The main goal of this thesis is to develop real-time computer vision algorithms in order to detect and to track targets in uncertain complex environments purely based on a visual sensor. Two major subjects addressed by this work are: 1. The development of fast and robust image segmentation algorithms that are able to search and automatically detect targets in a given image. 2. The development of sound filtering algorithms to reduce the effects of noise in signals from the image processing. The main constraint of this research is that the algorithms should work in real-time with limited computing power on an onboard computer in an aircraft. In particular, we focus on contour tracking which tracks the outline of the target represented by contours in the image plane. This thesis is concerned with three specific categories, namely image segmentation, shape modeling, and signal filtering. We have designed image segmentation algorithms based on geometric active contours implemented via level set methods. Geometric active contours are deformable contours that automatically track the outlines of objects in images. In this approach, the contour in the image plane is represented as the zero-level set of a higher dimensional function. (One example of the higher dimensional function is a three-dimensional surface for a two-dimensional contour.) This approach handles the topological changes (e.g., merging, splitting) of the contour naturally. Although geometric active contours prevail in many fields of computer vision, they suffer from the high computational costs associated with level set methods. Therefore, simplified versions of level set methods such as fast marching methods are often used in problems of real-time visual tracking. This thesis presents the development of a fast and robust segmentation algorithm based on up-to-date extensions of level set methods and geometric active contours, namely a fast implementation of Chan-Vese's (active contour) model (FICVM). The shape prior is a useful cue in the recognition of the true target. For the contour tracker, the outline of the target can be easily disrupted by noise. In geometric active contours, to cope with deviations from the true outline of the target, a higher dimensional function is constructed based on the shape prior, and the contour tracks the outline of an object by considering the difference between the higher dimensional functions obtained from the shape prior and from a measurement in a given image. The higher dimensional function is often a distance map which requires high computational costs for construction. This thesis focuses on the extraction of shape information from only the zero-level set of the higher dimensional function. This strategy compensates for inaccuracies in the calculation of the shape difference that occur when a simplified higher dimensional function is used. This is named as contour-based shape modeling. Filtering is an essential element in tracking problems because of the presence of noise in system models and measurements. The well-known Kalman filter provides an exact solution only for problems which have linear models and Gaussian distributions (linear/Gaussian problems). For nonlinear/non-Gaussian problems, particle filters have received much attention in recent years. Particle filtering is useful in the approximation of complicated posterior probability distribution functions. However, the computational burden of particle filtering prevents it from performing at full capacity in real-time applications. This thesis concentrates on improving the processing time of particle filtering for real-time applications. In principle, we follow the particle filter in the geometric active contour framework. This thesis proposes an advanced blob tracking scheme in which a blob contains shape prior information of the target. This scheme simplifies the sampling process and quickly suggests the samples which have a high probability of being the target. Only for these samples is the contour tracking algorithm applied to obtain a more detailed state estimate. Curve evolution in the contour tracking is realized by the FICVM. The dissimilarity measure is calculated by the contour based shape modeling method and the shape prior is updated when it satisfies certain conditions. The new particle filter is applied to the problems of low contrast and severe daylight conditions, to cluttered environments, and to the appearing/disappearing target tracking. We have also demonstrated the utility of the filtering algorithm for multiple target tracking in the presence of occlusions. This thesis presents several test results from simulations and flight tests. In these tests, the proposed algorithms demonstrated promising results in varied situations of tracking.

Made available in DSpace on 2014-12-17T15:54:51Z (GMT). No. of bitstreams: 1 BrunoCOS.pdf: 1429649 bytes, checksum: 1029e4832e6e3f29731061ed96530d73 (MD5) Previous issue date: 2009-02-03 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior / In the current climate of global public health there is the emergence of urban dengue, a disease regarded as acute infectious fever. The disease annually, has affected millions of people worldwide, mostly in the range of the intertropical globe. The disease's main vector in urban areas, the Aedes aegypti mosquito. Recent studies indicate that the distribution of the insect vector of dengue in the geographical area is directly tied to the behavior of environmental restrictions that area, especially among those, the air temperature and relative humidity. From that context, the work aims to estimate and spatializing, monthly, for each municipality in the state of Rio Grande do Norte, the potential of biophysical conditions conducive to the development of Aedes aegypti. Yet, made use of the following methodology: collection of epidemiological data and climatological, Normal climatological, descriptive statistics (measures of central tendency and scatter), uniform distribution, estimation geostatistics and sufer program, version 8.0. The results flagged for a behavior very heterogeneous, both in space and in time, in the case of the potential of biophysical conditions conducive to the development of Aedes aegypti in the state of Rio Grande do Norte. Still, he noted that there is a tendency for lifting the potential of development for the entire state, from the month of January, ending in the month of April mainly in central and western portions of the state. By contrast, there is the permanence of increased potential for development in the eastern portion of the state. The latter record maximum potential in the month of July, resulting probability of greater than 70% have been favorable conditions for the development of Aedes aegypti in that area. In the period between the months of August to December, it is small potential for development of Aedes aegypti in all parts of the state / Na atual conjuntura da sa?de p?blica mundial destaca-se a emerg?ncia da dengue urbana, doen?a infecciosa febril aguda que tem como principal vetor no meio urbano o Aedes aegypti. A doen?a, anualmente, tem acometido milh?es de pessoas em todo o mundo, principalmente na faixa intertropical do globo. A partir desse contexto, o trabalho objetiva estimar e espacializar mensalmente, sob a ?tima bioclim?tica, a potencialidade, para cada munic?pio do Rio Grande do Norte (RN), da condi??o biof?sica Favor?vel ao Desenvolvimento do Aedes aegypti. Para tanto, fez uso: coleta de dados epidemiol?gicos e climatol?gicos, Normal Climatol?gica, estat?stica descritiva (medidas de tend?ncia central e dispers?o), distribui??o uniforme, estima??o geoestat?stica e o programa sufer, vers?o 8.0. Os resultados sinalizaram para um comportamento bastante heterog?neo, tanto no espa?o como no tempo, em se tratando da potencialidade da condi??o biof?sica Favor?vel ao desenvolvimento do Aedes aegypti no estado RN. Ainda, observou-se que h? uma tend?ncia de eleva??o da potencialidade de desenvolvimento, para todo o estado, a partir do m?s de janeiro, cessando-se no m?s de abril principalmente nas por??es central e oeste do estado. Em contrapartida, para o mesmo per?odo, verifica-se a perman?ncia de aumento da potencialidade de desenvolvimento na por??o leste do estado. Esta ?ltima registrar? potencial m?ximo no m?s de julho, resultando em probabilidade maior que 70% de ter-se condi??es favor?veis ao desenvolvimento do Aedes aegypti naquela ?rea. No per?odo compreendido entre os meses de agosto a dezembro, constatou-se diminuto potencial de desenvolvimento do Aedes aegypti em todas as por??es do estado

L’imagerie polarimétrique permet d’estimer certaines caractéristiques d’un milieu qui peuvent ne pas être révélées par imagerie d’intensité standard. Cependant, les mesures effectuées peuvent être fortement perturbées par des fluctuations inhérentes aux processus physiques d’acquisition. Ces fluctuations sont difficiles à atténuer, notamment à cause de la fragilité des milieux observés ou de l’inhomogénéité des images acquises. Il est alors utile de caractériser la précision des estimations qu’il est possible d’obtenir. Dans cette thèse, cette question est abordée au travers de deux applications d’imagerie polarimétrique : la microscopie non-linéaire de second harmonique résolue en polarisation (PSHG) pour l’analyse de l’organisation structurale d’objets biomoléculaires, et l’imagerie radar polarimétrique interférométrique à synthèse d’ouverture (PolInSAR) pour l’estimation des paramètres du couvert forestier. Pour la première application, la précision d’estimation en présence de bruit de Poisson est caractérisée pour l’ensemble des assemblages moléculaires présentant une symétrie cylindrique, ce qui permet notamment d'aboutir à une procédure de détection des mesures qui ne permettent pas d’atteindre une précision d’estimation requise. Pour l’imagerie PolInSAR, on analyse une modalité d'acquisition intéressante pour les futures missions satellitaires. En particulier, on étudie dans ce contexte la précision d'estimation de la hauteur de végétation en présence de bruit de speckle en s'appuyant sur l'analyse du contraste polarimétrique. Une interprétation simple des comportements de cette modalité d'acquisition est obtenue dans la sphère de Poincaré. / Polarimetric imaging allows one to estimate some characteristics of a medium which might not be revealed by standard intensity imaging. However, the measurements can be strongly perturbed by fluctuations that are inherent in the physical acquisition processes. These fluctuations are difficult to attenuate, for instance because of the fragility of the observed media or because of the inhomogeneity of the obtained images. It is then useful to characterize the estimation precision that can be reached. In this thesis, this question is addressed through two polarimetric imaging applications: polarized-resolved second-harmonic generation non-linear microscopy (PSHG) for the analysis of the structural organization of biomolecular objects, and polarimetric interferometric synthetic aperture radar imaging (PolInSAR) for the estimation of vegetation parameters. For the first application, the estimation precision in the presence of Poisson noise is characterized for any molecular assembly that presents a cylindrical symmetry. This study results in particular in a procedure to detect the measurements that do not lead to a required precision. For PolInSAR imaging, we analyze an acquisition system that is interesting for future spatial missions. In particular, the estimation precision of the vegetation height is studied in this context in the presence of speckle noise by relying on the analysis of the polarimetric contrast. A simple interpretation of the behavior of this acquisition system is obtained in the Poincaré sphere.

Since the Morris worm was released in 1988, Internet worms continue to be one of top security threats. For example, the Conficker worm infected 9 to 15 million machines in early 2009 and shut down the service of some critical government and medical networks. Moreover, it constructed a massive peer-to-peer (P2P) botnet. Botnets are zombie networks controlled by attackers setting out coordinated attacks. In recent years, botnets have become the number one threat to the Internet. The objective of this research is to characterize spatial-temporal infection structures of Internet worms, and apply the observations to study P2P-based botnets formed by worm infection. First, we infer temporal characteristics of the Internet worm infection structure, i.e., the host infection time and the worm infection sequence, and thus pinpoint patient zero or initially infected hosts. Specifically, we apply statistical estimation techniques on Darknet observations. We show analytically and empirically that our proposed estimators can significantly improve the inference accuracy. Second, we reveal two key spatial characteristics of the Internet worm infection structure, i.e., the number of children and the generation of the underlying tree topology formed by worm infection. Specifically, we apply probabilistic modeling methods and a sequential growth model. We show analytically and empirically that the number of children has asymptotically a geometric distribution with parameter 0.5, and the generation follows closely a Poisson distribution. Finally, we evaluate bot detection strategies and effects of user defenses in P2P-based botnets formed by worm infection. Specifically, we apply the observations of the number of children and demonstrate analytically and empirically that targeted detection that focuses on the nodes with the largest number of children is an efficient way to expose bots. However, we also point out that future botnets may self-stop scanning to weaken targeted detection, without greatly slowing down the speed of worm infection. We then extend the worm spatial infection structure and show empirically that user defenses, e.g., patching or cleaning, can significantly mitigate the robustness and the effectiveness of P2P-based botnets. To counterattack, we evaluate a simple measure by future botnets that enhances topology robustness through worm re-infection.

Target tracking is an important part of video surveillance from a UAV. Tracking a target in an urban environment can be difficult because of the number of occlusions present in the environment. If multiple UAVs are used to track a target and the target behavior is learned autonomously by the UAV then the task may become easier. This thesis explores the hypothesis that an existing cooperative control algorithm can be enhanced by a language modeling algorithm to improve over time the target tracking performance of one or more ground targets in a dense urban environment. Observations of target behavior are reported to the Sequence Memoizer which uses the observations to create a belief model of future target positions. This belief model is combined with a kinematic belief model and then used in a cooperative auction algorithm for UAV path planning. The results for tracking a single target using the combined belief model outperform other belief models and improve over the duration of the mission. Results from tracking multiple targets indicate that algorithmic enhancements may be needed to find equivalent success. Future target tracking algorithms should involve machine learning to enhance tracking performance.

L'expansion du marché des semi-conducteurs dans tous les secteurs d'activité résulte de la capacité de créer de nouvelles applications grâce à l'intégration de plus en plus de fonctionnalités sur une surface de plus en plus faible. Pour chaque entreprise, la compétitivité dépend du coût de fabrication mais aussi de la fiabilité du produit. Ainsi, la phase de test d'un circuit intégré, et plus particulièrement des circuits analogiques et mixtes, est le facteur prédominant dans les choix d'un compromis entre ces deux critères antagonistes, car son coût est désormais proche du coût de production. Cette tendance contraint les acteurs du marché à mettre en place de nouvelles solutions moins onéreuses. Parmi les recherches dans ce domaine, la conception en vue du test (DfT) consiste à intégrer pendant le développement de la puce, une circuiterie additionnelle susceptible d'en faciliter le test, voire d'effectuer un auto-test (BIST). Mais la sélection d'une de ces techniques nécessite une évaluation de leur capacité de différencier les circuits fonctionnels des circuits défaillants. Ces travaux de recherche introduisent une méthodologie d'estimation de la qualité d'une DfT ou d'un BIST dans le flot de conception de circuits analogiques et mixtes. Basée sur la génération d'un large échantillon prenant en compte l'impact des variations d'un procédé technologique sur les performances et les mesures de test du circuit, cette méthodologie calcule les métriques de test exprimant la capacité de chaque technique de détecter les circuits défaillants sans rejeter des circuits fonctionnels et d'accepter les circuits fonctionnels en rejetant les circuits défaillant. Ensuite, le fonctionnement d'un auto-test numérique adapté aux convertisseurs sigma-delta est présenté ainsi qu'une nouvelle méthode de génération et d'injection du stimulus de test. La qualité de ces techniques d'auto-test est démontrée en utilisant la méthodologie d'estimation des métriques de test. Enfin, un démonstrateur développé sur un circuit programmable démontre la possibilité d'employer une technique d'auto-test dans un système de calibrage intégré. / The pervasiveness of the semiconductor industry in an increasing range of applications that span human activity stems from our ability to integrate more and more functionalities onto a small silicon area. The competitiveness in this industry, apart from product originality, is mainly defined by the manufacturing cost, as well as the product reliability. Therefore, finding a trade-off between these two often contradictory objectives is a major concern and calls for efficient test solutions. The focus nowadays is mainly on Analog and Mixed-Signal (AMS) circuits since the associated testing cost can amount up to 70% of the overall manufacturing cost despite that AMS circuits typically occupy no more than 20% of the die area. To this end, there are intensified efforts by the industry to develop more economical test solutions without sacrificing product quality. Design-for-Test (DfT) is a promising alternative to the standard test techniques. It consists of integrating during the development phase of the chip extra on-chip circuitry aiming to facilitate testing or even enable a built-in self-test (BIST). However, the adoption of a DFT technique requires a prior evaluation of its capability to distinguish the functional circuits from the defective ones. In this thesis, we present a novel methodology for estimating the quality of a DfT technique that is readily incorporated in the design flow of AMS circuits. Based on the generation of a large synthetic sample of circuits that takes into account the impact of the process ariations on the performances and test measurements, this methodology computes test metrics that determine whether the DFT technique is capable of rejecting defective devices while passing functional devices. In addition, the thesis proposes a novel, purely digital BIST technique for Sigma-Delta analog-to-digital converters. The efficiency of the test metrics evaluation methodology is demonstrated on this novel BIST technique. Finally, a hardware prototype developed on an FPGA shows the possibility of adapting the BIST technique within a calibration system.