De nos jours, la compression de l'image médicale est un processus essentiel dans les systèmes de cybersanté. Compresser des images médicales de haute qualité est une exigence vitale pour éviter de mal diagnostiquer les examens médicaux par les radiologues. WAAVES est un algorithme de compression d'images médicales prometteur basé sur la transformée en ondelettes discrètes (DWT) qui permet d'obtenir une performance de compression élevée par rapport à l'état de la technique. Les principaux objectifs de ce travail sont d'améliorer la qualité d'image lors de la compression à l'aide de WAAVES et de fournir une architecture DWT haute vitesse pour la compression d'image sur des systèmes embarqués. En ce qui concerne l'amélioration de la qualité, les systèmes de nombres logarithmiques (LNS) ont été explorés pour être utilisés comme une alternative à l'arithmétique linéaire dans les calculs de DWT. Une nouvelle bibliothèque LNS a été développée et validée pour réaliser le DWT logarithmique. En outre, une nouvelle méthode de quantification appelée (LNS-Q) basée sur l'arithmétique logarithmique a été proposée. Un nouveau schéma de compression (LNS-WAAVES) basé sur l'intégration de la méthode Hybrid-DWT et de la méthode LNS-Q avec WAAVES a été développé. Hybrid-DWT combine les avantages des domaines logarithmique et linéaire conduisant à l'amélioration de la qualité d'image et du taux de compression. Les résultats montrent que LNS-WAAVES est capable d'obtenir une amélioration de la qualité d'un pourcentage de 8% et de 34% par rapport aux WAAVES en fonction des paramètres de configuration de compression et des modalités d'image. Pour la compression sur les systèmes embarqués, le défi majeur consistait à concevoir une architecture 2D DWT qui permet d'obtenir un débit de 100 trames full HD. Une nouvelle architecture unifiée de calcul 2D DWT a été proposée. Cette nouvelle architecture effectue à la fois des transformations horizontale et verticale simultanément et élimine le problème des accès de pixel d'image en colonne à partir de la RAM DDR hors-puce. Tous ces facteurs ont conduit à une réduction de la largeur de bande DDR RAM requise de plus de 2X. Le concept proposé utilise des tampons de ligne à 4 ports conduisant à quatre opérations en parallèle pipeline: la DWT verticale, la transformée DWT horizontale et les opérations de lecture / écriture vers la mémoire externe. L'architecture proposée a seulement 1/8 de cycles par pixel (CPP) lui permettant de traiter plus de 100fps Full HD et est considérée comme une solution prometteuse pour le futur traitement vidéo 4K et 8K. Enfin, l'architecture développée est hautement évolutive, surperforme l'état de l'art des travaux connexes existants, et est actuellement déployé dans un prototype médical EEG vidéo. / Nowadays, medical image compression is an essential process in eHealth systems. Compressing medical images in high quality is a vital demand to avoid misdiagnosing medical exams by radiologists. WAAVES is a promising medical images compression algorithm based on the discrete wavelet transform (DWT) that achieves a high compression performance compared to the state of the art. The main aims of this work are to enhance image quality when compressing using WAAVES and to provide a high-speed DWT architecture for image compression on embedded systems. Regarding the quality improvement, the logarithmic number systems (LNS) was explored to be used as an alternative to the linear arithmetic in DWT computations. A new LNS library was developed and validated to realize the logarithmic DWT. In addition, a new quantization method called (LNS-Q) based on logarithmic arithmetic was proposed. A novel compression scheme (LNS-WAAVES) based on integrating the Hybrid-DWT and the LNS-Q method with WAAVES was developed. Hybrid-DWT combines the advantages of both the logarithmic and the linear domains leading to enhancement of the image quality and the compression ratio. The results showed that LNS-WAAVES is able to achieve an improvement in the quality by a percentage of 8% and up to 34% compared to WAAVES depending on the compression configuration parameters and the image modalities. For compression on embedded systems, the major challenge was to design a 2D DWT architecture that achieves a throughput of 100 full HD frame/s. A novel unified 2D DWT computation architecture was proposed. This new architecture performs both horizontal and vertical transform simultaneously and eliminates the problem of column-wise image pixel accesses to/from the off-chip DDR RAM. All of these factors have led to a reduction of the required off-chip DDR RAM bandwidth by more than 2X. The proposed concept uses 4-port line buffers leading to pipelined parallel four operations: the vertical DWT, the horizontal DWT transform, and the read/write operations to the external memory. The proposed architecture has only 1/8 cycles per pixel (CPP) enabling it to process more than 100fps Full HD and it is considered a promising solution for future 4K and 8K video processing. Finally, the developed architecture is highly scalable, outperforms the state of the art existing related work, and currently is deployed in a video EEG medical prototype.

No description available.

This thesis aims to investigate the use of the discrete wavelet transform of a stock index as a means to forecast intraday returns. This will be done by having the discrete wavelet transform as an input in a Transformers neural network with binary labels signifying a positive or negative next-day return. The input will be limited to a time horizon of 30 days since the entire history is likely not necessary, meaning we do not care about the discrete wavelet transform 5 years ago when we are trying to predict the next day's return. The network will be evaluated in terms of accuracy and a "trading strategy" on the OMXS30 index, where we compare the performance of the network with that of the original index. Overall, the performance of the discrete wavelet transform and the Transformers network was okay. The performance was slightly better than simply going long on the index, but not by much, and when factoring in transaction costs it is probably not a worthwhile strategy to use this setup. / Detta examensarbete syftar till att undersöka användningen av den diskreta wavelet-transformen av ett aktieindex som ett sätt att prognostisera nästa dags avkastning. Detta kommer att göras genom att ha den diskreta wavelet-transformen som en input i ett Transformersnätverk med målet att utföra binär klassificiering. Inputen kommer att vara begränsad till en tidshorisont på 30 dagar eftersom hela historien sannolikt inte är nödvändigt, vilket betyder att vi inte bryr oss om den diskreta wavelet-transformen för 5 år sedan när vi försöker prognostisera nästa dags avkastning. Nätverket kommer att utvärderas med hjälp av accuracy och en tradingstrategi som kommer utvärderas på OMXS30-indecet, där vi jämför tradingstrategins prestation med det ursprungliga indexet. Slutsatsen man kan dra av det här examensarbetet är att den diskreta wavelet-transformen och Transformers-nätverkets prestanda var acceptabel. Trading strategin var något bättre än att bara gå lång på indexet, men inte mycket, och när man räknar in transaktionskostnader är det förmodligen inte en lönsam strategi.

No / Study is presented into the performance of Fast Fourier Transform (FFT) and Discrete Wavelet Transform (DWT) and MIMO-DWT with transmit beamforming. Feedback loop has been used between the equalizer at the transmitter to the receiver which provided the channel state information which was then used to construct a steering matrix for the transmission sequence such that the received signals at the transmitter can be combined constructively in order to provide a reliable and improved system for next generation wireless systems. As convolution in time domain equals multiplication in frequency domain no such counterpart exist for the symbols in space, means linear convolution and Intersymbol Interference (ISI) generation so both zero forcing (ZF) and minimum mean squared error (MMSE) equalizations have been employed. The results show superior performance improvement and in addition allow keeping the processing, power and implementation cost at the transmitter which has less constraints and the results also show that both equalization algorithms perform alike in wavelets and the ISI is spread equally between different wavelet domains.

Digital Signal Processing has been dominated by the Fourier transform since the Fast Fourier Transform (FFT) was developed in 1965 by Cooley and Tukey. In the 1980's a new transform was developed called the wavelet transform, even though the first wavelet goes back to 1910. With the Fourier transform, all information about localized changes in signal features are spread out across the entire signal space, making local features global in scope. Wavelets are able to retain localized information about the signal by applying a function of a limited duration, also called a wavelet, to the signal. As with the Fourier transform, the discrete wavelet transform has an inverse transform, which allows us to make changes in a signal in the wavelet domain and then transform it back in the time domain. In this thesis, we have investigated the filtering properties of this technique and analyzed its performance under various settings. Another popular application of wavelet transform is data compression, such as described in the JPEG 2000 standard and compressed digital storage of fingerprints developed by the FBI. Previous work on filtering has focused on the discrete wavelet transform. Here, we extended that method to the stationary wavelet transform and found that it gives a performance boost of as much as 9 dB over that of the discrete wavelet transform. We also found that the SNR of noise filtering decreases as a frequency of the base signal increases up to the Nyquist limit for both the discrete and stationary wavelet transforms. Besides filtering the signal, the discrete wavelet transform can also be used to estimate the standard deviation of the white noise present in the signal. We extended the developed estimator for the discrete wavelet transform to the stationary wavelet transform. As with filtering, it is found that the quality of the estimate decreases as the frequency of the base signal increases. Many interesting signals are self-similar, which means that one of their properties is invariant on many different scales. One popular example is strict self-similarity, where an exact copy of a signal is replicated on many scales, but the most common property is statistical self-similarity, where a random segment of a signal is replicated on many different scales. In this work, we investigated wavelet-based methods to detect statistical self-similarities in a signal and their performance on various types of self-similar signals. Specifically, we found that the quality of the estimate depends on the type of the units of the signal being investigated for low Hurst exponent and on the type of edge padding being used for high Hurst exponent. / Master of Science

The task of detecting human faces within either a still image or a video frame is one of the most popular object detection problems. For the last twenty years researchers have shown great interest in this problem because it is an essential pre-processing stage for computing systems that process human faces as input data. Example applications include face recognition systems, vision systems for autonomous robots, human computer interaction systems (HCI), surveillance systems, biometric based authentication systems, video transmission and video compression systems, and content based image retrieval systems. In this thesis, non-traditional methods are investigated for detecting human faces within color images or video frames. The attempted methods are chosen such that the required computing power and memory consumption are adequate for real-time hardware implementation. First, a standard color image database is introduced in order to accomplish fair evaluation and benchmarking of face detection and skin segmentation approaches. Next, a new pre-processing scheme based on skin segmentation is presented to prepare the input image for feature extraction. The presented pre-processing scheme requires relatively low computing power and memory needs. Then, several feature extraction techniques are evaluated. This thesis introduces feature extraction based on Two Dimensional Discrete Cosine Transform (2D-DCT), Two Dimensional Discrete Wavelet Transform (2D-DWT), geometrical moment invariants, and edge detection. It also attempts to construct a hybrid feature vector by the fusion between 2D-DCT coefficients and edge information, as well as the fusion between 2D-DWT coefficients and geometrical moments. A self organizing map (SOM) based classifier is used within all the experiments to distinguish between facial and non-facial samples. Two strategies are tried to make the final decision from the output of a single SOM or multiple SOM. Finally, an FPGA based framework that implements the presented techniques, is presented as well as a partial implementation. Every presented technique has been evaluated consistently using the same dataset. The experiments show very promising results. The highest detection rate of 89.2% was obtained when using a fusion between DCT coefficients and edge information to construct the feature vector. A second highest rate of 88.7% was achieved by using a fusion between DWT coefficients and geometrical moments. Finally, a third highest rate of 85.2% was obtained by calculating the moments of edges. / Master of Science

Submitted by Renata Lopes (renatasil82@gmail.com) on 2017-04-24T15:28:35Z No. of bitstreams: 1 paolajohanasaboyalopez.pdf: 1042873 bytes, checksum: a46dd95de00e062cba39ef4b9b642462 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2017-04-24T17:09:24Z (GMT) No. of bitstreams: 1 paolajohanasaboyalopez.pdf: 1042873 bytes, checksum: a46dd95de00e062cba39ef4b9b642462 (MD5) / Made available in DSpace on 2017-04-24T17:09:24Z (GMT). No. of bitstreams: 1 paolajohanasaboyalopez.pdf: 1042873 bytes, checksum: a46dd95de00e062cba39ef4b9b642462 (MD5) Previous issue date: 2013-06-03 / A presente dissertação tem por objetivo propor e avaliar cinco técnicas de detecção de ruídos impulsivos para a melhoria da transmissão digital de dados via redes de energia elétrica (do inglês, Power Line Communications) (PLC). As técnicas propostas contemplam a detecção de ruídos impulsivos no domínio do tempo discreto, no domínio da transformada wavelet discreta (do inglês, Discrete Wavelet Transform) (DWT) e no domínio da transformada discreta de Fourier (do inglês, Discrete Fourier Transform) (DFT). Tais técnicas fazem uso de métodos de extração e seleção de características, assim como métodos de detecção de sinais baseados na teoria de Bayes e redes neurais. Análises comparativas explicitam as vantagens e desvantagens de cada uma das técnicas propostas para o problema em questão, e ainda indicam que estas são bastante adequadas para a solução do mesmo. / This dissertation aims to propose and evaluate five techniques for impulsive noise detection in order to improve digital communications through power line channels. The imput signals for the proposed detection techniques are impulsive noise signals on discrete-time domain, on the Discrete Wavelet Transform domain and on the Discrete Fourier Transform domain and it makes use of feature extraction and selection techniques, as well as detection techniques supported on Bayes Theory and Multi-layer Perceptron Neural Networks. Comparative analysis show some advantages and disadvantages of each proposed technique and the relevance of them to solve the impulsive noise detection problem.

Este trabalho apresenta o estudo e o desenvolvimento de um algoritmo híbrido para detecção, classificação e localização de faltas em sistemas com três terminais utilizando como principal ferramenta a transformada wavelet (TW) em suas versões discreta (TWD) e estacionária (TWE). O algoritmo é dito híbrido, pois alia duas metodologias para localizar a falta. A primeira baseada na análise de componentes de alta freqüência (ondas viajantes) e a segunda, baseada na extração dos componentes fundamentais para o cálculo da impedância aparente. A metodologia proposta foi concebida de maneira a trabalhar com dados sincronizados dos três terminais ou apenas dados locais para estimar a localização da falta. O localizador híbrido escolhe automaticamente qual a melhor técnica de localização ser utilizada para alcançar uma localização confiável e precisa. Deste modo, um método pode suprir as dificuldades do outro, ou, no mínimo, fornecer mais informações para que, junto ao conhecimento do operador, uma localização próxima da ótima possa ser alcançada. Com o objetivo de testar e validar a aplicabilidade do algoritmo de localização de faltas híbrido para linhas com três terminais, utilizou-se de dados de sinais faltosos obtidos através de simulações do software ATP (Altenative Transients Program), levando-se em conta a variação de diversos parâmetros que poderiam influenciar o desempenho do algoritmo proposto. Os resultados alcançados pelo algoritmo frente às situações avaliadas são bastante animadores, apontando a uma promissora aplicabilidade do mesmo. / This work presents a study and development of a hybrid algorithm for fault detection, classification and location in tree terminal lines based on wavelet transform (WT). It will be presented in two versions: discrete wavelet transform (DWT) and stationary wavelet transform (SWT). The algorithm is called hybrid because it uses two fault location methodologies: one based on fundamental components and other based on traveling waves. The proposed methodology works either with synchronized tree terminal data or only local data. The hybrid fault locator chooses automatically which location technique to be used in order to reach a reliable and accurate fault location. In this manner, this technique can avoid some difficulties present in other techniques, aiming to reach an optimized fault location. The proposed hybrid fault location was evaluated by simulated fault signals obtained by alternative transient program (ATP). In the tests, several parameters, which would influence the performance of the hybrid algorithm, were varied, such as: fault inception angle, fault resistance, fault type, etc. The results obtained by the proposed methodology are very encouraging and it points out to a very promising application.

Submitted by Angela Maria de Oliveira (amolivei@uepg.br) on 2018-02-08T13:10:32Z No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) Ronaldo Mendes Evaristo.pdf: 2297085 bytes, checksum: 293f7e08aca7690caa0d317480f9e18e (MD5) / Made available in DSpace on 2018-02-08T13:10:32Z (GMT). No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) Ronaldo Mendes Evaristo.pdf: 2297085 bytes, checksum: 293f7e08aca7690caa0d317480f9e18e (MD5) Previous issue date: 2017-12-08 / Nos ultimos anos, o aumento da incidência de doenças cardiovasculares na população mundial vem motivando a comunidade científica a buscar novas técnicas ou inovações tecnológicas para complementar os métodos existentes para avaliação do desempenho do coração. Dentre elas, destaca-se a análise da Variabilidade da Frequência Cardíarca (VFC) via eletrocardiograma (ECG), método não invasivo importante na detecção de patologias leves e moderadas cada vez mais frequentes nos seres humanos como doenças coronarianas, arritmias, bradicardias e taquicardias, além de distúrbios na relação entre os sistemas nervosos simpático e parassimpático. Neste trabalho é introduzida uma inovação em um modelo matemático baseado em modulações de exponenciais gaussianas utilizado para reproduzir a morfologia do ECG de seres humanos. Trata-se da introdução de tacogramas gerados por um processo estocástico autorregressivo (AR), previamente à integração das equações diferenciais do modelo, capaz de reproduzir com maior fidelidade a VFC quando comparados com dados experimentais de adultos saudaveis e de adultos com doença arterial coronariana (DAC). Para validar o modelo, os resultados simulados são comparados com dados experimentais via espectro de potencia da transformada wavelet discreta (TWD), gráficos de Poincare e pela analise de flutuação sem tendência. Verificamos que a DAC não altera a morfologia do ECG em situação de repouso, mas influencia significativamente na VFC, sendo que o modelo matemático proposto absorve e reproduz esse comportamento. / In recent years, the increase in the incidence of cardiovascular diseases in the world population has motivated the scientific community to seek new techniques or technological innovations to complement existing methods for assessing heart performance. Among them, stands out the analysis of the Heart Rate Variability (HRV) by electrocardiogram (ECG), an important non-invasive method for the detection of mild and moderate pathologies that are increasingly frequent in humans such as coronary diseases, arrhythmias, bradycardia and tachycardias, besides disturbances in the relationship between the sympathetic and parasympathetic nervous systems. This work introduces an innovation in a mathematical model based on Gaussian exponential modulations used to reproduce the ECG morphology of humans. This is the introduction of tachograms generated by an autoregressive stochastic process (AR), prior to the integration of the diferential equations of the model, capable of reproducing with better delity the HRV when compared with experimental data of healthy adults and adults with coronary artery disease (CAD). In order to validate the model, the simulated results are compared with experimental data using the discrete wavelet transform (DWT) power spectrum, Poincare plots and the detrended uctuation analysis (DFA). We verified that CAD does not change the morphology of ECG in resting state, but it has a significant in uence on HRV, and the proposed mathematical model absorbs and reproduces this behavior.

Cette thèse a pour objectif de fournir aux utilisateurs de SMPS (Scanning Mobility Particle Sizer) une méthodologie pour calculer les incertitudes associées à l’estimation de la granulométrie en nombre des aérosols. Le résultat de mesure est le comptage des particules de l’aérosol en fonction du temps. Estimer la granulométrie en nombre de l’aérosol à partir des mesures CNC revient à considérer un problème inverse sous incertitudes.Une revue des modèles existants est présentée dans le premier chapitre. Le modèle physique retenu fait consensus dans le domaine d’application.Dans le deuxième chapitre, un critère pour l’estimation de la granulométrie en nombre qui couple les techniques de régularisation et de la décomposition sur une base d’ondelettes est décrit.La nouveauté des travaux présentés réside dans l’estimation de granulométries présentant à la fois des variations lentes et des variations rapides. L’approche multi-échelle que nous proposons pour la définition du nouveau critère de régularisation permet d’ajuster les poids de la régularisation sur chaque échelle du signal. La méthode est alors comparée avec la régularisation classique. Les résultats montrent que les estimations proposées par la nouvelle méthode sont meilleures (au sens du MSE) que les estimations classiques.Le dernier chapitre de cette thèse traite de la propagation de l’incertitude à travers le modèle d’inversiondes données. C’est une première dans le domaine d’application car aucune incertitude n’est associée actuellement au résultat de mesure. Contrairement à l’approche classique qui utilise un modèle fixe pour l’inversion en faisant porter l’incertitude sur les entrées, nous proposons d’utiliser un modèle d’inversion aléatoire (tirage Monte-Carlo) afin d’intégrer les erreurs de modèle. Une estimation moyenne de la granulométrie en nombre de l’aérosol et une incertitude associée sous forme d’une région de confiance à 95 % est finalement présentée sur quelques mesures réelles. / This thesis aims to provide SMPS (Scanning Mobility Particle Sizer) users with a methodology to compute the uncertainties associated with the estimation of aerosol size distributions. Recovering aerosol size distribution yields to consider an inverse problem under uncertainty.The first chapter of his thesis presents a review of physical models and it shows that competitive theories exist to model the physic.A new criterion that couples regularization techniques and decomposition on a wavelet basis is described in chapter 2 to perform the estimation of the size distribution.Main improvement of this work is brought when size distributions to be estimated show both broad and sharp profiles. The multi-scale approach helps to adjust the weights of the regularization on each scale of the signal. The method is then tested against common regularization and shows better estimates (in terms of the mean square error).Last chapter of this thesis deals with the propagation of the uncertainty through the data inversion process.Results from SMPS measurements are not given with any uncertainty at this time so providing end-users with an uncertainty is already a real improvement. Common approach is to consider a fixed physical model and to model the inputs (particle count) as random. We choose to consider both the physical model as well as the inputs as random to account for the model error. The result is expressed as a mean estimate of the size distribution with a 95% coverage region. The all methodology is finally tested on real measurements.