Reconhecimento de padrões em imagens é uma área de grande interesse no mundo científico. Os chamados métodos de extração de características, possuem as habilidades de extrair características das imagens e também de reduzir a dimensionalidade dos dados gerando assim o chamado vetor de características. Considerando uma imagem de consulta, o foco de um sistema de reconhecimento de imagens de faces humanas é pesquisar em um banco de imagens, a imagem mais similar à imagem de consulta, de acordo com um critério dado. Este trabalho de pesquisa foi direcionado para a geração de vetores de características para um sistema de reconhecimento de imagens, considerando bancos de imagens de faces humanas, para propiciar tal tipo de consulta. Um vetor de características é uma representação numérica de uma imagem ou parte dela, descrevendo seus detalhes mais representativos. O vetor de características é um vetor n-dimensional contendo esses valores. Essa nova representação da imagem propicia vantagens ao processo de reconhecimento de imagens, pela redução da dimensionalidade dos dados. Uma abordagem alternativa para caracterizar imagens para um sistema de reconhecimento de imagens de faces humanas é a transformação do domínio. A principal vantagem de uma transformação é a sua efetiva caracterização das propriedades locais da imagem. As wavelets diferenciam-se das tradicionais técnicas de Fourier pela forma de localizar a informação no plano tempo-freqüência; basicamente, têm a capacidade de mudar de uma resolução para outra, o que as fazem especialmente adequadas para análise, representando o sinal em diferentes bandas de freqüências, cada uma com resoluções distintas correspondentes a cada escala. As wavelets foram aplicadas com sucesso na compressão, melhoria, análise, classificação, caracterização e recuperação de imagens. Uma das áreas beneficiadas onde essas propriedades tem encontrado grande relevância é a área de visão computacional, através da representação e descrição de imagens. Este trabalho descreve uma abordagem para o reconhecimento de imagens de faces humanas com a extração de características baseado na decomposição multiresolução de wavelets utilizando os filtros de Haar, Daubechies, Biorthogonal, Reverse Biorthogonal, Symlet, e Coiflet. Foram testadas em conjunto as técnicas PCA (Principal Component Analysis) e IMPCA (Image Principal Component Analysis), sendo que os melhores resultados foram obtidos utilizando a wavelet Biorthogonal com a técnica IMPCA / Image pattern recognition is an interesting area in the scientific world. The features extraction method refers to the ability to extract features from images, reduce the dimensionality and generates the features vector. Given a query image, the goal of a features extraction system is to search the database and return the most similar to the query image according to a given criteria. Our research addresses the generation of features vectors of a recognition image system for human faces databases. A feature vector is a numeric representation of an image or part of it over its representative aspects. The feature vector is a n-dimensional vector organizing such values. This new image representation can be stored into a database and allow a fast image retrieval. An alternative for image characterization for a human face recognition system is the domain transform. The principal advantage of a transform is its effective characterization for their local image properties. In the past few years researches in applied mathematics and signal processing have developed practical wavelet methods for the multi scale representation and analysis of signals. These new tools differ from the traditional Fourier techniques by the way in which they localize the information in the time-frequency plane; in particular, they are capable of trading on type of resolution for the other, which makes them especially suitable for the analysis of non-stationary signals. The wavelet transform is a set basis function that represents signals in different frequency bands, each one with a resolution matching its scale. They have been successfully applied to image compression, enhancement, analysis, classification, characterization and retrieval. One privileged area of application where these properties have been found to be relevant is computer vision, especially human faces imaging. In this work we describe an approach to image recognition for human face databases focused on feature extraction based on multiresolution wavelets decomposition, taking advantage of Biorthogonal, Reverse Biorthogonal, Symlet, Coiflet, Daubechies and Haar. They were tried in joint the techniques together the PCA (Principal Component Analysis) and IMPCA (Image Principal Component Analysis)

Este trabalho investiga a possibilidade de reconstruir a variabilidade da precipitação pretérita na América do Sul a partir de uma série temporal de razão isotópica de oxigênio (δ18O) em um testemunho de gelo obtido na Bolívia, para isso é analisado as relações espaço-temporais entre a precipitação sobre a América do Sul e a série do δ18O. Utilizou-se dados do δ18O dos primeiros 50 metros do testemunho de gelo do Nevado Illimani (6.350 m, 16º37’S, 67º46’W) e totais mensais da precipitação coletada em 890 estações meteorológicas sobre o continente sul-americano no período 1979–2008. As amostras da precipitação foram dispostas em uma grade equiespaçada com resolução de ~2º de latitude e longitude. A variabilidade espacial e temporal da precipitação foram analisadas pela Análise das Componentes Principais no Modo T e S, respectivamente, com a identificação dos principais Padrões Espaciais Anômalos (PEA) e das zonas espaciais em que a variação temporal das anomalias da precipitação é correlacionada. No núcleo de cada zona foi selecionada uma série temporal anômala da precipitação para comparar com a série do δ18O. Constatou-se que há 12 PEA da precipitação e que são originários, predominantemente, da variação na temperatura superficial dos oceanos Pacífico Equatorial e Atlântico Tropical Norte e 8 zonas espaciais em que a variação da precipitação é parcialmente homogênea. Analisando características dos PEA da precipitação com a variação do δ18O constata-se que os registros de máximo (mínimo) fracionamento isotópico no verão estão relacionados à seca (chuvas acima da média) na Amazônia ou no nordeste brasileiro. Já no inverno, os registros de máximo (mínimo) fracionamento isotópico estão relacionados a secas (chuvas acima da média no litoral) na região tropical e chuvas acima da média (secas) na região extratropical. O índice de correlação entre todas séries de precipitação com a série do δ18O mostrou que a variação sincronizada das variáveis altera-se sazonalmente em função da migração espacial dos mecanismos de transporte e convergência da umidade. A comparação das séries de precipitação das 8 zonas com o δ18O possibilitou identificar possíveis áreas para a reconstrução paleoclimática. Nas regiões equatorial e subtropical foi testado a reconstrução. Nessas áreas, os fatores que controlam as duas variáveis são os mesmos, predominando os oceanos Pacífico Equatorial e Atlântico Norte nas alterações das séries temporais com frequência entre 24 e 60 meses. Para reconstrução da variabilidade da precipitação pretérita a partir da variação do δ18O foram testada duas técnicas, uma utilizando a análise de dependência por regressão e outra pela adoção de PEA para casos específicos da variação do δ18O. A primeira técnica não gerou resultados satisfatórios, por isso não foi usada. Com a segunda técnica foi possível identificar com 80% de chance de acerto as secas na Amazônia, no nordeste brasileiro e na região subtropical e com 68% de chance de acerto as chuvas acima da média na região subtropical, quando a avaliação não considera a magnitude das anomalias da precipitação. Por outro lado, a técnica é limitada, porque os PEAs adotados como referência não representam a grande variabilidade da precipitação nessas regiões. / This thesis investigates the possibility of reconstructing the South American past precipitation variability based on an oxygen isotope ratio (δ18O) time series from a Bolivian ice core, for this we analysed the spatial-temporal relations between these two variables. We used δ18O data from the upper 50 meters of Nevado Illimani ice core (6,350 m, 16°37'S, 67°46'W) and monthly total precipitation collected at 890 weather stations in South America in the period 1979–2008. Precipitation samples were arranged in an equally spaced grid with a ~2° latitude and longitude. The precipitation temporal and spatial variability were analysed by Principal Component Analysis on Mode S and T, respectively, with the identification of the main Anomalous Spatial Patterns (ASP) and spatial zones in which temporal rainfall anomalies variations are correlated. At each zone core, we selected one anomalous precipitation time series to compare with the δ18O series. There are 12 precipitations ASP that originate predominantly from surface temperature variations in the equatorial Pacific and tropical North Atlantic oceans, and 8 spatial zones in which the precipitation variation is partially homogeneous. Analysing the precipitation ASP characteristics with the δ18O variation, it is observed that records of maximum (minimum) isotopic fractionation in summer are related to droughts (above average precipitations) in the Amazon and northeastern Brazil. In the winter, records of maximum (minimum) isotopic fractionation are related to droughts (above average rainfall in the coastal area) in the tropical region and above average rainfall (droughts) in the extratropical region. The correlation index among all series of precipitation with δ18O series showed that these variables synchronized variation alters seasonally depending on the transport mechanisms spatial migration and on the moisture convergence. The comparison of the 8 precipitation zones with the δ18O series enabled us to identify possible areas for paleoclimatic reconstruction, which are the equatorial and subtropical regions. In these areas, the factors controlling the two variables are the same, prevailing the equatorial Pacific and the North Atlantic changes in time series with frequency from 24 to 60 months. To reconstruct the precipitation variability from the past δ18O variation we tested two techniques, one using dependency analysis for regression and other adopting ASP for specific cases of δ18O variations. The first technique yielded no satisfactory results, so it was disregarded. Using the second technique, it was possible to identify droughts in Amazonia, the Brazilian northeast and in the subtropical region with 80% changes of success. Above the average rainfalls in the subtropical region are identify with 68% changes of success, if this assessment does not consider the magnitude of the precipitation anomalies. On the other hand, the technique has limitations because the ASPs adopted as reference do not represent the wide precipitation variability in these regions.

When stimuli processing time varies in an oddball paradigm, the latency of the P300 will vary across trials. In an oddball task requiring difficult response selections, as the variation of stimuli processing time increases, so does the variation of the P300 latency, causing latency jitters in the measurement. Averaging the P300 across different trials without adjusting this latency jitter will lead to diminished P300 amplitude, resulting in inaccurate conclusions from the data. Verleger et al. (2014) reported a diminished P300 amplitude in a difficult oddball task that required subjects to make response selections among stimuli that are difficult to distinguish, but his work did not correct for any latency jitter observed within his sample. The current study replicated the easy and hard oddball tasks conducted in Verleger et al.. Raw ERPs obtained from 16 subjects indicated a successful replication of the study. An examination of the behavioral data showed that there was substantial variation in the P300 during the hard oddball tasks, and a latency jitter correction was applied in the analysis. Results indicated that there was a significant increase in the amplitude of P300 after latency jitter correction, and that this P300 amplitude did not differ significantly between easy and hard oddball tasks. These results suggest that difficult decision requirement does not reduce the amplitude of the P300, and that latency jitter should be accounted for when analyzing data from tasks involving a difficult decision requirement.

In 2010, approximately two-thirds of adults and one-fifth of the adolescent population in the United States were considered either overweight or obese, resulting in the United States having the highest per capita obesity rate among all OECD countries. A considerable body of literature regarding health behavior, health outcomes, and public policy exists on what the Centers for Disease Control and Prevention considers an obesity epidemic. In response to the growing problem of childhood obesity, the Child Nutrition and WIC Reauthorization Act of 2004 (CNRA), which required that schools participating in the National School Lunch Program and/or School Breakfast Program have wellness policies on file, was passed. The purpose of this research is to provide additional insight into the origin of the geographic variation in adolescent obesity rates between the U.S. states. Previous research has looked at differences in built environments, maternal employment, food prices, agriculture policies, and technology factors in an effort to explain the variation in adolescent obesity prevalence. This dissertation contributes to the literature by examining the hypothesis that state-level school wellness policies also played a role in determining the rates of childhood obesity. Using School Health Policies and Practices Study (SHPPS) surveys from 2000 – 2012, I derived a state-level school wellness policy measure. This, together with Youth Risk Behavior Surveillance survey data on adolescent BMI was used to measure the effect of the wellness policy mandate on adolescent obesity prevalence. Several models were applied to first demonstrate that the state of residence for an adolescent is indeed related to BMI trends and then to investigate various determinants of adolescent obesity including the primary variable of interest, state school wellness policies. The results of this research provide evidence of a statistically significant, although very small positive effect of school wellness policies on adolescent BMI that is contrary to my hypothesis. Dominance analysis showed that of the four wellness policy factors considered in the principal component composition of the wellness policy measure, policy components that met state requirements rather than those meeting health screen criteria, state recommendations, and national standards were most important in explaining the overall variance of the regression model. Interestingly, the public school attendance rate itself was also associated with a substantial decrease in adolescent BMI. Understanding the determinants of adolescent obesity and how to effect change in the rising trend is a national concern. Obese adolescents are at significant risk of becoming obese adults and previous research has already shown the high economic costs associated with adult obesity and its comorbidities. Policies implemented in school, where adolescents consume a considerable portion of their daily calories and participate in physical activity, can help to build healthy habits that have the potential to lower the probability of an adolescent becoming an obese adult. Over time, a healthier adult population may result in lower economic costs associated with medical care and lost productivity.

The aim of this study was to propose a nonlinear multiscale principal component analysis (NLMSPCA) methodology for process monitoring and fault detection based upon multilevel wavelet decomposition and nonlinear principal component analysis via an input-training neural network. Prior to assessing the capabilities of the monitoring scheme on a nonlinear industrial process, the data is first pre-processed to remove heavy noise and significant spikes through wavelet thresholding. The thresholded wavelet coefficients are used to reconstruct the thresholded details and approximations. The significant details and approximations are used as the inputs for the linear and nonlinear PCA algorithms in order to construct detail and approximation conformance models. At the same time non-thresholded details and approximations are reconstructed and combined which are used in a similar way as that of the thresholded details and approximations to construct a combined conformance model to take account of noise and outliers. Performance monitoring charts with non-parametric control limits are then applied to identify the occurrence of non-conforming operation prior to interrogating differential contribution plots to help identify the potential source of the fault. A novel summary display is used to present the information contained in bivariate graphs in order to facilitate global visualization. Positive results were achieved. / Dissertation (M Eng (Control Engineering))--University of Pretoria, 2007. / Chemical Engineering / unrestricted

Vector-borne diseases, such as those transmitted by mosquitoes, pose a significant public health concern in many countries worldwide. In this thesis, I explored the role of a number of risk factors defined at multiple scales on vector-borne disease prevalence, focusing on malaria and arboviral infections in several regions of North-Eastern Tanzania, with the principal aim of improving the overall diagnosis of febrile illness in this region. First, I investigated the influence of household-wealth on prevalence of malaria and arboviral infections using principal component analysis (PCA), and then described the methodological challenges associated with this statistical technique when used to compute indices from smaller datasets. I then employed a multilevel modelling approach to simultaneously incorporate household-level anthropogenic factors and village-level environmental characteristics to investigate key determinants of Anopheles malaria vector density among rural households. These analyses provided methodologically rigorous approaches to studying vector-borne diseases at a very fine-scale and also have significant public health relevance as the research findings can assist in guiding policy decisions regarding surveillance efforts as well as inform where and when to prioritize interventions.

Recent inquiry into the nomenclature of several species within Spinocyrtia has led to questions concerning name applicability and validity, particularly whether Delthyris granulosa and Spinocyrtia (Spirifer) granulosa are synonymous. This study utilized two-dimensional outline landmark analysis, a form of geometric morphometric analysis, to evaluate interspecific variation among these species. I took over a thousand photographs of over a hundred specimens of brachiopods belonging to the family Spinocyrtiidae. Ninety-six specimens originated from the Givetian outcrop belt of New York state, three were from northwestern Ohio, there was single Canadian specimen, and there was a single German specimen. The results from these analyses indicate that the mophospaces of Spinocyrtia (Spirifer) congesta, S. (Spirifer) granulosa, and S?. (Spirifer) marcyi are statistically (p < 0.05) distinct from one another.

There are numerous approaches to quantitatively measure well-being. Most well-beingresearch are based on income or health situation from economics perspective. The needfor research on women’s relationship between education and well-being is an area thathas not been fully investigated. It is also important to know how the situation ofwomen’s well-being compare with men’s. The purpose of this research is to estimatewomen’s well-being and understand how well-being women is compared with men inChina. Different characteristics of men and women is considered and estimate thespecific relationship between education and well-being. Two measure of well-being areused: self-assessed unidimensional subjective well-being and parametrically estimatedmultidimensional well-being. Two measurement will help to understand the differencebetween subjectivity and objectivity of well-being. To achieve this goal, this researchcomputes and compares the well-being of 34,054 women and men by using ChineseGeneral Social Survey in 2012, 2013 and 2015. Well-being is measured by computingmultidimensionally by principal component analysis which depend on differentdomains of identity, capability, material well-being. All the domains contribute toindividual’s well-being. The findings suggest that, multidimensional well-being indexdiffer from the subjective well-being in ranking individual grouped by importantcommon characterizes. The difference is attributed to multidimensionality of the well-being index. Under this circumstance, education still does influence well-beingpositively conditional on controlling for identity, capability and material well-being.

Gait initiation (GI) is an important locomotor transition task that includes anticipatory postural adjustments and the joint propulsion necessary for the first step of walking. Metrics associated with this task are known to change across the lifespan and may provide valuable information for fall risk indication, as falls often occur during transitional tasks. Assessments of discrete variables between fallers and non-fallers at GI have provided insight into differences between groups. However, more complex approaches such as time-series principal component analysis (PCA) may allow the examination of changes in magnitude, pattern, and timing not detectable using discrete comparisons alone. Therefore, this thesis aims to characterize differences between fallers and non-fallers by examining the kinematics and kinetics of gait initiation using time-series PCA. A sample of 56 community-dwelling older adults was recruited for this study and completed five walking trials where GI was measured by two force platforms. PCA of centre of pressure kinematics and kinetics time-series data were used to identify the critical features of the signal, and multivariate analysis of covariance was used to compare the individual loading scores of each principal component for each phase between groups. It was revealed that fallers demonstrated differences in the range of mediolateral movement during weight transfer and forward progression, a greater range of anteroposterior movement in forward progression, and a more gradual rise in vertical forces in the first step, associated with a shorter first step length. These findings point to a tendency for fallers to prioritize stability over forward progression performance, and differences in postural control strategies, compared to non-fallers. Further, the use of time-series PCA helped to highlight differences not detectable using discrete analysis alone. / Graduate

A principal component least mean square (PC-LMS) adaptive algorithm is described that has considerable benefits for large control systems used to implement feedforward control of single frequency disturbances. The algorithm is a transform domain version of the multichannel filtered-x LMS algorithm. The transformation corresponds to the principal components of the transfer function matrix between the sensors and actuators in a control system at a single frequency. The method is similar to other transform domain LMS algorithms because the transformation can be used to accelerate convergence when the control system is ill-conditioned. This ill-conditioning is due to actuator and sensor placement on a continuous structure. The principal component transformation rotates the control filter coefficient axes to a more convenient coordinate system where (1) independent convergence factors can be used on each coordinate to accelerate convergence, (2) insignificant control coordinates can be eliminated from the controller, and (3) coordinates that require excessive control effort can be eliminated from the controller. The resulting transform domain algorithm has lower computational requirements than the filtered-x LMS algorithm. The formulation of the algorithm given here applies only to single frequency control problems, and computation of the decoupling transforms requires an estimate of the transfer function matrix between control actuators and error sensors at the frequency of interest. The feasibility of the method was demonstrated in real-time noise control experiments involving 48 microphones and 12 control actuators mounted on a closed cylindrical shell. Convergence of the PC-LMS algorithm was more stable than the filtered-x LMS algorithm. In addition, the PC-LMS controller produced more noise reduction with less control effort than the filtered-x LMS controller in several tests. / Ph. D.