Laser vibration sensing can be used to classify military targets by its unique vibration signature. A coherent laser radar receives the target´s rapidly oscillating surface vibrations and by using proper demodulation and Doppler technique, stationary, radially moving and even accelerating targets can be taken care of. A frequency demodulation method developed at the former FOA, is for the first time validated against real data with turbulence, scattering, rain etc. The issue is to find a robust and reliable system for target recognition and its performance is therefore compared with some frequency distribution methods. The time frequency distributions have got a crucial drawback, they are affected by interference between the frequency and amplitude modulated multicomponent signals. The system requirements are believed to be fulfilled by combining the FOA method with the new statistical method proposed here, the combination being suggested as aimpoint for future investigations.

Music recordings contain the mixed contribution of multiple overlapping instruments. In order to better understand the music, it would be beneficial to understand each instrument independently. This thesis focuses on separating the individual instrument recordings within a song. In particular, we propose novel algorithms for separating instrument recordings given only their mixture. When the number of source signals does not exceed the number of mixture signals, we focus on a subclass of source separation algorithms based on joint diagonalization. Each approach leverages a different form of source structure. We introduce repetitive structure as an alternative that leverages unique repetition patterns in music and compare its performance against the other techniques. When the number of source signals exceeds the number of mixtures (i.e. the underdetermined problem), we focus on spectrogram factorization techniques for source separation. We extend single-channel techniques to utilize the additional spatial information in multichannel recordings, and use phase information to improve the estimation of the underlying components.

In this thesis work, time-frequency filtering of nonstationary signals in noise using Wigner-Ville Distribution is investigated. Continuous-time, discrete-time and discrete Wigner Ville Distribution definitions, their relations, and properties are given. Time-Frequency Peak Filtering Method is presented. The effects of different parameters on the performance of the method are investigated, and the results are presented. Time-Varying Wiener Filter is presented. Using simulations it is shown that the performance of the filter is good at SNR levels down to -5 dB. It is proposed and shown that the performance of the filter improves by using Support Vector Machines. The presented time-frequency filtering techniques are applied on test signals and on a real world signal. The results obtained by the two methods and also by classical zero-phase low-pass filtering are compared. It is observed that for low sampling rates Time-Varying Wiener Filter, and for high sampling rates Time-Frequency Peak Filter performs better.

It is commonly accepted that the induced noise on DFT phase spectrum has a negligible effect on speech intelligibility for short durations of analysis windows, as the early intelligibility studies pointed out. This fact is confirmed by recent intelligibility studies as well. Based on this phenomenon, classical speech enhancement algorithms do not modify DFT phase spectrum and only make changes in the DFT magnitude spectrum. However, in recent studies it is also indicated that these classical speech enhancement algorithms are not capable of improving the intelligibility scores of noise degraded speech signals. In other words, the contained information in a noise degraded signal cannot be increased by classical enhancement methods. Instead the ease of listening, i.e. quality, can be improved. Hence additional effort can be made to increase the amount of quality improvement using both DFT magnitude and DFT phase. Therefore if the performances of the classical methods are to be improved in terms of speech quality, the effect of DFT phase on speech quality needs to be studied. In this work, the contribution of DFT phase on speech quality is investigated through some simulations using an objective quality assessment criterion. It is concluded from these simulations that, the phase spectrum has a significant effect on speech quality for short durations of analysis windows. Furthermore, phase values of low frequency components are found to have the largest contribution to this quality improvement. Under the motivation of these results, a new enhancement method is proposed which modifies the phase of certain low frequency components as well as the magnitude spectrum. The proposed algorithm is implemented in MATLAB environment. The results indicate that the proposed system improves the performance of the classical methods in terms of speech quality.

No description available.

This thesis applies an electromagnetic feature extraction technique to design electromagnetic target classifiers for conductors, dielectrics and dielectric coated conductors using their natural resonance related late-time scattered responses. Classifier databases contain scattered data at only a few aspects for each candidate target. The targets are dielectric spheres of varying sizes and refractive indices, perfectly conducting spheres varying sizes and dielectric coated conducting spheres of varying refractive indices and thickness in coating. The applied classifier design technique is suitable for real-time target classification because of the computational efficiency of feature extraction and decision making approaches. The Wigner-Ville Distribution (WD) is employed in this study in addition to the Principal Components Analysis (PCA) technique to extract target features mainly from late-time target responses. WD is applied to the back-scattered responses at different aspects. To decrease aspect dependency, feature vectors are extracted from selected late-time portions of the WD outputs that include natural resonance related information. Principal components analysis is also used to fuse the feature vectors and/or late-time target responses extracted from reference aspects of a given target into a single characteristic feature vector for each target to further reduce aspect dependency.

Made available in DSpace on 2014-07-29T15:08:19Z (GMT). No. of bitstreams: 1 Reconhecimento automatico tinamidae.pdf: 1584398 bytes, checksum: 2f87f01e251266076588ff37c7d6322f (MD5) Previous issue date: 2012-04-11 / This work presents a comprehensive approach to develop a system for recognizing birds by vocalization. The approach specifically addresses the recognition of birds of the Tinamidae family proposing the analysis of data related to the frequency and song of the bird and also classifying and determining the species of bird. The study differs from related research primarily for performing the pre-processing stage automatically. This stage determines the following characteristics: the minimum, the maximum and the stronger frequencies. It s still made a segmentation of the bird singing in periods of sound and silence. The time of singing is also used as a characteristic peculiar to each species analyzed. For the automatic determination of the characteristics of the frequency and song of the bird, an analysis of the power spectral density was made for each time period specified in the frequency using the spectrogram of the song. The recognition and classification technique adopted was the nearest neighbor, using Euclidean distance normalized by the standard deviation. The accuracy of the technique used was 94.12%. / O presente trabalho apresenta uma abordagem completa para o desenvolvimento de um sistema de reconhecimento de aves através da vocalização. A abordagem trata especificamente do reconhecimento de aves da família Tinamidae, propondo analisar dados relacionados à frequência e canto da ave, classificando e determinando a espécie da ave. O estudo feito diferencia-se das pesquisas correlatas principalmente por realizar a etapa de pré-processamento de maneira automática, determinando nesta fase as características de frequência mínima, máxima e mais intensa. Ainda é feita uma segmentação do canto da ave, dividindo o mesmo em períodos de som e silêncio, sendo o tempo de canto também usado como característica peculiar de cada espécie analisada. Para a determinação das características de frequência e canto da ave de maneira automática foi feita uma análise da densidade espectral de potência para cada período de tempo especificado em relação à frequência, utilizando para isso o espectrograma do canto. A técnica de classificação e reconhecimento adotada foi a do vizinho mais próximo, aplicando a distância Euclidiana normalizada pelo desvio padrão. A acurácia da técnica utilizada foi de 94,12%.

A presente dissertação tem como objetivo principal estudar a propagação acústica em escoamentos monofásicos. Para tal, são analisados sinais transientes de pressão fornecidos por sensores instalados em posições conhecidas na linha de teste, através do estudo de técnicas de análise de sinais, a fim de investigar se as variações do conteúdo espectral dos sinais são influenciadas pela ocorrência de vazamentos no duto. A análise dos sinais foi realizada nos planos temporal, frequencial, tempo-frequência e estatístico. Os resultados experimentais foram obtidos no oleoduto piloto do NETeF - Núcleo de Engenharia Térmica e Fluidos da USP - Universidade de São Paulo, com uma seção de testes com 1500 metros e diâmetro de 51,2 mm, com escoamento monofásico de água. Os resultados obtidos através da análise tempo-frequência mostraram-se satisfatórios, sendo esta técnica capaz de identificar a composição espectral instantânea de um sinal, ou seja, foi eficiente na identificação de picos de amplitude da frequência ao longo do eixo temporal. Além disso, a análise probabilística, através do desvio-padrão do sinal também mostrou-se eficiente exibindo uma disparidade significativa entre os sinais com e sem vazamento. / The present dissertation reports on the study of the acoustic propagation in single-phase flow. It analyzes the transient signals provided by pressure sensors in known locations in the test line through the study of signal analysis techniques to investigate if the variations in spectral content of the signals are influenced by the occurrence of leaks in the pipe. The analysis of signals was performed in the time, frequency, time-frequency and statistical plans. The experimental results were obtained in a 1500 meter-long and 51.2 millimeter-diameter pilot pipeline at the Center of Thermal Engineering and Fluids, with single-phase flow of water. The results obtained by time-frequency analysis were satisfactory, allowing identifying the spectral composition of an instantaneous signal, i.e., the analysis was effective in identifying the frequency amplitude peaks along the time axis. Moreover, probabilistic analysis using the standard deviation of the signal was also efficient, displaying a significant disparity between the signals with and without leakage.

Neste trabalho simulou-se o escoamento do fluxo de ar em um tubo, com um anteparo de formato triangular com arestas cortantes, posicionado no centro do tubo. O objetivo do estudo é a análise do comportamento do número de Strouhal em função do número de Reynolds. Para isto, foi utilizada a técnica da análise tempo-freqüência, baseada na transformada de Fourier e na transformada de Gabor. Os ensaios foram realizados com o fluxo com velocidades médias de escoamento de 3 a 10 m/s, sendo utilizado um sensor de pressão tipo piezo-resistivo para a detecção da flutuação de pressão ocasionada pelo desprendimento e formação dos vórtices. Os ensaios foram realizados em cinco etapas com o objetivo de se verificar a influência dos seguintes parâmetros na coleta de sinais e no fenômeno: ruído da rede elétrica; influência do anteparo e do ruído proveniente do escoamento do fluxo de ar; número de pontos da amostragem na coleta dos dados; do comprimento da tubulação; e posicionamento do sensor. Pode-se observar, a sensibilidade do sistema de medição através do ensaio realizado sem o anteparo, sendo verificada a influência do ruído do escoamento de ar pelo tubo; pode-se observar também uma pequena interferência do ruído da rede elétrica predominantemente para velocidades abaixo de 3 m/s. Apesar das influências citadas, e utilizando a transformada de Gabor para análise dos sinais, observou-se um sinal mais intenso na freqüência dos vórtices para as velocidades de escoamento, podendo-se comprovar que o número de Strouhal permanece quase constante e é independente do número de Reynolds, devendo-se ressaltar que esta conclusão é valida para números de Reynolds compreendidos na faixa de 3000 a 100000. No experimento obteve-se um fator de sensibilidade (freqüência vórtices/velocidade média) de 8,2 Hz/m/s, e número de Strouhal médio de 0,196. / This work concerns the simulation of an air flux through a pipe with a triangular bluff body positioned inside it. In order to study the behavior of the Strouhal number in function of the Reynolds number. For this, the time-frequency analysis technique was used, based on Fourier transform and the Gabor transform. The experiments were carried out with an air flux velocity ranging from 3 to 10 m/s and using a piezoresistive pressure sensor to detect pressure fluctuations caused by the shedding and vortex formation. The experimental procedures were divided in five stages to make it possible to verify the influence of the following parameters in the signal data acquisition: electric network noise, the bluff body presence and the noise generated due to its presence, number of sampling data points, tubing length and sensor positioning. The sensitivity of the experiment could be observed testing the air flowing with no bluff body inside the pipe. Thus, it was possible to investigate the influence of the noise generated due to this flux limiting body. It could be also observed, mainly at 3 m/s or less, the noise generated due to the electric network. Despite the listed influences, and with the use of the Gabor transform, a more intense signal on the vortex frequency for the flow velocity was observed, showing that the Strouhal number remains almost constant and is independent of the Reynolds number. It is important to recall that this conclusion is valid for Reynolds numbers between 3000 and 100000. In the experiments the factor of sensitivity (vortex frequency/mean velocity) obtained was 8,2 Hz/m/s and the mean Strouhal number 0,196.

Cette thèse porte sur l'utilisation de la magnétoencéphalographie (MEG) comme outil d'étude de la dynamique des réponses corticales durant la production de mots. Les données empiriques accumulées dans la littérature sont évaluées au regard des modèles psycho- et neuro-linguistiques de la production des mots et du langage. Nous réalisons une exploration de l'évolution des modèles psycho-linguistiques et effectuons en ce sens une revue de la littérature MEG. Les forces et limites de la technique et des données empiriques existantes sont considérées et utilisées pour établir un protocole de dénomination d'images qui soit compatible avec la MEG. Nous développons ensuite une étude empirique réalisée en MEG, en utilisant une manipulation visuo-sémantique pour explorer la dynamique des réponses corticales. Cette étude démontre une large réponse bi-hémisphérique avec des différences inter-conditions précoces (~100ms) dans la BA8 et des différences dans le gyrus cingulaire antérieur, le cortex médial temporal antérieur droit à 207ms et dans la jonction temporo-pariétale à 233ms après apparition du stimulus. Des différences entre les conditions apparaissent tardivement dans le cuneus droit et suggèrent également un traitement visuel en cours. Nos résultats questionnent le timing estimé pour les traitements phonologiques et sémantiques suggérés par les modèles sériels actuels de production du langage. A la lumière de la revue de la littérature et de l'étude empirique conduite, nous évaluons les modèles existants et discutons des directions potentielles pour les recherches futures. / This thesis concerns the use of magnetoencephalography [MEG] as a tool for investigating the dynamics of the cortical response during word production. The evidence gathered is considered in the context of existing psycho- and neuro-linguistic models of word and speech production. An exploration of the evolution of psycholinguistic models is performed, motivating a review of the MEG literature. The strengths and limitations of the technique and existing evidence are considered, and used to guide the design of a picture naming protocol compatible with MEG. An empirical MEG study is then developed and implemented using a visuo-semantic manipulation to explore the dynamics of the cortical response. This study demonstrates a broad, bi-hemispheric response with early (~100ms) between-conditions differences in bilateral BA8 and anterior cingulate cortex, in right anterior medial temporal cortex at 207ms, and a difference in right temporo-parietal junction at 233ms post stimulus. Late between conditions differences in the right cuneus also suggest ongoing visual processing. Our findings question the timing estimated for semantic and phonological processing suggested by current serial models of speech processing. In the light of the review and empirical study, a contextual evaluation of existing models is performed and potential future avenues of investigation are discussed.