A digital speckle-pattern interferometer was built utilizing a 100 x 100 element Reticon diode array interfaced to an HP-9836C desk-top computer. A single-mode optical fiber mounted in the center of the system's aperture stop creates a spherical-wave reference beam. Secondary interference fringes are calculated inside the computer by subtracting speckle patterns before and after a deformation, and squaring this difference. This technique has been shown superior to that of taking the absolute value of the difference. The traditional vibrational observation technique of low-pass filtering a single speckle pattern and squaring the result is emulated in software. It is compared to four other vibration observation techniques. A new technique records the self-interference terms in a reference frame, and subtracts these from the time-averaged vibration data. It provides very good fringe contrast for moderately unstable objects, as well as interferometers which have not been optimized to minimize self-interference terms. The best vibration fringe contrast is obtained by subtracting two time-averaged speckle patterns of a single object resonance. One exposure has a relative π phase-shift between object and reference beams to cancel self-interference terms. This last technique is not real-time; whereas, the new technique is. Double-exposure speckle interferograms are averaged using a technique which utilizes a stepping motor to change the object illumination angle. Results of averaging deformation measurements and double-exposure vibration techniques show a large increase in fringe contrast with an accompanying reduction in speckle noise. Quantitative measurements of object deformations is demonstrated by applying phase-shifting interferometry techniques. A deformation's phase is found by subtracting modulo 2π phases measured for each object state. Phase discontinuities are removed with the aid of noise reduction algorithms. Limitations are low intensity modulation as the phase is shifted, and speckle decorrelation during object deformation. It is shown that 10 waves of object deformation are measurable to λ/10 across the detector array. The double-exposure phase-measurement technique works well and is applicable to many different metrological measurements. To show the versatility of this technique, an optically smooth surface is contoured using two illumination wavelengths.

München, Techn. Univ., Diss., 2002. / Computerdatei im Fernzugriff.

München, Techn. Univ., Diss., 2002. / Computerdatei im Fernzugriff.

München, Techn. Universiẗat, Diss., 2004.

München, Techn. Universiẗat, Diss., 2002.

Oldenburg, University, Diss., 2002.

Techn. University, Diss., 2003--Barcelona.

Orientadores : Inacio Maria Dal Fabbro, Hector Jorge Rabal / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Agricola / Made available in DSpace on 2018-07-27T01:05:18Z (GMT). No. of bitstreams: 1 BragaJunior_RobertoAlves_D.pdf: 6979173 bytes, checksum: b525fe6093bb3a41b164123459962f4b (MD5) Previous issue date: 2000 / Doutorado

Speckle noise is an integral part of any laser projection display because it is the nature of laser (coherent) illumination to form interference patterns with high visibility. The granularity of the image due to the speckle formed on the viewer's retina degrades the image quality, thus stimulating the need for speckle reduction techniques applied to laser displays. An instrument to measure image plane speckle contrast was built based on a linear CCD detector array interfaced to an LSI-11 microcomputer. Speckle reduction techniques were then evaluated by comparing the contrast obtained with each method. The effect of the spatial frequency response (MTF) on the measured contrast was studied, along with the statistical significance of the measurement which is limited by the finite sample space of 1024 detector pixels per CCD frame. The lowering of the contrast due to the array MTF can be minimized by working at extremely high F numbers (> 100). The sample space can be widened by taking more than one frame of data and treating all the frames as a single data set. Techniques to reduce speckle noise in laser displays fall into two broad categories: reducing the coherence of light forming the speckle, and incoherently adding multiple uncorrelated (or partially correlated) speckle patterns. The first technique (effective only for monochromatic displays) was implemented by coating a screen with various dyes, phosphors, or fluorescent paints. Using the 514 nm line from an Argon laser, the contrast can be reduced by almost 30% by spraying a thin layer of fluorescent paint on the screen. More speckle reduction can be achieved with an accompanying loss in image brightness. The second technique involved creating a multiplicity of partially correlated speckle patterns that appear from the same position on the screen over the integration period of the eye. The different speckle patterns are produced by changing the angle of illumination while keeping a portion of the laser spot focused on the same point on the screen. The scan angle method (applicable to multi-color displays), can be implemented by properly synchronizing an acousto-optic modulator with the scan optics, and imaging the modulator on the screen. Using a beaded screen and a reasonable laser dither of 10 millirads, the contrast can be reduced by half.

L'objectif de la caractérisation des tissus par ultrasons ou ‘Quantitative Ultrasound (QUS)’ est de différencier les tissus pathologiques en associant les paramètres d’un modèle aux caractéristiques physiques du tissu. L'usage exclusif des ultrasons pour obtenir un diagnostic peut garantir que le patient ne subira pas une procédure invasive (e.g. une biopsie), utilisant des rayonnements ionisants (e.g. la tomographie) ou même inconfortable et coûteuse (e.g. IRM). Les méthodes de QUS extraient des informations sur la microstructure du tissu à partir du contenu spectral ou temporel des signaux ultrasonores. Le signal temporel radiofréquence (RF) et son enveloppe sont d'intérêt à cause du speckle crée par l’interférence des ondes, qui peut être modélisé par des distributions statistiques. Ce travail propose d'explorer la possibilité d'obtenir des estimations QUS fiables en utilisant des distributions statistiques comme modèles pour le speckle ultrasonore. Les estimations sont constituées des paramètres des distributions respectives et dépendent de la densité de diffuseurs dans le milieu. L’évaluation s’effectue sur des images simulées, des fantômes de particules et des biofantômes. Dans la première partie, la distribution Gaussienne Généralisée est utilisée pour modéliser le signal RF, et la distribution de Nakagami est utilisée pour modéliser son enveloppe. Les deux distributions sont limitées à discriminer les milieux avec une faible densité de diffuseurs, parce que les valeurs de leurs paramètres de forme saturent pour un speckle pleinement développé. Par conséquent, puisque la formation du speckle pleinement développé dépend de la résolution du système d'imagerie, la caractérisation peut se faire seulement à de très hautes résolutions, correspondant à des hautes fréquences qui ne sont pas communes en échographie clinique. Une application du modèle de Nakagami sur l’image crée par la seconde harmonique montre le potentiel du paramètre de forme de Nakagami en tant que mesure de la nonlinéarité du milieu. Dans la deuxième partie, l'enveloppe a été modélisée en utilisant la distribution K-Homodyne. Le paramètre de regroupement des diffuseurs α permet de discriminer entre les milieux denses jusqu’à une limite supérieure à celle du paramètre de Nakagami. Pourtant, cette limite est difficile à estimer avec précision, parce que les valeurs caractéristiques pour le speckle pleinement développé sont affectées par un biais et une variance élevés. Le biais et la variance peuvent être améliorés en augmentant la quantité de données utilisée pour l’estimation. Dans la dernière partie, une technique de déconvolution spécialement conçue pour la caractérisation des tissus a été évaluée. Des essais exhaustifs ont montré qu’elle n’est pas suffisamment robuste pour une application clinique, puisque les images déconvoluées ne sont pas fidèles à la réflectivité originale du milieu / The purpose of ultrasound tissue characterization or Quantitative Ultrasound (QUS) is to differentiate between tissue pathologies by associating model parameters to physical tissue features. The exclusive use of ultrasound for diagnosis would guarantee that the patient does not undergo a procedure that is invasive (e.g. a biopsy), using ionizing radiation (e.g. tomography) or simply uncomfortable and expensive (e.g. MRI). QUS methods extract information on the tissue microstructure from the temporal or spectral content of the acquired ultrasound signals. The temporal radiofrequency (RF) signal and its envelope are of interest because of the speckle patterns created by wave interference, which can be modeled by statistical distributions. The present work proposes to explore the possibility of obtaining reliable QUS estimates by using statistical distributions as models for ultrasound speckle. The estimates consist in the parameters of the respective distributions and are indicators of the scatterer density in the medium. The evaluation is conducted on simulated images, particle phantoms and biophantoms. In the first part, the Generalized Gaussian distribution is used to model the RF signal, and the Nakagami distribution is used to model its envelope. The two distributions show limitations in discriminating media with high scatterer densities, as the values of their shape parameters saturate in the fully developed speckle regime. Therefore, since the formation of fully developed speckle depends on the resolution of the imaging system, characterization can be done only at very high resolutions, corresponding to high frequencies that are not common in clinical ultrasound. An application of the Nakagami model on the second harmonic image shows the potential of the Nakagami shape parameter as a measure of the nonlinearity of the medium. In the second part, the echo envelope was modeled using the Homodyned-K distribution. The scatterer clustering parameter α allows the discrimination of dense media up to a concentration that is higher than the one that limits the Nakagami distribution. However, this limit is difficult to estimate precisely, because the values of α that are characteristic for fully developed speckle suffer from large estimation bias and variance. The bias and the variance can be improved by performing the estimation on a very large amount of data. In the final part, a deconvolution technique designed specifically for ultrasound tissue characterization has been analyzed. Extensive testing has shown it to not be sufficiently robust for clinical applications, since the deconvolved images are not reliable in terms of fidelity to the original reflectivity of the medium