensors are microcircuits with a configuration of wheatstone-bridge and convert magnetic fields into differential voltage, which after pass for the multichannel signal stage can be to measure magnetic signals about of 10-9 T. The system is capable of scanning planar samples with dimensions up to (16x18) cm square. A full experimental characterization of the magnetic imaging system was carried out. The calibration factor for all sensor supplied by 8 V, was approximately 10-6 T/V, confirming the data sheet nominal properties from the vendor. The spatial resolution and the resolution of the magnetic imaging system were experimentally confirmed to be 3 mm and 3 nT, respectively. The spectral density noise was about , for the experimental conditions used in these studies. The signals were pre-processed for offset remove and the interpolation for spatial resolution improves and generates images of (256x256) pixels. The point spread and modulation transference functions of multi-sensor system were studied and the sensors were spaced accordingly. In this thesis, all raw images were generated by mapping the response of the magnetoresistive magnetometers multichannel array at short distances due to the presence of uncooked magnetite powder dispersed in eight planar phantoms with complex geometries and called as: PhMão; PhNum; PhLines; PhCinco; PhTrês; PhCircle; PhQuadSmall and PhQuadBig. These phantoms were magnetized by a uniform pulse field of approximately of 81.6 mT produced by a Helmholtz coil system. The samples were moved under the magnetoresistive sensors and the voltages generated by the variation in remanent magnetic field due to different magnetized ferromagnetic particles concentrations were recorded and controlled by a personal computer. Using the experimental noisy and blurred magnetic field images (raw images), and some another, a priori information\'s, the reconstruction of the magnetic field source images, such as, the distribution of ferromagnetic particles inner of the phantoms which are related with magnetic susceptibility, was obtained by various inverse problem solution algorithms\', such as, the spatial Wiener and Fourier filtering, the Pseudo-inverse filtering; the conjugated gradient and Tikhonov and Decomposition of Truncated Singular Values approaches and others. These procedures were implemented by mean of the scripts set called SmaGimFM v1.0, that we developed using the MATLAB® language from MathWorks Inc. A preliminary result shows that this magnetic imaging system join to some deconvolution technique can be considered efficient to be used in functional images of the gastrointestinal tract, where a moderate resolution is required. We can affirm that at a distance sensor-source choose is a critical parameter and affects the resolution of the images; and we can conclude that this magnetic images method can be successfully used to generate planar blurred magnetic images and magnetic field sources images in the near field region at macroscopic level generated by ferromagnetic materials.