In this dissertation we address the problem of adapting frequency domain tiling using the curvelet transform as the basis algorithm. The optimal tiling, for a given class of images, is computed using denoising performance as the cost function. The major adaptations considered are: the number of scale decompositions, angular decompositions per scale/quadrant, and scale locations. A global optimization algorithm combining the three adaptations is proposed. Denoising performance of adaptive curvelets is tested on seismic and face data sets. The developed adaptation procedure is applied to a number of different application areas. Adaptive curvelets are used to solve the problem of sparse data recovery from subsampled measurements. Performance comparison with default curvelets demonstrates the effectiveness of the adaptation scheme. Adaptive curvelets are also used in the development of a novel image similarity index. The developed measure succeeds in retrieving correct matches from a variety of textured materials. Furthermore, we present an algorithm for classifying different types of seismic activities.

Στην παρούσα διπλωματική εργασία παρουσιάζεται η θεωρία στην οποία βασίζεται η δημιουργία των αλγόριθμων αποθορυβοποίησης Boll, Berouti, Multiband, Wiener και Ακουστου Θορύβου (από Τσουκαλά). Μεσω γραφικού περιβάλλοντος GUI (Matlab) πραγματοποιήθηκαν μετρήσεις με αντικειμενικές μεθόδους μέτρησης ομιλίας για την σύγκριση της απόδοσης των αλγόριθμων αυτών. / In this essay, the theory in which 5 basic denoising algorithms are based (Boll, Berouti, Multiband, Wiener και Tsoukalas) is presented. Through a GUI (Matlab), subjective measurements were taken place in order to compare the performance of these algorithms.

Tese (doutorado)—Universidade de Brasília, Faculdade de Tecnologia, Departamento de Engenharia Elétrica, 2015. / Submitted by Albânia Cézar de Melo (albania@bce.unb.br) on 2016-03-29T13:45:46Z No. of bitstreams: 1 2015_AnaPaulaBernardiSilva.pdf: 2760403 bytes, checksum: 22bc53172314b3b37f4ab1aa008f9399 (MD5) / Approved for entry into archive by Raquel Viana(raquelviana@bce.unb.br) on 2016-03-29T16:39:10Z (GMT) No. of bitstreams: 1 2015_AnaPaulaBernardiSilva.pdf: 2760403 bytes, checksum: 22bc53172314b3b37f4ab1aa008f9399 (MD5) / Made available in DSpace on 2016-03-29T16:39:10Z (GMT). No. of bitstreams: 1 2015_AnaPaulaBernardiSilva.pdf: 2760403 bytes, checksum: 22bc53172314b3b37f4ab1aa008f9399 (MD5) / A Velocidade de Condução (VC) é um parâmetro fisiológico básico que fornece informações sobre o sistema neuromuscular, utilizada para identificar fadiga e patologias. Este trabalho propõe algoritmos para o cálculo da VC no domínio temporal e uma sistematização de procedimentos para otimizar o seu cálculo. Dado um arranjo linear de sinais Eletromiográficos de Superfície (EMGS), a sistematização proposta consiste de três etapas distintas: (i) seleção dos canais que serão utilizados no cálculo; (ii) aplicação do denoising para minimização do ruído branco gaussiano nos canais selecionados; e (iii) cálculo da velocidade de condução. Na primeira etapa é apresentado um método (independente da intervenção humana) para selecionar os canais de um sinal de EMGS que representem o fenômeno desejado e que sejam similares. Esta seleção é realizada utilizando a distribuição do espectro de frequência dos sinais no intervalo de frequência útil. Na segunda etapa são propostos dois parâmetros de limiariarização para o denoising invariante ao deslocamento que minimizem o ruído branco obtidos de um processo experimental exaustivo. Nesta etapa, também é apresentado um comparativo de desempenho dos métodos e parâmetros mais conhecidos de limiarização utilizados no denoising com os aqui propostos. Na terceira etapa a VC é estimada, por meio de dois algoritmos no domínio temporal, que descrevem o parâmetro como uma dispersão de valores instantâneos. Um dos métodos utiliza o deslocamento dos potenciais de ação respectivos em diferentes canais. O segundo método calcula o coeficiente angular entre um conjunto de canais. O método proposto para a etapa seleção de canais apresentou aproximadamente 98% de correta classificação dos sinais. Não há como comparar este resultado, pois não foi encontrado trabalho similar. Na etapa referente a aplicação do Denoising foi constatado que, para os sinais utilizados neste trabalho, os limiares clássicos SURE e Hibrido apresentam os maiores valores de RSR. O limiar 1, proposto neste trabalho apresentou resultados bem semelhantes aos otimizadores clássicos. Os métodos propostos para o cálculo da VC não necessitaram de utilização de canais consecutivos. As dispersões apresentadas para representar a VC apontaram que existe uma variação do parâmetro ao longo do tempo. As etapas da sistematização se mostraram essenciais para reduzir as limitações de um método de cálculo da VC no domínio temporal. / The conduction velocity (CV) is a basic physiological parameter that measures how fast an electrochemical impulse propagates through the neuromuscular system. The CV parameter is used to identify fatigue and pathologies, for instance. This work deals with novel algorithms to infer conduction velocity of white fiber in the time domain. Also, I propose a systematization process in order to estimate the conduction velocity which minimizes the inherent drawbacks of the time domain. Taken a linear arrange of skin surface electromyographic signals (EMGS) the proposed systematization has three distinct stages: (i) channel selection that are utilized for the calculation; (ii) denoising to minimize white Gaussian noise (AWGN) of the selected channels; and (iii) the conduction velocity estimation itself. In the first stage, a method (free from human intervention) that selects the channel of the EMG signal which are representative of the expected phenomenon and are also similar, is presented. In the second stage, I propose two thresholds for denoising were proposed, which are invariant to the displacement that minimizes AWGN. Also, in this stage, I compare the proposed methods with standart methods in the current literature. In the last stage, the conduction velocity is estimated by means of a time domain pair of algorithms, which describe the parameter as a dispersion of the real time values. The first algorithms uses the delay of action potentials on difference channels. The second method computes the slope between a set of channels. Results: The proposed method for channel selection step has approximately a 98% correct classification of signals. There is no way to compare this outcome because it was not found similar work. The classical thresholds SURE and Hybrid showed the highest SNR (Signal Noise Ratio) values for signals used in this work. The first threshold, proposed in this work showed very similar results to classic optimizers. The methods proposed for the calculating the CV does not require the use of consecutive channels. Dipersions presented to represent the result of CV showed that there is a variation of the parameter over time. The stages of systematization proved essencial to reduce the limitations of a CV estimation in the time domain.

No description available.

Most digital cameras capture images through Color Filter Arrays (CFA), and reconstruct the full color image from the CFA image. Each CFA pixel only captures one primary color component at each pixel location; the other primary components will be estimated using information from neighboring pixels. During the demosaicking algorithm, the unknown color components will be estimated at each pixel location. Most of the demosaicking algorithms use the RGB Bayer CFA pattern with Red, Green and Blue filters. Some other CFAs contain four color filters. The additional filter is a panchromatic/white filter, and it usually receives the full light spectrum. In this research, we studied and compared different four channel CFAs with panchromatic/white filter, and compared them with three channel CFAs. An appropriate demosaicking algorithm has been developed for each CFA. The most well-known three-channel CFA is Bayer. The Fujifilm X-Trans pattern has been studied in this work as another three-channel CFA with a different structure. Three different four-channel CFAs have been discussed in this research: RGBW-Kodak, RGBW-Bayer and RGBW- $5 \times 5$. The structure and the number of filters for each color are different for these CFAs. Since the Least-Square Luma-Chroma Demultiplexing method is a state of the art demosaicking method for the Bayer CFA, we designed the Least-Square method for RGBW CFAs. The effect of noise on different CFA patterns will be discussed for four channel CFAs. The Kodak database has been used to evaluate our non-adaptive and adaptive demosaicking methods as well as the optimized algorithms with the least square method. The captured values of white (panchromatic/clear) filters in RGBW CFAs have been estimated using red, green and blue filter values. Sets of optimized coefficients have been proposed to estimate the white filter values accurately. The results have been validated using the actual white values of a hyperspectral image dataset. A new denoising-demosaicking method for RGBW-Bayer CFA has been presented in this research. The algorithm has been tested on the Kodak dataset using the estimated value of white filters and a hyperspectral image dataset using the actual value of white filters, and the results have been compared. The results in both cases have been compared with the previous works on RGB-Bayer CFA, and it shows that the proposed algorithm using RGBW-Bayer CFA is working better than RGB-Bayer CFA in presence of noise.

No description available.

No description available.

No description available.

No description available.

No description available.