Melhorando a estimação de pose com o RANSAC preemptivo generalizado e múltiplos geradores de hipóteses

The camera motion estimation represents one of the fundamental problems in Computer Vision and it may be solved by several methods. Preemptive RANSAC is one of them, which in spite of its robustness and speed possesses a lack of flexibility related to the requirements of applications and hardware platforms using it. In this work, we propose an improvement to the structure of Preemptive RANSAC in order to overcome such limitations and make it feasible to execute on devices with heterogeneous resources (specially low budget systems) under tighter time and accuracy constraints. We derived a function called BRUMA from Preemptive RANSAC, which is able to generalize several preemption schemes, allowing previously fixed parameters (block size and elimination factor) to be changed according the applications constraints. We also propose the Generalized Preemptive RANSAC method, which allows to determine the maximum number of hipotheses an algorithm may generate. The experiments performed show the superiority of our method in the expected scenarios. Moreover, additional experiments show that the multimethod hypotheses generation achieved more robust results related to the variability in the set of evaluated motion directions

Implementação Hardware/Software da estimação de movimento segundo o padrão H.264

Motion estimation is the main responsible for data reduction in digital video encoding. It is also the most computational damanding step. H.264 is the newest standard for video compression and was planned to double the compression ratio achievied by previous standards. It was developed by the ITU-T Video Coding Experts Group (VCEG) together with the ISO/IEC Moving Picture Experts Group (MPEG) as the product of a partnership effort known as the Joint Video Team (JVT). H.264 presents novelties that improve the motion estimation efficiency, such as the adoption of variable block-size, quarter pixel precision and multiple reference frames. This work defines an architecture for motion estimation in hardware/software, using a full search algorithm, variable block-size and mode decision. This work consider the use of reconfigurable devices, soft-processors and development tools for embedded systems such as Quartus II, SOPC Builder, Nios II and ModelSim

Arquitetura com elevada taxa de processamento e reduzida largura de banda de memória para a estimação de movimento em vídeos digitais

Nowadays several electronics devices support digital videos. Some examples of these devices are cellphones, digital cameras, video cameras and digital televisions. However, raw videos present a huge amount of data, millions of bits, for their representation as the way they were captured. To store them in its primary form it would be necessary a huge amount of disk space and a huge bandwidth to allow the transmission of these data. The video compression becomes essential to make possible information storage and transmission. Motion Estimation is a technique used in the video coder that explores the temporal redundancy present in video sequences to reduce the amount of data necessary to represent the information. This work presents a hardware architecture of a motion estimation module for high resolution videos according to H.264/AVC standard. The H.264/AVC is the most advanced video coder standard, with several new features which allow it to achieve high compression rates. The architecture presented in this work was developed to provide a high data reuse. The data reuse schema adopted reduces the bandwidth required to execute motion estimation. The motion estimation is the task responsible for the largest share of the gains obtained with the H.264/AVC standard so this module is essential for final video coder performance. This work is included in Rede H.264 project which aims to develop Brazilian technology for Brazilian System of Digital Television

Sistema não-Intrusivo para Estimação da Direção do olhar utilizando redes neurais artificiais

The fundamental senses of the human body are: vision, hearing, touch, taste and smell. These senses are the functions that provide our relationship with the environment. The vision serves as a sensory receptor responsible for obtaining information from the outside world that will be sent to the brain. The gaze reflects its attention, intention and interest. Therefore, the estimation of gaze direction, using computer tools, provides a promising alternative to improve the capacity of human-computer interaction, mainly with respect to those people who suffer from motor deficiencies. Thus, the objective of this work is to present a non-intrusive system that basically uses a personal computer and a low cost webcam, combined with the use of digital image processing techniques, Wavelets transforms and pattern recognition, such as artificial neural network models, resulting in a complete system that performs since the image acquisition (including face detection and eye tracking) to the estimation of gaze direction. The obtained results show the feasibility of the proposed system, as well as several feature advantages.

Métodos estatísticos recursivos aplicados ao problema de estimação de vazão de gás em plantas de Plunger-lift

This work has as main objective to find mathematical models based on linear parametric estimation techniques applied to the problem of calculating the grow of gas in oil wells. In particular we focus on achieving grow models applied to the case of wells that produce by plunger-lift technique on oil rigs, in which case, there are high peaks in the grow values that hinder their direct measurement by instruments. For this, we have developed estimators based on recursive least squares and make an analysis of statistical measures such as autocorrelation, cross-correlation, variogram and the cumulative periodogram, which are calculated recursively as data are obtained in real time from the plant in operation; the values obtained for these measures tell us how accurate the used model is and how it can be changed to better fit the measured values. The models have been tested in a pilot plant which emulates the process gas production in oil wells

Estimação paramétrica e não-paramétrica em modelos de markov ocultos

In this work we study the Hidden Markov Models with finite as well as general state space. In the finite case, the forward and backward algorithms are considered and the probability of a given observed sequence is computed. Next, we use the EM algorithm to estimate the model parameters. In the general case, the kernel estimators are used and to built a sequence of estimators that converge in L1-norm to the density function of the observable process

Estimação de parâmetros genéticos para produção de leite de vacas da raça Holandesa via regressão aleatória

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Aplicação de redes neurais na estimação da temperatura interna de transformadores de distribuição imersos em óleo

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Uma abordagem neural para estimação de conjugado em motores de indução

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Inferência bayesiana aplicada à estimação de herdabilidades dos parâmetros da curva de crescimento de fêmeas da raça Nelore

Objetivou-se estimar parâmetros genéticos, utilizando inferência Bayesiana, para as estimativas dos parâmetros individuais de peso à maturidade (Â) e taxa de crescimento, obtidos pela função de crescimento Brody. O arquivo estava constituído de 14.563 registros de pesos e idades referentes a 1.158 fêmeas da raça Nelore, participantes do Programa de Melhoramento Genético da Raça Nelore. Para a análise das estimativas dos parâmetros da curva, via inferência bayesiana, foi proposto um modelo animal unicaráter, que incluiu como fixo o efeito de grupo contemporâneo (animais nascidos no mesmo estado, no mesmo trimestre do ano, mesmo ano e mesmo regime alimentar) e como aleatórios os efeitos genético direto e residual. Nessa análise, foram utilizados dois diferentes tamanhos para as cadeias geradas pelo algoritmo de amostragem de Gibbs, de 550 e 1.100 mil ciclos, com períodos de descarte amostral de 50 e 100 mil ciclos, respectivamente, e amostragens a cada 500 e 1.000 ciclos, respectivamente. As médias posteriores da variância genética aditiva e residual foram próximas, tanto para  quanto para a, mesmo quando implementados diferentes tamanhos para as cadeias geradas pelo algoritmo de amostragem de Gibbs. Os coeficientes de herdabilidade estimados para Â, variaram de 0,44 a 0,46, amplitude semelhante aos 0,46 a 0,48 obtidos para as estimativas de. Essas magnitudes indicam que a seleção pode ser usada como instrumento para alterar a forma da curva de crescimento desses animais. Entretanto, o uso das informações obtidas, visando à alteração da curva de crescimento dos animais, deve ser feito com grande cautela, uma vez que as características a serem trabalhadas na modificação do formato da curva de crescimento, de acordo com resultados da literatura especializada, são negativamente correlacionadas.

Estimação de parâmetros genéticos para a produção de leite no dia do controle e em 305 dias para primeiras lactações de vacas da raça Holandesa

Dados de 39.578 controles leiteiros de 3.766 primeiras lactações de vacas da raça Holandesa, ocorridas de 1994 a 2002, foram analisados com os objetivos de estimar parâmetros genéticos para as produções de leite no dia do controle (PLDC) e para a produção até 305 dias de lactação (P305) e comparar estes dois critérios de seleção. Os componentes de variância foram estimados pelo método da máxima verossimilhança restrita, sob modelo animal univariado ou bivariado. Para as PLDC, os modelos incluíram o efeito aleatório genético aditivo, o efeito fixo de grupo contemporâneo e, como covariáveis, a idade da vaca ao parto (efeitos linear e quadrático) e os dias em lactação (efeito linear). Para a P305, foi utilizado o mesmo modelo, substituindo dias em lactação por duração da lactação. Os grupos de contemporâneos foram formados por ano, mês do controle e rebanho (para as PLDC) e por ano, época do parto e rebanho (para a P305). As herdabilidades estimadas para a P305 foram de 0,27 e 0,25 para as análises univariadas e bivariadas, respectivamente. Para as PLDC, as herdabilidades variaram de 0,11 a 0,31. Para o modelo bivariado (pelo qual avaliaram-se simultaneamente P305 e as PLDC), as herdabilidades para os controles (PLDC) foram menores, variando de 0,08 a 0,25. As maiores estimativas ocorreram para as produções do 4º e 5º controles, correspondendo aos 2º e 3º meses de lactação. As correlações genéticas entre P305 e os controles individuais foram positivas e elevadas, variando de 0,83 a 1,00. Os resultados indicaram que a seleção direta para P305, como tradicionalmente realizada, implicaria maiores ganhos genéticos para a produção de leite (PL) na maioria dos controles quinzenais. Além disso, a seleção direta para as produções parciais poderia proporcionar ganhos correlacionados também para a P305, mas estes ganhos seriam menores que os obtidos via seleção direta.

Relação do perímetro escrotal com a média de peso do grupo contemporâneo para estimação de um modelo de ajuste

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Estudo de modelo alternativo para estimação de componentes de (co)variância e predição de valores genéticos de características de crescimento em bovinos da raça Nelore

The objectives of this study were to estimate genetic parameters for non-standardized weights at nursing (PR120), at weaning (PR240), at yearling (PR365) and at post yearling (PR550), and to predict EPD's (expected progeny differences) for these traits using records from 29,769 Nellores. Covariance components and genetic parameters were estimated by mixed-model methodology, REML, using an animal model. Models for PR120, PR240, PR365 and PR455 included the random direct and maternal animal effects, the dam permanent environmental effect and the error. Fixed effects were contemporary group (CG) and age of cow at parturition (CIVP) and the covariate age of the calf at measuring. Two additional models for PR365, PR455 and PR550 analyses were used: the first included CG and CIVP, animal and maternal direct effect, residual and age of the calf (as covariate), and the second included CG and CIVP (as fixed effects), animal direct effect, residual and age of calf at measuring. Observed means±standard deviations were: 127±25kg (PR120); 191±34kg (PR240); 225±42kg (PR365); 266±51kg (PR455) and 310±56kg (PR550). From single-trait analyses, direct and maternal heritabilities for PR120, PR240, PR365 and PR455 were, respectively, .23 and .08; .19 and .10; .24 and .04; .30 and .04. Direct heritabilities were .39; .44 and .43, respectively, for PR365, PR455 and PR550. In the model without permanent effect, direct and maternal heritabilities for PR365, PR455 and PR550 were .25 and .08; .32 and .07; .38 and .03, respectively. When the estimates for standardized traits at the same period were compared, no differences in magnitude were found. Rank correlation had important changes when standardized and non-standardized traits were compared.

Estimação de frequência usando sensores de erro com atrasos adaptativos

This paper proposes a solution to improve the performance of the first order Early Error Sensing (EES) Adaptive Time Delay Tanlock Loops (ATDTL) presented in (Al-Zaabi, Al-Qutayri e Al-Araji, 2005), regarding to frequency estimation and tracking time. The EES-ATDTL are phaselocked-loops (PLL) used to hardware implementations, due to their simple structure. Fixed-points theorems are used to determine conditions for rapid convergence of the estimation process and a estimative of the frecuency input is obtained with a Gaussian filter that improves the gain adaptation. The mathematical models are the presented by (Al-Araji, Al-Qutayri e Al-Zaabi, 2006). Simulations have been performed to evaluate the theoretical results.

Fluxo de potência trifásico probabilístico para redes de distribuição usando o método de estimação por pontos

In this paper the point estimation method is applied to solve the probabilistic power flow problem for unbalanced three-phase distribution systems. Through the implementation of this method the probability distribution functions of voltages (magnitude and angle) as well as the active and reactive power flows in the branches of the distribution system are determined. Two different approaches of the point estimation method are presented (2m and 2m+1 point schemes). In order to test the proposed methodology, the IEEE 34 and 123 bus test systems are used. The results obtained with both schemes are compared with the ones obtained by a Monte Carlo Simulation (MCS).