abstract: Light field imaging is limited in its computational processing demands of high sampling for both spatial and angular dimensions. Single-shot light field cameras sacrifice spatial resolution to sample angular viewpoints, typically by multiplexing incoming rays onto a 2D sensor array. While this resolution can be recovered using compressive sensing, these iterative solutions are slow in processing a light field. We present a deep learning approach using a new, two branch network architecture, consisting jointly of an autoencoder and a 4D CNN, to recover a high resolution 4D light field from a single coded 2D image. This network decreases reconstruction time significantly while achieving average PSNR values of 26-32 dB on a variety of light fields. In particular, reconstruction time is decreased from 35 minutes to 6.7 minutes as compared to the dictionary method for equivalent visual quality. These reconstructions are performed at small sampling/compression ratios as low as 8%, allowing for cheaper coded light field cameras. We test our network reconstructions on synthetic light fields, simulated coded measurements of real light fields captured from a Lytro Illum camera, and real coded images from a custom CMOS diffractive light field camera. The combination of compressive light field capture with deep learning allows the potential for real-time light field video acquisition systems in the future. / Dissertation/Thesis / Masters Thesis Computer Engineering 2017

abstract: Compressive sensing theory allows to sense and reconstruct signals/images with lower sampling rate than Nyquist rate. Applications in resource constrained environment stand to benefit from this theory, opening up many possibilities for new applications at the same time. The traditional inference pipeline for computer vision sequence reconstructing the image from compressive measurements. However,the reconstruction process is a computationally expensive step that also provides poor results at high compression rate. There have been several successful attempts to perform inference tasks directly on compressive measurements such as activity recognition. In this thesis, I am interested to tackle a more challenging vision problem - Visual question answering (VQA) without reconstructing the compressive images. I investigate the feasibility of this problem with a series of experiments, and I evaluate proposed methods on a VQA dataset and discuss promising results and direction for future work. / Dissertation/Thesis / Masters Thesis Computer Engineering 2017

Orientador: Jefferson Sidney Camacho / Co-orientador: Guilherme Aris Parsekian / Banca: Mauro Mitsuuchi Tashima / Banca: Gilberto Carbonari / Resumo: O concreto celular, que se caracteriza por ter um peso próprio reduzido pela incorporação artificial de ar no concreto, pode constituir uma alternativa ao concreto armado nas estruturas de pequeno e médio porte. A baixa solicitação dos elementos estruturais permite aproveitar as principais vantagens do material: melhoria da isolação, economia de material para produzir o concreto e facilidade do transporte dos elementos pré-moldados. A fim de conhecer melhor o desempenho mecânico do material para aplicações estruturais, o presente trabalho trata da avaliação da aderência entre o concreto celular e as barras de aço. Para estudar esse fenômeno, foram realizados ensaios de arrancamento com barras de diâmetro igual a 4,2 mm, parcialmente imersas em corpos-de-prova de concreto cilíndrico. A variação da massa específica do concreto e a presença ou não de uma barra transversal dentro do cilindro constituem as variáveis da pesquisa. No total foram realizados três tipos de concreto, com massa específica aparente aos 28 dias de 2255, 1565 e 1510 kg/m 3 . Entre outros resultados, a pesquisa revelou, pelo arrancamento de barras simples de diâmetro igual a 4,2 mm, o forte decréscimo da tensão de aderência junto com o rebaixamento da massa específica aparente do concreto. Confrontando os resultados com trabalhos semelhantes, a elevada relação água/cimento parece ser o parâmetro que mais influenciou os resultados. Também, parâmetros que podem ter influenciado a tensão de aderência são a quantidade de agente espumígeno colocada nos concretos, a utilização de agregado graúdo e talvez o diâmetro das barras arrancadas. A introdução da barra transversal no meio do cilindro permitiu obter um ganho de resistência ao arrancamento da barra longitudinal, que cresceu junto com o rebaixamento da massa específica do concreto celular. Assim, a utilização de ancoragens especiais (não exclusivamente retas), que permite... / Abstract: Cellular concrete is a lightweight concrete obtained by aerating agent, which produces air-voids into the mixture. Low loads acting on houses or small buildings permits to use cellular concrete advantages: good thermal insulation, lower concrete production costs and simplification of precast transportation. To improve the mechanical characteristic of structural cellular concrete knowledge, this work deals with bond between cellular concrete and steel rebars. Pull-out tests of 4,2 mm diameter rebars partly immersed into concrete cylinders were made. Concrete unit weight variation and setting of a transverse rebar into the concrete cylinder center constitute the variables of the work. Three types of mixture were prepared: one without aerating agent (with a dry unit weight of 2255 kg/m 3 ), and two with different aerating agent rates (with a dry unit weight equal to 1565 and 1510 kg/m 3 ). One of the conclusions of the study is the large decrease of bond stress between 4,2 diameter rebars and cellular concrete, due to concrete unit weight reduction. Comparison of results with others studies reveals that water/cement ratio could be the main variable which influences bond stress. Moreover, other variables like aerating agent quantity introduced into the mixture, coarse aggregate use or rebar diameter could influence the bond stress value. Transverse rebar introduced into concrete cylinder center increased the ultimate pull-out force of the test: the gain of bond produced by the transverse rebar grew up when the cellular concrete unit weight had been reduced. Therefore, special anchorages (not exclusively straight), permitting bond improvement between cellular concrete and 4,2 mm diameter rebar could be a solution to improve the mechanical performance of cellular concrete / Mestre

Este estudo objetivou avaliar a resistência máxima às forças de compressão entre dois grupos diferentes de coroas em porcelanas livres de metal cimentadas sobre dentes naturais e um grupo formado por dentes caninos naturais hígidos e comparar os resultados obtidos com os de (Chaves, 2001), em seu trabalho de Doutorado onde também se avaliou a resistência máxima às forças de compressão entre coroa metalo cerâmica fraturada e reparada com resina composta com coroa metalo cerâmica integra e dente natural permanente. No presente estudo foram feitos vinte preparos para coroa total em dentes naturais caninos e divididos em dois grupos para confecção de dez coroas do sistema cerâmico aluminizado infiltrado por vidro (Angelus Brasil) e dez coroas em porcelana feldspática Noritake ® (Noritake Kisai CO. Cada grupo de coroas foi cimentado com cimento resinoso auto adesivo RelyX Unicem ® , (3M ESPE, Brasil) com o objetivo de aumentar a resistência da porcelana frente às forças de compressão. O teste de compressão foi realizado em uma máquina de ensaio universal KRATOS, através de uma carga aplicada axialmente no ápice da borda incisal de cada espécime até a fratura. Os dados foram submetidos à análise de Variância ANOVA (p < 0,05) e teste de Tukey para comparação entre grupos. Os resultados mostraram que não houve diferença estatisticamente significante entre a coroa de porcelana aluminizada infiltrada por vidro e o dente natural. A coroa de porcelana feldspática apresentou-se menos resistente com relação ao dente natural e coroa de porcelana aluminizada infiltrada por vidro. / The objective of this study is evaluate the maximum load resistance between two different groups of porcelain metal free crown seated on natural teeth and a group of natural canine and to compare the results gotten with the ones of (Chaves, 2001), in its work of Doutorado where also the ultimate strength to the forces of compression between metalo ceramic broken and repaired with composite resin with ceramic crown metaloceramic integrates and permanent natural tooth. In the present study twenty natural canine teeth were prepared for crown canines and divided in two groups: aluminized porcelain crown infiltrated for glass (Angelus Brazil) and feldspathic Noritake Kisai Noritake (co). Crowns were seated using self adhesive resin cement RelyX Unicem (3M ESPE, Brazil) Compression testing was performed by a universal testing machine (KRATOS) by a load applied axially in incisal of each specimen until occurs the fracture. The data were analyzed by one way analyses of variance (ANOVA) and Tukey test for comparison between groups. The results showed that it did not have statistical significant difference between aluminized porcelain crown infiltrated by glass and natural tooth. The feldspathic porcelain crown presented less resistant than natural tooth and aluminized porcelain crown infiltrated by glass. .

Next generation wireless systems will support higher data rates, improved spectral efficiency, and less latency. Massive multiple-input multiple-output (MIMO) is proposed to satisfy these demands. In massive MIMO, many benefits come from employing hundreds of antennas at the base station (BS) and serving dozens of user terminals (UTs) per cell. As the number of antennas increases at the BS, the channel becomes sparse. By exploiting sparse channel in massive MIMO, compressive sensing (CS) methods can be implemented to estimate the channel. In CS methods, the length of pilot sequences can be shortened compared to pilot-based methods. In this paper, a novel channel estimation algorithm based on a CS method called elastic net is proposed. Channel estimation accuracy of pilot-based, lasso, and elastic-net based methods in massive MIMO are compared. It is shown that the elastic-net based method gives the best performance in terms of error for the less pilot symbols and SNR values.

The variation in the compressive strength of concrete block masonry was studied at elevated temperatures. Small specimens known as wallettes were used to obtain the compressive strength under steady state conditions. Eighteen wallettes were made of lightweight concrete blocks and 1:1:5 mortar proportion. The target temperatures were 20°C, 200°C, 400°C, 600°C, 700°C and 800°C. Initially load-deflection relationships were determined from the experimental wallettes and later they were converted into stress-strain relationships. Although the goal was to determine the compressive strength, other parameters were also studied such as modulus of elasticity, temperature-time relationships, modes of failure, material degradation, and change of colour.Lightweight concrete blocks were also tested to determine the compressive strength at equal temperatures applied for the wallettes. The blocks belonged to the same batch used for the wallettes. The tensile behaviour of mortar was determined at 20°C, 200°C and 400°C.Once the mechanical properties of the masonry wallettes, units and mortar were determined, they were used as input data to develop finite element models to simulate the same behaviour of the experimental wallettes. Finally, using the experimental and numerical results from the wallettes, they were used to predict the behaviour of 3m height walls.

The literature on compressive parameter estimation has been mostly focused on the use of sparsity dictionaries that encode a discretized sampling of the parameter space; these dictionaries, however, suffer from coherence issues that must be controlled for successful estimation. To bypass such issues with discretization, we propose the use of statistical parameter estimation methods within the Approximate Message Passing (AMP) algorithm for signal recovery. Our method leverages the recently proposed use of custom denoisers in place of the usual thresholding steps (which act as denoisers for sparse signals) in AMP. We introduce the design of analog denoisers that are based on statistical parameter estimation algorithms, and we focus on two commonly used examples: frequency estimation and bearing estimation, coupled with the Root MUSIC estimation algorithm. We first analyze the performance of the proposed analog denoiser for signal recovery, and then link the performance in signal estimation to that of parameter estimation. Numerical experiments show significant improvements in estimation performance versus previously proposed approaches for compressive parameter estimation.

Seismic exploration utilizes controlled sources, which emit seismic waves that propagate through the earth subsurface and get reflected off subsurface interfaces and scatterers. The reflected and scattered waves are recorded by recording stations installed along the earth surface or down boreholes. Seismic imaging is a powerful tool to map these reflected and scattered energy back to their subsurface scattering or reflection points. Seismic imaging is conventionally based on the single-scattering assumption, where only energy that bounces once off a subsurface scatterer and recorded by a receiver is projected back to its subsurface position. The internally multiply scattered seismic energy is considered as unwanted noise and is usually suppressed or removed from the recorded data. Conventional seismic imaging techniques yield subsurface images that suffer from low spatial resolution, migration artifacts, and acquisition fingerprint due to the limited acquisition aperture, number of sources and receivers, and bandwidth of the source wavelet. Hydrocarbon traps are becoming more challenging and considerable reserves are trapped in stratigraphic and pinch-out traps, which require highly resolved seismic images to delineate them. This thesis focuses on developing and implementing new advanced cost-effective seismic imaging techniques aiming at enhancing the resolution of the migrated images by exploiting the sparseness of the subsurface reflectivity distribution and utilizing the multiples that are usually neglected when imaging seismic data. I first formulate the seismic imaging problem as a Basis pursuit denoise problem, which I solve using an L1-minimization algorithm to obtain the sparsest migrated image corresponding to the recorded data. Imaging multiples may illuminate subsurface zones, which are not easily illuminated by conventional seismic imaging using primary reflections only. I then develop an L2-norm (i.e. least-squares) inversion technique to image internally multiply scattered seismic waves to obtain highly resolved images delineating vertical faults that are otherwise not easily imaged by primaries. Seismic interferometry is conventionally based on the cross-correlation and convolution of seismic traces to transform seismic data from one acquisition geometry to another. The conventional interferometric transformation yields virtual data that suffers from low temporal resolution, wavelet distortion, and correlation/convolution artifacts. I therefore incorporate a least-squares datuming technique to interferometrically transform vertical-seismic-profile surface-related multiples to surface-seismic-profile primaries. This yields redatumed data with high temporal resolution and less artifacts, which are subsequently imaged to obtain highly resolved subsurface images. Tests on synthetic examples demonstrate the efficiency of the proposed techniques, yielding highly resolved migrated sections compared with images obtained by imaging conventionally redatumed data. I further advance the recently developed cost-effective Generalized Interferometric Multiple Imaging procedure, which aims to not only image first but also higher-order multiples as well. I formulate this procedure as a linearized inversion framework and solve it as a least-squares problem. Tests of the least-squares Generalized Interferometric Multiple imaging framework on synthetic datasets and demonstrate that it could provide highly resolved migrated images and delineate vertical fault planes compared with the standard procedure. The results support the assertion that this linearized inversion framework can illuminate subsurface zones that are mainly illuminated by internally scattered energy.

Ultrafast plane-wave ultrasound imaging offers very high frame rates (exceeding thousands of frames per second) but entails large volumes of backscattered data collected by a sensor array over multiple plane-wave emissions at different angles. We propose a simple method for reducing the total amount of sampled data. First, we acquire the zero-angle data in full, and then we perform deterministic subsampling of the remaining nonzero-angle data. Our subsampling patterns are angle-specific and derived based on the zero-angle data using a Fourier-domain migration technique. We use two experimental datasets to evaluate the benefits and drawbacks of our proposed method in terms of spatial resolution and contrast-to-noise ratio, observed in the resulting B-mode images. / Graduate

This work investigates the optimal measurement matrices that can be used in compressive sensing (CS) image sensors. It also optimizes CMOS current-model pixel cell circuits for CS image sensors. Based on the outcomes from these optimization studies, three CS image senor circuits with compression ratios of 4, 6, and 8 are designed with using a 130 nm CMOS technology. The pixel arrays used in the image sensors has a size of 256X256. Circuit simulations with benchmark image Lenna show that the three images sensors can achieve peak signal to noise ratio (PSNR) values of 37.64, 33.29, and 32.44 dB respectively.