Post-processing and analysis of tracked hand trajectories

In this thesis, the suitability of different trackers for finger tracking in high-speed videos was studied. Tracked finger trajectories from the videos were post-processed and analysed using various filtering and smoothing methods. Position derivatives of the trajectories, speed and acceleration were extracted for the purposes of hand motion analysis. Overall, two methods, Kernelized Correlation Filters and Spatio-Temporal Context Learning tracking, performed better than the others in the tests. Both achieved high accuracy for the selected high-speed videos and also allowed real-time processing, being able to process over 500 frames per second. In addition, the results showed that different filtering methods can be applied to produce more appropriate velocity and acceleration curves calculated from the tracking data. Local Regression filtering and Unscented Kalman Smoother gave the best results in the tests. Furthermore, the results show that tracking and filtering methods are suitable for high-speed hand-tracking and trajectory-data post-processing.

Adaptive Markov chain Monte Carlo and Bayesian filtering for state space models

This thesis is concerned with the state and parameter estimation in state space models. The estimation of states and parameters is an important task when mathematical modeling is applied to many different application areas such as the global positioning systems, target tracking, navigation, brain imaging, spread of infectious diseases, biological processes, telecommunications, audio signal processing, stochastic optimal control, machine learning, and physical systems. In Bayesian settings, the estimation of states or parameters amounts to computation of the posterior probability density function. Except for a very restricted number of models, it is impossible to compute this density function in a closed form. Hence, we need approximation methods. A state estimation problem involves estimating the states (latent variables) that are not directly observed in the output of the system. In this thesis, we use the Kalman filter, extended Kalman filter, Gauss���Hermite filters, and particle filters to estimate the states based on available measurements. Among these filters, particle filters are numerical methods for approximating the filtering distributions of non-linear non-Gaussian state space models via Monte Carlo. The performance of a particle filter heavily depends on the chosen importance distribution. For instance, inappropriate choice of the importance distribution can lead to the failure of convergence of the particle filter algorithm. In this thesis, we analyze the theoretical L��� particle filter convergence with general importance distributions, where p ���2 is an integer. A parameter estimation problem is considered with inferring the model parameters from measurements. For high-dimensional complex models, estimation of parameters can be done by Markov chain Monte Carlo (MCMC) methods. In its operation, the MCMC method requires the unnormalized posterior distribution of the parameters and a proposal distribution. In this thesis, we show how the posterior density function of the parameters of a state space model can be computed by filtering based methods, where the states are integrated out. This type of computation is then applied to estimate parameters of stochastic differential equations. Furthermore, we compute the partial derivatives of the log-posterior density function and use the hybrid Monte Carlo and scaled conjugate gradient methods to infer the parameters of stochastic differential equations. The computational efficiency of MCMC methods is highly depend on the chosen proposal distribution. A commonly used proposal distribution is Gaussian. In this kind of proposal, the covariance matrix must be well tuned. To tune it, adaptive MCMC methods can be used. In this thesis, we propose a new way of updating the covariance matrix using the variational Bayesian adaptive Kalman filter algorithm.

Parameter identification in time series models

Time series analysis can be categorized into three different approaches: classical, Box-Jenkins, and State space. Classical approach makes a basement for the analysis and Box-Jenkins approach is an improvement of the classical approach and deals with stationary time series. State space approach allows time variant factors and covers up a broader area of time series analysis. This thesis focuses on parameter identifiablity of different parameter estimation methods such as LSQ, Yule-Walker, MLE which are used in the above time series analysis approaches. Also the Kalman filter method and smoothing techniques are integrated with the state space approach and MLE method to estimate parameters allowing them to change over time. Parameter estimation is carried out by repeating estimation and integrating with MCMC and inspect how well different estimation methods can identify the optimal model parameters. Identification is performed in probabilistic and general senses and compare the results in order to study and represent identifiability more informative way.

Automatic recognition of fish from video sequences

The problem of automatic recognition of the fish from the video sequences is discussed in this Master���s Thesis. This is a very urgent issue for many organizations engaged in fish farming in Finland and Russia because the process of automation control and counting of individual species is turning point in the industry. The difficulties and the specific features of the problem have been identified in order to find a solution and propose some recommendations for the components of the automated fish recognition system. Methods such as background subtraction, Kalman filtering and Viola-Jones method were implemented during this work for detection, tracking and estimation of fish parameters. Both the results of the experiments and the choice of the appropriate methods strongly depend on the quality and the type of a video which is used as an input data. Practical experiments have demonstrated that not all methods can produce good results for real data, whereas on synthetic data they operate satisfactorily.

Scalable Feature Selection Applications for Genome-Wide Association Studies of Complex Diseases

Personalized medicine will revolutionize our capabilities to combat disease. Working toward this goal, a fundamental task is the deciphering of geneticvariants that are predictive of complex diseases. Modern studies, in the formof genome-wide association studies (GWAS) have a���orded researchers with the opportunity to reveal new genotype-phenotype relationships through the extensive scanning of genetic variants. These studies typically contain over half a million genetic features for thousands of individuals. Examining this with methods other than univariate statistics is a challenging task requiring advanced algorithms that are scalable to the genome-wide level. In the future, next-generation sequencing studies (NGS) will contain an even larger number of common and rare variants. Machine learning-based feature selection algorithms have been shown to have the ability to e���ectively create predictive models for various genotype-phenotype relationships. This work explores the problem of selecting genetic variant subsets that are the most predictive of complex disease phenotypes through various feature selection methodologies, including ���lter, wrapper and embedded algorithms. The examined machine learning algorithms were demonstrated to not only be e���ective at predicting the disease phenotypes, but also doing so e���ciently through the use of computational shortcuts. While much of the work was able to be run on high-end desktops, some work was further extended so that it could be implemented on parallel computers helping to assure that they will also scale to the NGS data sets. Further, these studies analyzed the relationships between various feature selection methods and demonstrated the need for careful testing when selecting an algorithm. It was shown that there is no universally optimal algorithm for variant selection in GWAS, but rather methodologies need to be selected based on the desired outcome, such as the number of features to be included in the prediction model. It was also demonstrated that without proper model validation, for example using nested cross-validation, the models can result in overly-optimistic prediction accuracies and decreased generalization ability. It is through the implementation and application of machine learning methods that one can extract predictive genotype���phenotype relationships and biological insights from genetic data sets.

Fusion of Data from Quadcopter���s Inertial Measurement Unit Using Complementary Filter

A quadcopter is a helicopter with four rotors, which is mechanically simple device, but requires complex electrical control for each motor. Control system needs accurate information about quadcopter���s attitude in order to achieve stable flight. The goal of this bachelor���s thesis was to research how this information could be obtained. Literature review revealed that most of the quadcopters, whose source-code is available, use a complementary filter or some derivative of it to fuse data from a gyroscope, an accelerometer and often also a magnetometer. These sensors combined are called an Inertial Measurement Unit. This thesis focuses on calculating angles from each sensor���s data and fusing these with a complementary filter. On the basis of literature review and measurements using a quadcopter, the proposed filter provides sufficiently accurate attitude data for flight control system. However, a simple complementary filter has one significant drawback ��� it works reliably only when the quadcopter is hovering or moving at a constant speed. The reason is that an accelerometer can���t be used to measure angles accurately if linear acceleration is present. This problem can be fixed using some derivative of a complementary filter like an adaptive complementary filter or a Kalman filter, which are not covered in this thesis.

Bayesian analysis of SEIR epidemic models

This thesis concerns the analysis of epidemic models. We adopt the Bayesian paradigm and develop suitable Markov Chain Monte Carlo (MCMC) algorithms. This is done by considering an Ebola outbreak in the Democratic Republic of Congo, former Za��re, 1995 as a case of SEIR epidemic models. We model the Ebola epidemic deterministically using ODEs and stochastically through SDEs to take into account a possible bias in each compartment. Since the model has unknown parameters, we use different methods to estimate them such as least squares, maximum likelihood and MCMC. The motivation behind choosing MCMC over other existing methods in this thesis is that it has the ability to tackle complicated nonlinear problems with large number of parameters. First, in a deterministic Ebola model, we compute the likelihood function by sum of square of residuals method and estimate parameters using the LSQ and MCMC methods. We sample parameters and then use them to calculate the basic reproduction number and to study the disease-free equilibrium. From the sampled chain from the posterior, we test the convergence diagnostic and confirm the viability of the model. The results show that the Ebola model fits the observed onset data with high precision, and all the unknown model parameters are well identified. Second, we convert the ODE model into a SDE Ebola model. We compute the likelihood function using extended Kalman filter (EKF) and estimate parameters again. The motivation of using the SDE formulation here is to consider the impact of modelling errors. Moreover, the EKF approach allows us to formulate a filtered likelihood for the parameters of such a stochastic model. We use the MCMC procedure to attain the posterior distributions of the parameters of the SDE Ebola model drift and diffusion parts. In this thesis, we analyse two cases: (1) the model error covariance matrix of the dynamic noise is close to zero , i.e. only small stochasticity added into the model. The results are then similar to the ones got from deterministic Ebola model, even if methods of computing the likelihood function are different (2) the model error covariance matrix is different from zero, i.e. a considerable stochasticity is introduced into the Ebola model. This accounts for the situation where we would know that the model is not exact. As a results, we obtain parameter posteriors with larger variances. Consequently, the model predictions then show larger uncertainties, in accordance with the assumption of an incomplete model.

Forecasting the Yield Curve of Government Bonds: A Comparative Study

For the past 20 years, researchers have applied the Kalman filter to the modeling and forecasting the term structure of interest rates. Despite its impressive performance in in-sample fitting yield curves, little research has focused on the out-of-sample forecast of yield curves using the Kalman filter. The goal of this thesis is to develop a unified dynamic model based on Diebold and Li (2006) and Nelson and Siegel���s (1987) three-factor model, and estimate this dynamic model using the Kalman filter. We compare both in-sample and out-of-sample performance of our dynamic methods with various other models in the literature. We find that our dynamic model dominates existing models in medium- and long-horizon yield curve predictions. However, the dynamic model should be used with caution when forecasting short maturity yields

Une m��thode d'inf��rence bay��sienne pour les mod��les espace-��tat affines faiblement identifi��s appliqu��e �� une strat��gie d'arbitrage statistique de la dynamique de la structure �� terme des taux d'int��r��t

Th��se num��ris��e par la Division de la gestion de documents et des archives de l'Universit�� de Montr��al

Estimation of State Space Models and Stochastic Volatility

Ma th��se est compos��e de trois chapitres reli��s �� l'estimation des mod��les espace-��tat et volatilit�� stochastique. Dans le premi��re article, nous d��veloppons une proc��dure de lissage de l'��tat, avec efficacit�� computationnelle, dans un mod��le espace-��tat lin��aire et gaussien. Nous montrons comment exploiter la structure particuli��re des mod��les espace-��tat pour tirer les ��tats latents efficacement. Nous analysons l'efficacit�� computationnelle des m��thodes bas��es sur le filtre de Kalman, l'algorithme facteur de Cholesky et notre nouvelle m��thode utilisant le compte d'op��rations et d'exp��riences de calcul. Nous montrons que pour de nombreux cas importants, notre m��thode est plus efficace. Les gains sont particuli��rement grands pour les cas o�� la dimension des variables observ��es est grande ou dans les cas o�� il faut faire des tirages r��p��t��s des ��tats pour les m��mes valeurs de param��tres. Comme application, on consid��re un mod��le multivari�� de Poisson avec le temps des intensit��s variables, lequel est utilis�� pour analyser le compte de donn��es des transactions sur les march��s financi��res. Dans le deuxi��me chapitre, nous proposons une nouvelle technique pour analyser des mod��les multivari��s �� volatilit�� stochastique. La m��thode propos��e est bas��e sur le tirage efficace de la volatilit�� de son densit�� conditionnelle sachant les param��tres et les donn��es. Notre m��thodologie s'applique aux mod��les avec plusieurs types de d��pendance dans la coupe transversale. Nous pouvons modeler des matrices de corr��lation conditionnelles variant dans le temps en incorporant des facteurs dans l'��quation de rendements, o�� les facteurs sont des processus de volatilit�� stochastique ind��pendants. Nous pouvons incorporer des copules pour permettre la d��pendance conditionnelle des rendements sachant la volatilit��, permettant avoir diff��rent lois marginaux de Student avec des degr��s de libert�� sp��cifiques pour capturer l'h��t��rog��n��it�� des rendements. On tire la volatilit�� comme un bloc dans la dimension du temps et un �� la fois dans la dimension de la coupe transversale. Nous appliquons la m��thode introduite par McCausland (2012) pour obtenir une bonne approximation de la distribution conditionnelle �� posteriori de la volatilit�� d'un rendement sachant les volatilit��s d'autres rendements, les param��tres et les corr��lations dynamiques. Le mod��le est ��valu�� en utilisant des donn��es r��elles pour dix taux de change. Nous rapportons des r��sultats pour des mod��les univari��s de volatilit�� stochastique et deux mod��les multivari��s. Dans le troisi��me chapitre, nous ��valuons l'information contribu��e par des variations de volatilite r��alis��e �� l'��valuation et pr��vision de la volatilit�� quand des prix sont mesur��s avec et sans erreur. Nous utilisons de mod��les de volatilit�� stochastique. Nous consid��rons le point de vue d'un investisseur pour qui la volatilit�� est une variable latent inconnu et la volatilit�� r��alis��e est une quantit�� d'��chantillon qui contient des informations sur lui. Nous employons des m��thodes bay��siennes de Monte Carlo par cha��ne de Markov pour estimer les mod��les, qui permettent la formulation, non seulement des densit��s a posteriori de la volatilit��, mais aussi les densit��s pr��dictives de la volatilit�� future. Nous comparons les pr��visions de volatilit�� et les taux de succ��s des pr��visions qui emploient et n'emploient pas l'information contenue dans la volatilit�� r��alis��e. Cette approche se distingue de celles existantes dans la litt��rature empirique en ce sens que ces derni��res se limitent le plus souvent �� documenter la capacit�� de la volatilit�� r��alis��e �� se pr��voir �� elle-m��me. Nous pr��sentons des applications empiriques en utilisant les rendements journaliers des indices et de taux de change. Les diff��rents mod��les concurrents sont appliqu��s �� la seconde moiti�� de 2008, une p��riode marquante dans la r��cente crise financi��re.

D��veloppement et validation d'un mod��le de simulation num��rique personnalis�� �� une athl��te de plongeon

Les entra��neurs en sports acrobatiques disposent de peu d���outils permettant d���am��liorer leur compr��hension des saltos vrill��s et la performance des athl��tes. L���objectif de ce m��moire ��tait de d��velopper un environnement graphique de simulation num��rique r��aliste et utile des acrobaties a��riennes. Un mod��le compos�� de 17 segments et de 42 degr��s de libert�� a ��t�� d��velopp�� et personnalis�� �� une athl��te de plongeon. Un syst��me opto��lectronique ��chantillonn�� �� 300 Hz a permis l���acquisition de huit plongeons en situation r��elle d���entra��nement. La cin��matique articulaire reconstruite avec un filtre de Kalman ��tendu a ��t�� utilis��e comme entr��e du mod��le. Des erreurs quadratiques moyennes de 20�� (salto) et de 9�� (vrille) entre les performances simul��es et r��elles ont permis de valider le mod��le. Enfin, une formation bas��e sur le simulateur a ��t�� offerte �� 14 entra��neurs en sports acrobatiques. Une augmentation moyenne de 11 % des r��sultats aux questionnaires post-test a permis de constater le potentiel p��dagogique de l���outil pour la formation.

Improving sampling, optimization and feature extraction in Boltzmann machines

L���apprentissage supervis�� de r��seaux hi��rarchiques �� grande ��chelle conna��t pr��sentement un succ��s fulgurant. Malgr�� cette effervescence, l���apprentissage non-supervis�� repr��sente toujours, selon plusieurs chercheurs, un ��l��ment cl�� de l���Intelligence Artificielle, o�� les agents doivent apprendre �� partir d���un nombre potentiellement limit�� de donn��es. Cette th��se s���inscrit dans cette pens��e et aborde divers sujets de recherche li��s au probl��me d���estimation de densit�� par l���entremise des machines de Boltzmann (BM), mod��les graphiques probabilistes au coeur de l���apprentissage profond. Nos contributions touchent les domaines de l�����chantillonnage, l���estimation de fonctions de partition, l���optimisation ainsi que l���apprentissage de repr��sentations invariantes. Cette th��se d��bute par l���exposition d���un nouvel algorithme d'��chantillonnage adaptatif, qui ajuste (de fa ��con automatique) la temp��rature des cha��nes de Markov sous simulation, afin de maintenir une vitesse de convergence ��lev��e tout au long de l���apprentissage. Lorsqu���utilis�� dans le contexte de l���apprentissage par maximum de vraisemblance stochastique (SML), notre algorithme engendre une robustesse accrue face �� la s��lection du taux d���apprentissage, ainsi qu���une meilleure vitesse de convergence. Nos r��sultats sont pr��sent ��es dans le domaine des BMs, mais la m��thode est g��n��rale et applicable �� l���apprentissage de tout mod��le probabiliste exploitant l�����chantillonnage par cha��nes de Markov. Tandis que le gradient du maximum de vraisemblance peut-��tre approxim�� par ��chantillonnage, l�����valuation de la log-vraisemblance n��cessite un estim�� de la fonction de partition. Contrairement aux approches traditionnelles qui consid��rent un mod��le donn�� comme une bo��te noire, nous proposons plut��t d���exploiter la dynamique de l���apprentissage en estimant les changements successifs de log-partition encourus �� chaque mise �� jour des param��tres. Le probl��me d���estimation est reformul�� comme un probl��me d���inf��rence similaire au filtre de Kalman, mais sur un graphe bi-dimensionnel, o�� les dimensions correspondent aux axes du temps et au param��tre de temp��rature. Sur le th��me de l���optimisation, nous pr��sentons ��galement un algorithme permettant d���appliquer, de mani��re efficace, le gradient naturel �� des machines de Boltzmann comportant des milliers d���unit��s. Jusqu����� pr��sent, son adoption ��tait limit��e par son haut co��t computationel ainsi que sa demande en m��moire. Notre algorithme, Metric-Free Natural Gradient (MFNG), permet d�����viter le calcul explicite de la matrice d���information de Fisher (et son inverse) en exploitant un solveur lin��aire combin�� �� un produit matrice-vecteur efficace. L���algorithme est prometteur: en terme du nombre d�����valuations de fonctions, MFNG converge plus rapidement que SML. Son impl��mentation demeure malheureusement inefficace en temps de calcul. Ces travaux explorent ��galement les m��canismes sous-jacents �� l���apprentissage de repr��sentations invariantes. �� cette fin, nous utilisons la famille de machines de Boltzmann restreintes ���spike & slab��� (ssRBM), que nous modifions afin de pouvoir mod��liser des distributions binaires et parcimonieuses. Les variables latentes binaires de la ssRBM peuvent ��tre rendues invariantes �� un sous-espace vectoriel, en associant �� chacune d���elles, un vecteur de variables latentes continues (d��nomm��es ���slabs���). Ceci se traduit par une invariance accrue au niveau de la repr��sentation et un meilleur taux de classification lorsque peu de donn��es ��tiquet��es sont disponibles. Nous terminons cette th��se sur un sujet ambitieux: l���apprentissage de repr��sentations pouvant s��parer les facteurs de variations pr��sents dans le signal d���entr��e. Nous proposons une solution �� base de ssRBM bilin��aire (avec deux groupes de facteurs latents) et formulons le probl��me comme l���un de ���pooling��� dans des sous-espaces vectoriels compl��mentaires.

Analyse cin��matique 3D de la coordination des articulations de l'��paule par l'entremise du rythme scapulo-hum��ral

L'��paule est souvent affect��e par des troubles musculo-squelettiques. Toutefois, leur ��valuation est limit��e �� des mesures qualitatives qui nuisent �� la sp��cificit�� et justesse du diagnostic. L'analyse de mouvement tridimensionnel pourrait compl��menter le traitement conventionnel �� l'aide de mesures quantitatives fonctionnelles. L'interaction entre les articulations de l'��paule est estim��e par le rythme scapulo-hum��ral, mais la variabilit�� prononc��e qu'il affiche nuit �� son utilisation clinique. Ainsi, l'objectif g��n��ral de cette th��se ��tait de r��duire la variabilit�� de la mesure du rythme scapulo-hum��ral. L'effet de la m��thode de calcul du rythme scapulo-hum��ral et des conditions d'ex��cution du mouvement (rotation axiale du bras, charge, vitesse, activit�� musculaire) ont ��t�� test��es. La cin��matique des articulations de l'��paule a ��t�� calcul�� par cha��ne cin��matique et filtre de Kalman ��tendu sur des sujets sains avec un syst��me opto��lectronique. La m��thode usuelle de calcul du rythme scapulo-hum��ral extrait les angles d'��l��vation gl��no-hum��rale et de rotation lat��rale scapulo-thoracique. Puisque ces angles ne sont pas co-planaires au thorax, leur somme ne correspond pas �� l'angle d'��l��vation du bras. Une nouvelle approche de contribution articulaire incluant toutes les rotations de chaque articulation est propos��e et compar��e �� la m��thode usuelle. La m��thode usuelle surestimait syst��matiquement la contribution gl��no-hum��rale par rapport �� la m��thode propos��e. Ce nouveau calcul du rythme scapulo-hum��ral permet une ��valuation fonctionnelle dynamique de l'��paule et r��duit la variabilit�� inter-sujets. La comparaison d'exercices de r��adaptation du supra-��pineux contrastant la rotation axiale du bras a ��t�� r��alis��e, ainsi que l'effet d'ajouter une charge externe. L'exercice ��full-can�� augmentait le rythme scapulo-hum��ral et la contribution gl��no-hum��rale ce qui concorde avec la fonction du supra-��pineux. Au contraire, l'exercice ��empty-can�� augmentait la contribution scapulo-thoracique ce qui est associ�� �� une compensation pour ��viter la contribution gl��no-hum��rale. L'utilisation de charge externe lors de la r��adaptation du supra-��pineux semble justifi��e par un rythme scapulo-hum��ral similaire et une ��l��vation gl��no-hum��rale sup��rieure. Le mouvement de l'��paule est souvent mesur�� ou ��valu�� en condition statique ou dynamique et passive ou active. Cependant, l'effet de ces conditions sur la coordination articulaire demeure incertain. La comparaison des ces conditions r��v��lait des diff��rences significatives qui montrent l'importance de consid��rer les conditions de mouvement pour l'acquisition ou la comparaison des donn��es.

Modeling and Analysis of Some Heavy Tailed Time Series

The thesis has covered various aspects of modeling and analysis of finite mean time series with symmetric stable distributed innovations. Time series analysis based on Box and Jenkins methods are the most popular approaches where the models are linear and errors are Gaussian. We highlighted the limitations of classical time series analysis tools and explored some generalized tools and organized the approach parallel to the classical set up. In the present thesis we mainly studied the estimation and prediction of signal plus noise model. Here we assumed the signal and noise follow some models with symmetric stable innovations.We start the thesis with some motivating examples and application areas of alpha stable time series models. Classical time series analysis and corresponding theories based on finite variance models are extensively discussed in second chapter. We also surveyed the existing theories and methods correspond to infinite variance models in the same chapter. We present a linear filtering method for computing the filter weights assigned to the observation for estimating unobserved signal under general noisy environment in third chapter. Here we consider both the signal and the noise as stationary processes with infinite variance innovations. We derived semi infinite, double infinite and asymmetric signal extraction filters based on minimum dispersion criteria. Finite length filters based on Kalman-Levy filters are developed and identified the pattern of the filter weights. Simulation studies show that the proposed methods are competent enough in signal extraction for processes with infinite variance.Parameter estimation of autoregressive signals observed in a symmetric stable noise environment is discussed in fourth chapter. Here we used higher order Yule-Walker type estimation using auto-covariation function and exemplify the methods by simulation and application to Sea surface temperature data. We increased the number of Yule-Walker equations and proposed a ordinary least square estimate to the autoregressive parameters. Singularity problem of the auto-covariation matrix is addressed and derived a modified version of the Generalized Yule-Walker method using singular value decomposition.In fifth chapter of the thesis we introduced partial covariation function as a tool for stable time series analysis where covariance or partial covariance is ill defined. Asymptotic results of the partial auto-covariation is studied and its application in model identification of stable auto-regressive models are discussed. We generalize the Durbin-Levinson algorithm to include infinite variance models in terms of partial auto-covariation function and introduce a new information criteria for consistent order estimation of stable autoregressive model.In chapter six we explore the application of the techniques discussed in the previous chapter in signal processing. Frequency estimation of sinusoidal signal observed in symmetric stable noisy environment is discussed in this context. Here we introduced a parametric spectrum analysis and frequency estimate using power transfer function. Estimate of the power transfer function is obtained using the modified generalized Yule-Walker approach. Another important problem in statistical signal processing is to identify the number of sinusoidal components in an observed signal. We used a modified version of the proposed information criteria for this purpose.

Underwater Target Localization, Tracking and Classification

Underwater target localization and tracking attracts tremendous research interest due to various impediments to the estimation task caused by the noisy ocean environment. This thesis envisages the implementation of a prototype automated system for underwater target localization, tracking and classification using passive listening buoy systems and target identification techniques. An autonomous three buoy system has been developed and field trials have been conducted successfully. Inaccuracies in the localization results, due to changes in the environmental parameters, measurement errors and theoretical approximations are refined using the Kalman filter approach. Simulation studies have been conducted for the tracking of targets with different scenarios even under maneuvering situations. This system can as well be used for classifying the unknown targets by extracting the features of the noise emanations from the targets.