Wavetrack, implementazione ottimizzata di un algoritmo di pitch tracking basato su wavelet

WaveTrack é un'implementazione ottimizzata di un algoritmo di pitch tracking basato su wavelet, nello specifico viene usata la trasformata Fast Lifting Wavelet Transform con la wavelet di Haar. La libreria è stata scritta nel linguaggio C e tra le sue peculiarità può vantare tempi di latenza molto bassi, un'ottima accuratezza e una buona flessibilità d'uso grazie ad alcuni parametri configurabili.

Trasformate wavelet

La presente tesi vuole dare una descrizione delle Trasformate Wavelet indirizzata alla codifica dell’immagine in formato JPEG2000. Dopo aver quindi descritto le prime fasi della codifica di un’immagine, procederemo allo studio dei difetti derivanti dall’analisi tramite la Trasformata Discreta del Coseno (utilizzata nel formato predecessore JPEG). Dopo aver quindi descritto l’analisi multirisoluzione e le caratteristiche che la differenziano da quest’ultima, analizzeremo la Trasformata Wavelet dandone solo pochi accenni teorici e cercando di dedurla, in una maniera più indirizzata all’applicazione. Concluderemo la tesi descrivendo la codifica dei coefficienti calcolati, e portando esempi delle innumerevoli applicazioni dell’analisi multirisoluzione nei diversi campi scientifici e di trasmissione dei segnali.

Analyse zur Auswirkung der irreversiblen Wavelet-Kompression (JPEG2000) auf CT-Bildverarbeitungs- und Darstellungsalgorithmen

Der technische Fortschritt konfrontiert die medizinische Bildgebung wie keine andere Sparte der Medizin mit einem rasanten Anstieg zu speichernder Daten. Anschaffung, Wartung und Ausbau der nötigen Infrastruktur entwickeln sich zunehmend zu einem ökonomischen Faktor. Ein Verfahren, welches diesem Trend etwas entgegensetzten könnte ist die irreversible Bilddatenkompression. Sie ist seit über 10 Jahren Gegenstand vieler Studien, deren Ergebnisse sich wiederum in Empfehlungen zum Einsatz irreversibler Kompression mehrerer nationaler und internationaler Organisation, wie CAR, DRG, RCR und ESR wiederspiegeln. Tenor dieser Empfehlungen ist, dass der Einsatz von moderater irreversibler Bilddatenkompression sicher und sinnvoll ist. Teil dieser Empfehlungen sind auch Angaben über das Maß an Kompression, ausgedrückt in Kompressionsraten, welche je nach Untersuchung und anatomischer Region als sicher anwendbar gelten und keinen diagnostisch relevanten Verlust der komprimierten Bilder erzeugen.rnVerschiedene Kompressionsalgorithmen wurden vorgeschlagen. Letztendlich haben sich vor allem die beiden weit verbreiteten Algorithmen JPEG und JPEG2000 bewährt. Letzterer erfährt in letzter Zeit zunehmen Anwendung, aufgrund seiner einfacheren Handhabung und seiner umfangreichen Zusatzfunktionen.rnAufgrund rechtlich-ethischer Bedenken hat die irreversible Kompression keine breite praktische Anwendung finden können. Dafür verantwortlich ist unter anderem auch die Unklarheit, wie sich irreversible Kompression auf Nach- und Weiterverarbeitung (sog. Postprocessing) medizinischer Bilder, wie Segmentierung, Volumetrie oder 3D-Darstellung, auswirkt. Bisherige Studien zu diesem Thema umfassen vier verschiedene Postprocessing-Algorithmen. Die untersuchten Algorithmen zeigten sich bei verlustbehafteter Kompression im Bereich der erwähnten, publizierten Kompressionsraten weitgehend unbeeinflusst. Lediglich die computergestützte Messung von Stenosegraden in der digitalen Koronarangiographie kollidiert mit den in Großbritannien geltenden Empfehlungen. Die Verwendung unterschiedlicher Kompressionsalgorithmen schränkt die allgemeinernAussagekraft dieser Studienergebnisse außerdem ein.rnZur Erweiterung der Studienlage wurden vier weitere Nach- und Weiterverarbeitungsalgorithmen auf ihre Kompressionstoleranz untersucht. Dabei wurden die Kompressionsraten von 8:1, 10:1 und 15:1 verwendet, welche um die empfohlenen Kompressionsraten von CAR, DRG, RCR und ESR liegen und so ein praxisnahes Setting bieten. Als Kompressionsalgorithmus wurde JPEG2000 verwendet, aufgrund seiner zunehmenden Nutzung in Studien sowie seiner bereits erwähnten Vorzüge in Sachen Handhabung und Zusatzfunktionen. Die vier Algorithmen umfassten das 3D-Volume rendering von CT-Angiographien der Becken-Bein-Gefäße, die Computer-assistierte Detektion von Lungenrundherden, die automatisierte Volumetrie von Leberrundherden und die funktionelle Bestimmung der Ejektionsfraktion in computertomographischen Aufnahmen des Herzens.rnAlle vier Algorithmen zeigten keinen Einfluss durch irreversibler Bilddatenkompression in denrngewählten Kompressionsraten (8:1, 10:1 und 15:1). Zusammen mit der bestehenden Literatur deuten die Ergebnisse an, dass moderate irreversible Kompression im Rahmen aktueller Empfehlungen keinen Einfluss auf Nach- und Weiterverarbeitung medizinischer Bilder hat. Eine explizitere Vorhersage zu einem bestimmten, noch nicht untersuchten Algorithmus ist jedoch aufgrund der unterschiedlichen Funktionsweisen und Programmierungen nicht sicher möglich.rnSofern ein Postprocessing Algorithmus auf komprimiertes Bildmaterial angewendet werden soll, muss dieser zunächst auf seine Kompressionstoleranz getestet werden. Dabei muss der Test eine rechtlich-ethische Grundlage für den Einsatz des Algorithmus bei komprimiertem Bildmaterial schaffen. Es sind vor allem zwei Optionen denkbar, die Testung institutsintern, eventuell unter Zuhilfenahme von vorgefertigten Bibliotheken, oder die Testung durch den Hersteller des Algorithmus.

Trasformata di Fourier e trasformata Wavelet

Questo elaborato si concentra sullo studio della trasformata di Fourier e della trasformata Wavelet. Nella prima parte della tesi si analizzano gli aspetti fondamentali della trasformata di Fourier. Si definisce poi la trasformata di Fourier su gruppi abeliani finiti, richiamando opportunamente la struttura di tali gruppi. Si mostra che calcolare la trasformata di Fourier nel quoziente richiede un minor numero di operazioni rispetto a calcolarla direttamente nel gruppo di partenza. L'ultima parte dell'elaborato si occupa dello studio delle Wavelet, dette ondine. Viene presentato quindi il sistema di Haar che permette di definire una funzione come serie di funzioni di Haar in alternativa alla serie di Fourier. Si propone poi un vero e proprio metodo per la costruzione delle ondine e si osserva che tale costruzione è strettamente legata all'analisi multirisoluzione. Un ruolo cruciale viene svolto dall'identità di scala, un'identità algebrica che permette di definire certi coefficienti che determinano completamente le ondine. Interviene poi la trasformata di Fourier che riduce la ricerca dei coefficienti sopra citati, alla ricerca di certe funzioni opportune che determinano esplicitamente le Wavelet. Non tutte le scelte di queste funzioni sono accettabili. Ci sono vari approcci, qui viene presentato l'approccio di Ingrid Daubechies. Le Wavelet costituiscono basi per lo spazio di funzioni a quadrato sommabile e sono particolarmente interessanti per la decomposizione dei segnali. Sono quindi in relazione con l'analisi armonica e sono adottate in un gran numero di applicazioni. Spesso sostituiscono la trasformata di Fourier convenzionale.

Wavelet-denoising of electroencephalogram and the absolute slope method: a new tool to improve electroencephalographic localization and lateralization

OBJECTIVE: In ictal scalp electroencephalogram (EEG) the presence of artefacts and the wide ranging patterns of discharges are hurdles to good diagnostic accuracy. Quantitative EEG aids the lateralization and/or localization process of epileptiform activity. METHODS: Twelve patients achieving Engel Class I/IIa outcome following temporal lobe surgery (1 year) were selected with approximately 1-3 ictal EEGs analyzed/patient. The EEG signals were denoised with discrete wavelet transform (DWT), followed by computing the normalized absolute slopes and spatial interpolation of scalp topography associated to detection of local maxima. For localization, the region with the highest normalized absolute slopes at the time when epileptiform activities were registered (>2.5 times standard deviation) was designated as the region of onset. For lateralization, the cerebral hemisphere registering the first appearance of normalized absolute slopes >2.5 times the standard deviation was designated as the side of onset. As comparison, all the EEG episodes were reviewed by two neurologists blinded to clinical information to determine the localization and lateralization of seizure onset by visual analysis. RESULTS: 16/25 seizures (64%) were correctly localized by the visual method and 21/25 seizures (84%) by the quantitative EEG method. 12/25 seizures (48%) were correctly lateralized by the visual method and 23/25 seizures (92%) by the quantitative EEG method. The McNemar test showed p=0.15 for localization and p=0.0026 for lateralization when comparing the two methods. CONCLUSIONS: The quantitative EEG method yielded significantly more seizure episodes that were correctly lateralized and there was a trend towards more correctly localized seizures. SIGNIFICANCE: Coupling DWT with the absolute slope method helps clinicians achieve a better EEG diagnostic accuracy.

A Wavelet Analysis of Pliopleistocene Climate Indicators: A New View of Periodicity Evolution

Wavelet analysis offers an alternative to Fourier based time-series analysis, and is particularly useful when the amplitudes and periods of dominant cycles are time dependent. We analyse climatic records derived from oxygen isotopic ratios of marine sediment cores with modified Morlet wavelets. We use a normalization of the Morlet wavelets which allows direct correspondence with Fourier analysis. This provides a direct view of the oscillations at various frequencies, and illustrates the nature of the time-dependence of the dominant cycles.

Quality assessment by region in spot images fused by means dual-tree complex wavelet transform

This work is motivated in providing and evaluating a fusion algorithm of remotely sensed images, i.e. the fusion of a high spatial resolution panchromatic image with a multi-spectral image (also known as pansharpening) using the dual-tree complex wavelet transform (DT-CWT), an effective approach for conducting an analytic and oversampled wavelet transform to reduce aliasing, and in turn reduce shift dependence of the wavelet transform. The proposed scheme includes the definition of a model to establish how information will be extracted from the PAN band and how that information will be injected into the MS bands with low spatial resolution. The approach was applied to Spot 5 images where there are bands falling outside PAN’s spectrum. We propose an optional step in the quality evaluation protocol, which is to study the quality of the merger by regions, where each region represents a specific feature of the image. The results show that DT-CWT based approach offers good spatial quality while retaining the spectral information of original images, case SPOT 5. The additional step facilitates the identification of the most affected regions by the fusion process.

Bio-inspired FPGA Architecture for Self-Calibration of an Image Compression Core based on Wavelet Transforms in Embedded Systems

A generic bio-inspired adaptive architecture for image compression suitable to be implemented in embedded systems is presented. The architecture allows the system to be tuned during its calibration phase. An evolutionary algorithm is responsible of making the system evolve towards the required performance. A prototype has been implemented in a Xilinx Virtex-5 FPGA featuring an adaptive wavelet transform core directed at improving image compression for specific types of images. An Evolution Strategy has been chosen as the search algorithm and its typical genetic operators adapted to allow for a hardware friendly implementation. HW/SW partitioning issues are also considered after a high level description of the algorithm is profiled which validates the proposed resource allocation in the device fabric. To check the robustness of the system and its adaptation capabilities, different types of images have been selected as validation patterns. A direct application of such a system is its deployment in an unknown environment during design time, letting the calibration phase adjust the system parameters so that it performs efcient image compression. Also, this prototype implementation may serve as an accelerator for the automatic design of evolved transform coefficients which are later on synthesized and implemented in a non-adaptive system in the final implementation device, whether it is a HW or SW based computing device. The architecture has been built in a modular way so that it can be easily extended to adapt other types of image processing cores. Details on this pluggable component point of view are also given in the paper.

A region based approach to background modeling in a wavelet multi-resolution framework

In the field of detection and monitoring of dynamic objects in quasi-static scenes, background subtraction techniques where background is modeled at pixel-level, although showing very significant limitations, are extensively used. In this work we propose a novel approach to background modeling that operates at region-level in a wavelet based multi-resolution framework. Based on a segmentation of the background, characterization is made for each region independently as a mixture of K Gaussian modes, considering the model of the approximation and detail coefficients at the different wavelet decomposition levels. Background region characterization is updated along time, and the detection of elements of interest is carried out computing the distance between background region models and those of each incoming image in the sequence. The inclusion of the context in the modeling scheme through each region characterization makes the model robust, being able to support not only gradual illumination and long-term changes, but also sudden illumination changes and the presence of strong shadows in the scene

Glottal Parameter Estimation by Wavelet Transform for Voice Biometry

Voice biometry is classically based on the parameterization and patterning of speech features mainly. The present approach is based on the characterization of phonation features instead (glottal features). The intention is to reduce intra-speaker variability due to the `text'. Through the study of larynx biomechanics it may be seen that the glottal correlates constitute a family of 2-nd order gaussian wavelets. The methodology relies in the extraction of glottal correlates (the glottal source) which are parameterized using wavelet techniques. Classification and pattern matching was carried out using Gaussian Mixture Models. Data of speakers from a balanced database and NIST SRE HASR2 were used in verification experiments. Preliminary results are given and discussed.

Accelerating FPGA-based evolution of wavelet transform filters by optimized task scheduling

Adaptive embedded systems are required in various applications. This work addresses these needs in the area of adaptive image compression in FPGA devices. A simplified version of an evolution strategy is utilized to optimize wavelet filters of a Discrete Wavelet Transform algorithm. We propose an adaptive image compression system in FPGA where optimized memory architecture, parallel processing and optimized task scheduling allow reducing the time of evolution. The proposed solution has been extensively evaluated in terms of the quality of compression as well as the processing time. The proposed architecture reduces the time of evolution by 44% compared to our previous reports while maintaining the quality of compression unchanged with respect to existing implementations. The system is able to find an optimized set of wavelet filters in less than 2 min whenever the input type of data changes.

Wavelet-based image fusion in multi-view three-dimensional microscopy

Multi-view microscopy techniques such as Light-Sheet Fluorescence Microscopy (LSFM) are powerful tools for 3D + time studies of live embryos in developmental biology. The sample is imaged from several points of view, acquiring a set of 3D views that are then combined or fused in order to overcome their individual limitations. Views fusion is still an open problem despite recent contributions in the field. We developed a wavelet-based multi-view fusion method that, due to wavelet decomposition properties, is able to combine the complementary directional information from all available views into a single volume. Our method is demonstrated on LSFM acquisitions from live sea urchin and zebrafish embryos. The fusion results show improved overall contrast and details when compared with any of the acquired volumes. The proposed method does not need knowledge of the system's point spread function (PSF) and performs better than other existing PSF independent fusion methods.

Gear dynamics monitoring using discrete wavelet transformation and multi-layer perceptron neural networks

This paper presents a multi-stage algorithm for the dynamic condition monitoring of a gear. The algorithm provides information referred to the gear status (fault or normal condition) and estimates the mesh stiffness per shaft revolution in case that any abnormality is detected. In the first stage, the analysis of coefficients generated through discrete wavelet transformation (DWT) is proposed as a fault detection and localization tool. The second stage consists in establishing the mesh stiffness reduction associated with local failures by applying a supervised learning mode and coupled with analytical models. To do this, a multi-layer perceptron neural network has been configured using as input features statistical parameters sensitive to torsional stiffness decrease and derived from wavelet transforms of the response signal. The proposed method is applied to the gear condition monitoring and results show that it can update the mesh dynamic properties of the gear on line.

A wavelet-based approach for large wind power ramp characterisation

A wavelet-based approach for large wind power ramp characterisation