Multi-modal image fusion is an active research area with many civilian and military applications. Fusion is defined as strategic combination of information collected by various sensors from different locations or different types in order to obtain a better understanding of an observed scene or situation. Fusion of multi-modal images cannot be completed unless these two modalities are spatially aligned. In this research, I consider two important problems. Multi-modal, multi-perspective image registration and decision level fusion of multi-modal images. In particular, LiDAR and visual imagery. Multi-modal image registration is a difficult task due to the different semantic interpretation of features extracted from each modality. This problem is decoupled into three sub-problems. The first step is identification and extraction of common features. The second step is the determination of corresponding points. The third step consists of determining the registration transformation parameters. Traditional registration methods use low level features such as lines and corners. Using these features require an extensive optimization search in order to determine the corresponding points. Many methods use global positioning systems (GPS), and a calibrated camera in order to obtain an initial estimate of the camera parameters. The advantages of our work over the previous works are the following. First, I used high level-features, which significantly reduce the search space for the optimization process. Second, the determination of corresponding points is modeled as an assignment problem between a small numbers of objects. On the other side, fusing LiDAR and visual images is beneficial, due to the different and rich characteristics of both modalities. LiDAR data contain 3D information, while images contain visual information. Developing a fusion technique that uses the characteristics of both modalities is very important. I establish a decision-level fusion technique using manifold models.

This thesis presents a new output passive filter for voltage source inverter applications which is based on a shunt connected single tuned filter topology. The proposed circuit has the advantage of tracing harmonic components wherever its location in the frequency spectrum. The change in the harmonic location might be as a result of a change in the inverter operating frequency. Also, the proposed filter achieves harmonic reduction close to the traditional single tuned passive filter. In order to show the superiority of the proposed model, a comparison is introduced with other self tuning harmonic filters showing merits and drawbacks of each technique. The proposed circuit (when integrated in square wave inverter) has also shown a tremendous reduction in the switching losses in comparison with high frequency Pulse Width Modulation inverter. Mathematical analyses showing the design of the proposed filter together with extensive simulation results to verify the design are also introduced. The practical implementation of the system is presented and the results show excellent agreement with the theory and simulation. In order to appreciate the proposed filter a new method for classifying passive power filters is introduced. The review includes a comparison of these configurations showing their merit and drawbacks.

The thesis addresses the dynamic state estimation problem for the field of robotics, particularly for unmanned aerial vehicles (UAVs). Based on data collected from an UAV, we design several probabilistic models for estimation of its state (mainly speed and rotation angles), including the configurations where one of the sensors is not available. We use Kalman filter and Particle filter and focus on learning the model parameters using EM algorithm. The EM algorithm is then adjusted with respect to non-Gaussian density of some sensor errors and modified using model complexity penalization terms for better generalization. We implement these methods in MATLAB environment and evaluate on separate datasets. We also analyze data from a ground robot and use our implementation of Particle filter for estimation of its position. Powered by TCPDF (www.tcpdf.org)

Abstract This diploma thesis focuses on processes that structure desmid communities. Phylogenetic structure of communities helps to reveal those processes. Phylogenetic structures links ecology of communities and evolutionary history of species in those communities. Totally I analysed 89 desmid communities. Most of them were phylogenetically structured, which agrees with the hypothesis that communities are structured by classical niche-related processes, such as competition and environmental filtering. Usage of environmental information showed the influence of pH on the community structure. Low pH acts as a strong environmental filter. Under the influence of this filter communities are composed of closely relates species. In the absence of this filter communities are structured by competition. Localities with higher pH host communities of distantly related species - phylogenetically overdispersed. There was found no influence of conductivity and geographical distance on phylogenetic structure of desmids communities. Key words: phylogenetic structure, desmids, Desmidiales, environmental filtering, pH

The aim of this work is to explore recommender systems for prediction user's future film ratings according to their previous ratings. Author will describe available algorithms and compare their results with his own algorithm. The goal is to find algorithm with the highest prediction accuracy and find the most important parameters for a good predictions.

Nanoplasmonics has provided a way to control light with extremely high precision, into nanoscale volumes. In many circumstances, the nanoplasmonic devices which can be realised are fabricated using processing techniques which rely on planar technologies. This thesis provides a general method to make nanoplasmonic devices on a flexible membrane structure, which can be free standing, extremely thin (less than the wavelength of visible light), but retains the ability to be manipulated without loss of optical function. These devices are very pliant and conformable. Flexibility allows the integration of nanoplasmonic devices into many new applications where curved surfaces or the ability to conform to another object is required, as well as providing a route for post-fabrication tunability. Two specific applications are considered: lab-on-fibre technology and surface enhanced Raman spectroscopy. Lab-on-fibre technologies have been advancing the ability to miniaturise experiments which would normally require a whole laboratory. Fabricating a membrane and then later applying it to the fibre decouples the choice of fibre from the design of the device. Surface enhanced Raman spectroscopy is a powerful diagnostic tool which can uniquely identify an optical fingerprint of different molecules. The technique has been held back from widespread clinical adoption because of the difficulty of reproducibility of the substrates used. A repeatable and reliable rigid substrate is demonstrated, which can identify the concentration of a three component mixture of physiologically relevant biomolecules. This same design is then shown in a flexible form factor, which is applied to a non-planar landscape where it can identify the locations where a molecule of interest has been deposited. This thesis details the development of the fabrication protocol, the construction of experimental apparatus for characterisation, and the use of numerical modelling to advance the flexible nanoplasmonic membrane platform.

Extraire des mesures sur des surfaces est un problème dans de nombreux domaines de recherche. L'archaïsme de certains systèmes ou la cherté d'appareils perfectionnés empêchent l'extraction rapide et robuste de ces paramètres. Pourtant, ils sont essentiels dans de nombreux domaines, comme les paramètres de rugosité qui interviennent dans de nombreux phénomènes physiques ou les valeurs dendrométriques pour l'étude de la bio-diversité. En parallèle, l'utilisation et la production de contenus 3D se développent considérablement ces dernières années dans des domaines très divers. Le but de cette thèse est donc d'utiliser toutes ces innovations dans le domaine de la reconstruction 3D et de les appliquer à la mesure de paramètres de surfaces. Pour cela, il est nécessaire de créer une chaîne complète de reconstruction 3D, n'utilisant que des images prises sans contrainte dans le but d'être accessible au plus grand nombre. Dans cette chaîne nous utilisons des algorithmes de stéréo-vision robustes produisant, pour chaque couple d'images, un nuage de points. Après le passage de ces nuages dans un référentiel commun, une étape de filtrage des points 3D et de suppression des redondances est nécessaire. Une étape de lissage permet d'obtenir le nuage final. Pour justifier des bons résultats obtenus, une étape de validation nous a permis de vérifier et d'étudier la robustesse de la chaîne de traitements développée. Enfin, les paramètres de rugosités et dendrométriques seront extraits. Nous étudierons dans ces deux cas, comment extraire ces informations et leurs utilisations / Extracting measures on surfaces is a problem in many areas of research. The archaism of some systems or the costliness of sophisticated devices prevent the fast and robust extraction of these parameters. Yet these measures are essential in many areas, such as roughness parameters involved in many physical phenomena or dendrometric values for the study of biodiversity. In parallel, the use and production of 3D content has grown dramatically this past year in very diverse domains. The purpose of this thesis is to use these innovations in the context of surfaces parameter measurements. It is necessary to create a complete chain of 3D reconstruction, using pictures taken without constraint, in order to be open to as many people. This chain uses robust stereo-vision algorithms in order to produce a point cloud for each pair of images. After the generation of these point cloud in the same geometric frame, a filtering step of 3D points and a deletion step of redundancies are necessary and a smoothing step allows us to obtain the final point cloud. To reveal the good results, a validation step has enabled us to verify and investigate the robustness of the developed chain. The roughness and dendrometric parameters are finally extracted. We will study in both cases, how to extract this information and their uses

Median filtering is a non-linear filtering technique which is effective in removing impulsive noise from data. In this thesis, directional median filtering has been implemented using cumulative histogram of samples in several directions. Different methods to implement directional median filtering have been proposed. The filtered images are smoothed along the direction of the filtering window. All implementations aimed to generate outputs in the least amount of time, while reducing the resource utilization on hardware. The implementation methods were designed for Xilinx Virtex 5 FPGA devices but were also attempted on Spartan 3E. The proposed methods used less than 30% of the resources on Virtex 5 FPGA but the resource utilization on Spartan 3E exceeded the number of available resources. After an initial delay, methods 1 and 2 generate a new output for every 5 clock cycles while method 3 generates an output for every 1.5 clock cycles.

In battery management systems, the main figure of merit is the battery's SOC, typically obtained from voltage and current measurements. Present estimation methods use simplified battery models that do not fully capture the electrical characteristics of the battery, which are useful for system design. This thesis studied SOC estimation for a lithium-ion battery using a nonlinear, electrical-circuit battery model that better describes the electrical characteristics of the battery. The extended Kalman filter, unscented Kalman filter, third-order and fifth-order cubature Kalman filter, and the statistically linearized filter were tested on their ability to estimate the SOC through numerical simulation. Their performances were compared based on their root-mean-square error over one hundred Monte Carlo runs as well as the time they took to complete those runs. The results show that the extended Kalman filter is a good choice for estimating the SOC of a lithium-ion battery.

This report describes a distributed approach to social filtering based on the agent metaphor. Firstly, previous approaches are described, such as cognitive filtering and social filtering. Then a couple of previously implemented systems are presented and then a new system design is proposed. The main goal is to give the requirements and design of an agent-based system that recommends web-documents. The presented approach combines cognitive and social filtering to get the advantages from both techniques. Finally, a prototype implementation called WebCondor is described and results of testing the system are reported and discussed.