In this thesis, rotation, scale and translation (RST) invariant automatic target recognition (ATR) for satellite imagery is presented. Template matching is used to realize the target recognition. However, unlike most of the studies of template matching in the literature, RST invariance is required in our problem, since most of the time we will have only a small number of templates of each target, while the targets to be recognized in the scenes will have various orientations, scaling and translations. RST invariance is studied in detail and implemented with some of the competing methods in the literature, such as Fourier-Mellin transform and bipectrum combined with log-polar mapping. Phase correlation and normalized cross-correlation are used as similarity metrics. Encountered drawbacks were overcome with additional operations and modifications of the algorithms. ATR using reconstruction of the target image with respect to the template, based on bispectrum, log-polar mapping and phase correlation outperformed the other methods and successful recognition was realized for various target types, especially for targets on relatively simpler backgrounds, i.e. containing little or no other objects.

Disertacijoje gautas Dirichlė L funkcijų modifikuotosios Melino transformacijos pratęsimas į visą kompleksinę plokštumą. / In the thesis a meremorphic continuation of Dirichlet L- functions to the whole complex plane have been obtained.

In the thesis moromorphic continuation of modified Mellin transforms of Dirichlet L-functions to the whole complex plane have been obtained. / Disertacijoje gauta modifikuotosios Melino transformacijos L- funkcijai meromorfinis pratęsimas į visą kompleksinę plokštumą.

We give a rather self-contained introduction to the rare pion to dielectron decay which in nontrivial leading order is given by a QED triangle loop. We work within the dispersive framework where the imaginary part of the amplitude is obtained via the Cutkosky rules. We derive these rules in detail. Using the twofold Mellin-Barnes representation for the pion transition form factor, we derive a simple expression for the branching ratio B(π0 <img src="http://www.diva-portal.org/cgi-bin/mimetex.cgi?%5Crightarrow" /> e+e-) which we then test for various models. In particular a more recent form factor derived from a Lagrangian for light pseudoscalars and vector mesons inspired by effective field theories. Comparison with the KTeV experiment at Fermilab is made and we find that we are more than 3σ below the KTeV experiment for some of the form factors. This is in agreement with other theoretical models, such as the Vector Meson Dominance model and the quark-loop model within the constituent-quark framework. But we also find that we can be in agreement with KTeV if we explore some freedom of the form factor not fixed by the low-energy Lagrangian.

On s’intéresse dans cette thèse à l’étude des fonctions thêta intervenant dans la preuve de l’équation fonctionnelle des fonctions L de Dirichlet. En particulier, on adapte certains résultats obtenus dans le cadre des fonctions L au cas des fonctions thêta. S. Chowla a conjecturé que les fonctions L de Dirichlet associées à des caractères χ primitifs ne doivent pas s’annuler au point central de leur équation fonctionnelle. De façon analogue, il est conjecturé que les fonctions thêta ne s'annulent pas au point 1. Dans le but de prouver cette conjecture pour beaucoup de caractères, on étudie les moments de fonctions thêta dans plusieurs familles. On se focalise sur deux familles importantes. La première considérée est l’ensemble des caractères de Dirichlet modulo p où p est un nombre premier. On prouve des formules asymptotiques pour les moments d'ordre 2 et 4 en se ramenant à des problèmes de nature diophantienne. La seconde famille considérée est celle des caractères primitifs et quadratiques associés à des discriminants fondamentaux d inférieurs à une certaine borne fixée. On donne une formule asymptotique pour le premier moment et une majoration pour le moment d'ordre 2 en utilisant des techniques de transformée de Mellin ainsi que des estimations sur les sommes de caractères. Dans les deux cas, on en déduit des résultats de non-annulation des fonctions thêta. On propose également un algorithme qui, pour beaucoup de caractères, se révèle en pratique efficace pour prouver la non-annulation sur l'axe réel positif des fonctions thêta ce qui entraîne la non-annulation sur le même axe des fonctions L associées. / In this thesis, we focus on the study of theta functions involved in the proof of the functional equation of Dirichlet L- functions. In particular, we adapt some results obtained for L-functions to the case of theta functions. S. Chowla conjectured that Dirichlet L- functions associated to primitive characters χ don’t vanish at the central point of their functional equation. In a similar way to Chowla’s conjecture, it is conjectured that theta functions don't vanish at the central point of their functional equation for each primitive character. With the aim of proving this conjecture for a lot of characters, we study moments of theta functions in various families. We concentrate on two important families. The first one which we consider is the family of all Dirichlet characters modulo p where p is a prime number. In this case, we prove asymptotic formulae for the second and fourth moment of theta functions using diophantine techniques. The second family which we consider is the set of primitive quadratic characters associated to a fundamental discriminant less than a fixed bound. We give an asymptotic formula for the first moment and an upper bound for the second moment using techniques of Mellin transforms and estimation of character sums. In both cases, we deduce some results of non-vanishing. We also give an algorithm which, in practice, works well for a lot of characters to prove the non-vanishing of theta functions on the positive real axis. In this case, this implies in particular that the associated L-functions don’t vanish on the same axis.

Title: Options under Stable Laws. Author: Andrea Karlová Department: Department of Probability and Mathematical Statistics Supervisor: Doc. Petr Volf, CSc. Abstract: Stable laws play a central role in the convergence problems of sums of independent random variables. In general, densities of stable laws are represented by special functions, and expressions via elementary functions are known only for a very few special cases. The convenient tool for investigating the properties of stable laws is provided by integral transformations. In particular, the Fourier transform and Mellin transform are greatly useful methods. We first discuss the Fourier transform and we give overview on the known results. Next we consider the Mellin transform and its applicability on the problem of the product of two independent random variables. We establish the density of the product of two independent stable random variables, discuss the properties of this product den- sity and give its representation in terms of power series and Fox's H-functions. The fourth chapter of this thesis is focused on the application of stable laws into option pricing. In particular, we generalize the model introduced by Louise Bachelier into stable laws. We establish the option pricing formulas under this model, which we refer to as the Lévy Flight...

<div>The Subword Complexity of a character string refers to the number of distinct substrings of any length that occur as contiguous patterns in the string. The kth Subword Complexity in particular, refers to the number of distinct substrings of length k in a string of length n. In this work, we evaluate the expected value and the second factorial moment of the kth Subword Complexity for the binary strings over memory-less sources. We first take a combinatorial approach to derive a probability generating function for the number of occurrences of patterns in strings of finite length. This enables us to have an exact expression for the two moments in terms of patterns' auto-correlation and correlation polynomials. We then investigate the asymptotic behavior for values of k=a log n. In the proof, we compare the distribution of the kth Subword Complexity of binary strings to the distribution of distinct prefixes of independent strings stored in a trie. </div><div>The methodology that we use involves complex analysis, analytical poissonization and depoissonization, the Mellin transform, and saddle point analysis.</div>

[pt] Ao mesmo tempo em que a autonomia de robôs pessoais e domésticos aumenta, cresce a necessidade de interação dos mesmos com o ambiente. A interação mais básica de um robô com o ambiente é feita pela percepção deste e sua navegação. Para uma série de aplicações não é prático prover modelos geométricos válidos do ambiente a um robô antes de seu uso. O robô necessita, então, criar estes modelos enquanto se movimenta e percebe o meio em que está inserido através de sensores. Ao mesmo tempo é necessário minimizar a complexidade requerida quanto a hardware e sensores utilizados. No presente trabalho, um algoritmo iterativo baseado em entropia é proposto para planejar uma estratégia de exploração visual, permitindo a construção eficaz de um modelo em grafo do ambiente. O algoritmo se baseia na determinação da informação presente em sub-regiões de uma imagem panorâmica 2-D da localização atual do robô obtida com uma câmera fixa sobre o mesmo. Utilizando a métrica de entropia baseada na Teoria da Informação de Shannon, o algoritmo determina nós potenciais para os quais deve se prosseguir a exploração. Através de procedimento de Visual Tracking, em conjunto com a técnica SIFT (Scale Invariant Feature Transform), o algoritmo auxilia a navegação do robô para cada nó novo, onde o processo é repetido. Um procedimento baseado em transformações invariáveis a determinadas variações espaciais (desenvolvidas a partir de Fourier e Mellin) é utilizado para auxiliar o processo de guiar o robô para nós já conhecidos. Também é proposto um método baseado na técnica SIFT. Os processos relativos à obtenção de imagens, avaliação, criação do grafo, e prosseguimento dos passos citados continua até que o robô tenha mapeado o ambiente com nível pré-especificado de detalhes. O conjunto de nós e imagens obtidos são combinados de modo a se criar um modelo em grafo do ambiente. Seguindo os caminhos, nó a nó, um robô pode navegar pelo ambiente já explorado. O método é particularmente adequado para ambientes planos. As componentes do algoritmo proposto foram desenvolvidas e testadas no presente trabalho. Resultados experimentais mostrando a eficácia dos métodos propostos são apresentados. / [en] As the autonomy of personal service robotic systems increases so has their need to interact with their environment. The most basic interaction a robotic agent may have with its environment is to sense and navigate through it. For many applications it is not usually practical to provide robots in advance with valid geometric models of their environment. The robot will need to create these models by moving around and sensing the environment, while minimizing the complexity of the required sensing hardware. This work proposes an entropy-based iterative algorithm to plan the robot´s visual exploration strategy, enabling it to most efficiently build a graph model of its environment. The algorithm is based on determining the information present in sub-regions of a 2- D panoramic image of the environment from the robot´s current location using a single camera fixed on the mobile robot. Using a metric based on Shannon s information theory, the algorithm determines potential locations of nodes from which to further image the environment. Using a Visual Tracking process based on SIFT (Scale Invariant Feature Transform), the algorithm helps navigate the robot to each new node, where the imaging process is repeated. An invariant transform (based on Fourier and Mellin) and tracking process is used to guide the robot back to a previous node. Also, an SIFT based method is proposed to accomplish such task. This imaging, evaluation, branching and retracing its steps continues until the robot has mapped the environment to a pre-specified level of detail. The set of nodes and the images taken at each node are combined into a graph to model the environment. By tracing its path from node to node, a service robot can navigate around its environment. This method is particularly well suited for flat-floored environments. The components of the proposed algorithm were developed and tested. Experimental results show the effectiveness of the proposed methods.

Image registration, the process of finding the transformation that best maps one image to another, is an important tool in document image processing. Having properly-aligned microfilm images can help in manual and automated content extraction, zoning, and batch compression of images. An image registration algorithm is presented that quickly identifies the global affine transformation (rotation, scale, translation and/or shear) that maps one tabular document image to another, using the Fourier-Mellin Transform. Each component of the affine transform is recovered independantly from the others, dramatically reducing the parameter space of the problem, and improving upon standard Fourier-Mellin Image Registration (FMIR), which only directly separates translation from the other components. FMIR is also extended to handle shear, as well as different scale factors for each document axis. This registration method deals with all transform components in a uniform way, by working in the frequency domain. Registration is limited to foreground pixels (the document form and printed text) through the introduction of a novel, locally adaptive foreground-background segmentation algorithm, based on the median filter. The background removal algorithm is also demonstrated as a useful tool to remove ambient signal noise during correlation. Common problems with FMIR are eliminated by background removal, meaning that apodization (tapering down to zero at the edge of the image) is not needed for accurate recovery of the rotation parameter, allowing the entire image to be used for registration. An effective new optimization to the median filter is presented. Rotation and scale parameter detection is less susceptible to problems arising from the non-commutativity of rotation and "tiling" (periodicity) than for standard FMIR, because only the regions of the frequency domain directly corresponding to tabular features are used in registration. An original method is also presented for automatically obtaining blank document templates from a set of registered document images, by computing the "pointwise median" of a set of registered documents. Finally, registration is demonstrated as an effective tool for predictive image compression. The presented registration algorithm is reliable and robust, and handles a wider range of transformation types than most document image registration systems (which typically only perform deskewing).

La fonction de Möbius est définie par$$\mu(n)= \begin{cases} 1 & \textit{si $n=1$},\\ (-1)^k& \textit{si n est le produit de k nombres premiers distincts,}\\ 0 & \textit{si n contient un facteur carré. } \end{cases}$$Nous avons démontré que pour $x \ge \exp( 10^9) $ et $h=x^{1-\frac{1}{16000}}$, il existe dans chaque intervalle $[x-h,x]$ des entiers $n_1$ avec $\mu(n_1)=1$ et des entiers $n_2$ avec $\mu(n_2)=-1$.\\Ce résultat est une conséquence d'un résultat plus général.\\Pour $x \ge \exp(4\times 10^6)$, $\frac{1}{\sqrt{\log x}} \le \theta \le \frac{1}{2000}$, $h=x^{1-\theta}$ et $Q=(x/h)^{\frac{1}{20}}$, nous avons \\$$\sum_{q \leq Q} \log(Q/q)\sum_{\chi mod q}^*\left| \sum_{x.-h\le n \le x} \mu(n) \chi(n) \right| \leq 10^{20} h \theta \log(x) \exp( \frac{-1}{300 \theta}); $$la somme $\sum^*$ portant sur les caractères primitifs sauf l'éventuel caractère exceptionnel.\\Et en particulier pour $x \ge \exp( 10^9)$,$$ \left | \sum_{x.-x^{1-\frac{1}{16000}}\le n \le x} \mu(n) \right | \le \frac{1}{100} x^{1-\frac{1}{16000}}.\\$$ / The Möbius function is defined by$$\mu(n)= \begin{cases} 1 & \textit{if $n=1$},\\ (-1)^k& \textit{if n is a product of k distinct prime numbers,}\\ 0 & \textit{if n contains a square factor. } \end{cases}$$We demonstrate that for $x \ge \exp( 10^9) $ and $h=x^{1-\frac{1}{16000}}$, it exists in each interval $[x-h,x]$ integers $n_1$ with $\mu(n_1)=1$ and integers $n_2$ with $\mu(n_2)=-1$.\\This result is a consequence of a more general result. \\For $x \ge \exp(4\times 10^6)$, $\frac{1}{\sqrt{\log x}} \le \theta \le \frac{1}{2000}$, $h=x^{1-\theta}$ et $Q=(x/h)^{\frac{1}{20}}$, we have \\ $$\sum_{q \leq Q} \log(Q/q)\sum_{\chi mod q}^*\left| \sum_{x-h \le n \le x} \mu(n) \chi(n) \right| \leq 10^{20} h \theta \log(x) \exp( \frac{-1}{300 \theta}); $$the sum $\sum^*$ relating to primitive characters except for possible exceptional character.\\And in particular for $x \ge \exp( 10^9)$,$$\left | \sum_{x-.x^{1-\frac{1}{16000}}\le n \le x} \mu(n) \right | \le \frac{1}{100} x^{1-\frac{1}{16000}}.$$