Thesis (M.S.)--California Polytechnic State University, 2010. / Mode of access: Internet. Title from PDF title page; viewed on April 20, 2010. Major professor: John Saghri. "Presented to the Electrical Engineering Department faculty of California Polytechnic State University, San Luis Obispo, California." "In partial fulfillment of the requirements for the degree of Master of Science in Electrical Engineering." "March 2010." Includes bibliographical references (p. 119-125).

Thesis (Ph. D.)--Carleton University, 1999. / Also available in electronic format on the Internet.

Thesis (Ph. D.)--Oregon Graduate Center, 1987.

Orientador: João Marcos Travassos Romano / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-05T07:17:28Z (GMT). No. of bitstreams: 1 Ferrari_Rafael_M.pdf: 3590603 bytes, checksum: 36e018a11a02c47baafbf4b3682e9f1d (MD5) Previous issue date: 2005 / Resumo: Esta tese objetiva o estudo da utilização de fuzzy na equalização supervisionada e não-supervisionada de canais de comunicação digital. O trabalho se divide em basicamente duas partes. Na primeira, é feita uma revisão das técnicas de equalização supervisionada empregando filtros nebulosos mostradas na literatura. Na segunda parte concentramse as principais contribuições originais, voltadas ao estabelecimento de um paradigma sólido para equalização não-linear cega. Primeiramente, demonstramos que o teorema de Benveniste-Goursat-Ruget não garante a equalização quando do uso de não lineares como os fuzzy. Como alternativa, optamos por um critério baseado no erro de predição aliado a estruturas nebulosas, opção esta que se mostra plenamente justificada pela demonstração da equivalência entre os preditores fuzzy e o estimador de mínimo erro quadrático médio. Para efetuar o treinamento desta estrutura foi proposto um algoritmo baseado em clusterização não-supervisionada que combina estratégias evolutivas com técnicas de busca local. Porem, resultados de simulações computacionais são apresentados afim de avaliar e comparar com as soluções clássicas o desempenho dos equalizadores e das técnicas de treinamento descritos no trabalho / Abstract: The objective of this thesis is to study the application of fuzzy to supervised and unsupervised digital channel equalization. Our work is basically divided in two main parts. In the one, we make an extensive review of supervised fuzzy equalization techniques. In the second part, we present original contributions towards the establishment of a solid paradigm for blind nonlinear equalization. In this part, we demonstrate that the Benveniste-Goursat-Ruget theorem is not valid for nonlinear equalizers such as fuzzy liters. As a viable alternative, we propose an approach based on the predictionerror criterion and a fuzzy logic system. The eectiveness of which is cornered by the demonstration of the equivalence between the fuzzy predictor and the minimum eansquare error estimator. Secondly, we propose a training scheme founded on an unsupervised clustering algorithm that combines evolutionary strategies and local search techniques. Lastly, we present results of computational simulations to assess the performance of the equalizers and training techniques introduced in our work / Mestrado / Telecomunicações e Telemática / Mestre em Engenharia Elétrica

Orientadores: Roberto de Alencar Lotufo, Letícia Rittner / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-25T05:00:23Z (GMT). No. of bitstreams: 1 Souza_RobertoMedeirosde_M.pdf: 27462483 bytes, checksum: fd6e6b42169addd0201eeda81c058aea (MD5) Previous issue date: 2014 / Resumo: A Árvore de Componentes é uma estrutura de dados que representa uma imagem através da relação de hierarquia de seus componentes conexos. Ela é uma estrutura adequada para a implementação de filtros conexos e que foi utilizada com sucesso em muitas aplicações. A Árvore Máxima é uma estrutura compacta para a representação da Árvore de Componentes. A principal contribuiçãoo deste trabalho é a proposta do filtro de Simplificação Maximal da Árvore Máxima (MMS) com dois possíveis critérios para efetuar o seu cálculo: um critério de limiarização normalizada (MMS-T) e um critério de Regiões Extremais Maximamente Estáveis (MMS-MSER). Uma metodologia para aplicar o filtro MMS em associação com o filtro de Extinção, que é formalmente definido nesse trabalho, é apresentada. É mostrado que após a aplicação da metodologia de simplificação, a qual escolhe o número de máximos relevantes a serem mantidos na imagem, o número de nós da Árvore Máxima simplificada é no máximo duas vezes o número de máximos mantidos. Para definir o filtro MMS, novos conceitos, como nó composto e sub-ramo são apresentados. Esses conceitos são importantes para definir muitos algoritmos da Árvore Máxima, e eles possuem interpretações interessantes em termos de processamento de imagem. Possíveis aplicações da metodologia proposta, tais como localização de texto, simplificação/segmentação de imagens e reconhecimento de objetos são ilustrados para mostrar o potencial da metodologia. Também, estudos explortatórios de detecção de regiões salientes em imagens e análise da robustez da topologia da Árvore Máxima são apresentados / Abstract: The Component Tree is a data structure that represents an image through the hierarchical relationship of its connected components. It is an adequate structure to implement connected filters, and it has been successfully used in many applications. The Max-Tree is a compact structure for the Component Tree representation. The main contribution of this work is the proposal of the Maximal Max-Tree Simplification (MMS) filter with two possible criteria to compute the filter: a normalized threshold criterion (MMS-T) and a Maximally Stable Extremal Regions (MSER) criterion (MMS-MSER). A methodology to apply the MMS filter in association to the Extinction filter, which is formally defined in this work, is presented. It is shown that after applying our simplification methodology, which sets the number of relevant maxima in the image to be kept, the number of nodes in the simplified Max-Tree is at most twice this number. In order to define the MMS filter, new concepts, such as composite node and sub-branches are introduced. These concepts are important to define many Max-Tree algorithms, and they have interesting interpretations in terms of image processing. Possible applications of the methodology proposed, such as text location, object recognition, and image simplification/segmentation are illustrated to demonstrate the potential of this methodology. Also, exploratory studies, such as detection of distinguished regions in the image, and analysis of the robustness of the Max-tree topology are presented / Mestrado / Engenharia de Computação / Mestre em Engenharia Elétrica

This thesis presents an analysis of several smoothness energies (also called smoothing energies) in geometry processing, and introduces new methods as well as a mathematical proof of correctness and convergence for a well-established method. Geometry processing deals with the acquisition, modification, and output (be it on a screen, in virtual reality, or via fabrication and manufacturing) of complex geometric objects and data. It is closely related to computer graphics, but is also used by many other fields that employ applied mathematics in the context of geometry. The popular Laplacian energy is a smoothness energy that quantifies smoothness and that is closely related to the biharmonic equation (which gives it desirable properties). Minimizers of the Laplacian energy solve the biharmonic equation. This thesis provides a proof of correctness and convergence for a very popular discretization method for the biharmonic equation with zero Dirichlet and Neumann boundary conditions, the piecewise linear Lagrangian mixed finite element method. The same approach also discretizes the Laplacian energy. Such a proof has existed for flat surfaces for a long time, but there exists no such proof for the curved surfaces that are needed to represent the complicated geometries used in geometry processing. This proof will improve the usefulness of this discretization for the Laplacian energy. In this thesis, the novel Hessian energy for curved surfaces is introduced, which also quantifies the smoothness of a functions, and whose minimizers solve the biharmonic equation. This Hessian energy has natural boundary conditions that allow the construction of functions that are not significantly biased by the geometry and presence of boundaries in the domain (unlike the Laplacian energy with zero Neumann boundary conditions), while still conforming to constraints informed by the application. This is useful in any situation where the boundary of the domain is not an integral part of the problem itself, but just an artifact of data representation---be it, because of artifacts created by an imprecise scan of the surface, because information is missing outside of a certain region, or because the application simply demands a result that should not depend on the geometry of the boundary. Novel discretizations of this energy are also introduced and analyzed. This thesis also presents the new developability energy, which quantifies a different kind of smoothness than the Laplacian and Hessian energies: how easy is it to unfold a surface so that it lies flat on the plane without any distortion (surfaces for which this is possible are called developable surfaces). Developable surfaces are interesting, as they can be easily constructed from cheap material such as paper and plywood, or manufactured with methods such as 5-axis CNC milling. A novel definition of developability for discrete triangle meshes, as well as a variety of discrete developability energies are also introduced and applied to problems such as approximation of a surface by a piecewise developable surface, and the design and fabrication of piecewise developable surfaces. This will enable users to more easily take advantages of these cheap and quick fabrication methods. The novel methods, algorithms and the mathematical proof introduced in this thesis will be useful in many applications and fields, including numerical analysis of elliptic partial differential equations, geometry processing of triangle meshes, character animation, data denoising, data smoothing, scattered data interpolation, fabrication from simple materials, computer-controlled fabrication, and more.

Presented is a method for designing and implementing lattice filters to be used in Quadrature Mirror Filter (QMF) Banks. Quadrature Mirror Filter Banks find use in applications where a signal must be spilt into subbands operated on then reconstructed in the output. Because of their structure, lattice filters do this very well and allow perfect reconstruction, even when the lattice coefficients must be quantized. In this paper QMF's and Lattice Filters are derived and analyzed. Application of the lattice filter is presented along with a design program and example of its use to implement a QMF. The computer aided design procedure allows the user to input the stop-band frequency, normalized to the sampling frequency, and the desired attenuation. The resulting outputs are the lattice coefficients, and the Finite Impulse Response (FIR) coefficients of an FIR filter having the same characteristics. The program selects a set of coefficients based on optimal coefficients that are within the desired tolerance. The filter design program was written in FORTRAN, with the filter coefficients stored in a data file on disk. Programs were written in MATHCAD© to show the lattice filter response and to simulate the QMF using these coefficients.

An investigation of the non-uniform sampling technique as applied to digital filter designs will be made. The objective of the design is to reduce the interference problems as one would· encounter in using uniform sampling technique in the synthesis. An analysis of the error function which measures the goodness in approximating a desired frequency response will also be undertaken. An algorithm which determines the optimal parameters.for a high pass filter will be developed and used to synthesize the particular high pass filter. The results of this design, the frequency response and its approximation error will be studied and evaluated. / Master of Science

The potential is vast for the enhancement of graphic images that are used for human interpretation and/or image display. Sensor noise, blur due to camera misfocus, relative object-camera motion, and random atmospheric turbulence may contribute to photographic image deterioration. Image enhancement and restoration methods are useful for improving the quality and for augmenting specific characteristics of an image. Often, complex filtering of the image data is required. This paper presents a technique for enhancement of images using multistage filtering techniques which take advantage of a priori knowledge as to the images' content. Algorithms are designed and implemented which enhance edges as well as gray level contrast. In addition, the source images are passed through a sequence of controllable filter stages to provide varying degrees and types of enhancements. Image enhancement techniques do not increase the inherent information content in the data, but the techniques do accentuate distinct image features which result in an improved image display. The effects of the different filters and filter stages will be analyzed with regard to image enhancement and picture quality. Numerical results and graphic image results are included in the analysis. Further applications of the techniques analyzed are discussed as well.

No description available.