Visual target tracking problem consists of two topics: Obtaining targets from camera measurements and target tracking. Even though it has been studied for more than 30 years, there are still some problems not completely solved. Especially in the case of multiple targets, association of measurements to targets, creation of new targets and deletion of old ones are among those. What is more, it is very important to deal with the occlusion and crossing targets problems suitably. We believe that a slightly modified version of multiple hypothesis tracking can successfully deal with most of the aforementioned problems with sufficient success. Distance, track size, track color, gate size and track history are used as parameters to evaluate the hypotheses generated for measurement to track association problem whereas size and color are used as parameters for occlusion problem. The overall tracker has been fine tuned over some scenarios and it has been observed that it performs well over the testing scenarios as well. Furthermore the performance of the tracker is analyzed according to those parameters in both association and occlusion handling situations.

No description available.

No description available.

No description available.

Le suivi de particules est une méthode de choix pour comprendre les mécanismes intra-cellulaires car il fournit des moyens robustes et précis de caractériser la dynamiques des objets mobiles à l'échelle micro et nano métrique. Cette thèse traite de plusieurs aspects liés au problème du suivi de plusieurs centaines de particules dans des conditions bruitées. Nous présentons des techniques nouvelles basée sur des méthodes mathématiques robustes qui nous permettent des suivre des particules sous-résolutives dans les conditions variées qui sont rencontrées en imagerie cellulaire. Détection de particules : nous avons tout d'abord traité le problème de la détection de particules dans les images fluorescentes contenant un fond structuré. L'idée clé de la méthode est l'utilisation d'une technique de séparation de sources : l'algorithme d'Analyse en Composantes Morphologiques (ACM), pour séparer le fond des particules en exploitant leur différence de morphologie dans les images. Nous avons effectué un certain nombre de modifications à l'ACM pour l'adapter aux caractéristiques des images biologiques en fluorescence. Par exemple, nous avons proposé l'utilisation du dictionnaire de Curvelet et d'un dictionnaire de d'ondelettes, avec des à priori de parcimonie différents, afin de séparer le signal des particules du fond. Une fois la séparation de sources effectuée, l'image sans fond peut être analysée pour identifier de manière robuste la position des particules et pour les suivre au cours du temps. Modélisation du problème de suivi : nous avons proposé un cadre de travail statistique global qui tient compte des nombreux aspects du problème de suivi de particules dans des conditions bruitées. Le cadre de travail probabiliste que nous avons mis au point contient de nombreux modèles qui sont dédiés à l'imagerie biologique, tels que des modèles statistiques de mouvement des particules en milieu cellulaire. Nous avons aussi défini la concept de perceiability d'une cible dans le cas des particules biologiques. Grâce à ce modèle l'existence d'une particule est explicitement modélisée et quantifiée, ce qui nous permet de résoudre les problèmes de création et de terminaison des trajectoires au sein même de notre cadre probabiliste de suivi. Le cadre de travail proposé bénéficie d'une grande flexibilité mais reste facile à adapter car chaque paramètre du modèle trouve une interprétation simple et intuitive. Ainsi, notre modèle probabiliste de suivi nous a permis de modéliser de manière exhaustive un grand nombre de systèmes biologiques différents. Mise au point d'un algorithme de suivi : nous avons reformulé l'algorithme de suivi nommé Multiple Hypothesis Tracking (MHT) pour qu'il inclue notre modèle probabiliste de suivi dédié aux particules biologiques, et nous avons proposé une implémentation rapide qui permet de suivre de nombreuses particules dans des conditions d'imagerie dégradées. L'\textit{Enhanced} MHT (E-MHT) que nous avons proposé tire pleinement partie du modèle de suivi en incorporant la connaissance des images futures, ce qui augmente significativement le pouvoir discriminant des critères statistiques. En conséquence, l'E-MHT est capable d'identifier automatiquement les détections erronée et de détecter les événements d'apparition et de disparition des particules. Nous avons résolu le problème de la complexité de la tache de suivi grâce à un design de l'algorithme que exploite la topologie en arbre des solution et à la possibilité d'effectuer les calculs de manière parallèle. Une série de tests comparatifs entre l'E-MHT et des méthodes existantes de suivi a été réalisée avec des séquences d'images synthétiques 2D et avec des jeux de données réels 2D et 3D. Dans chaque cas l'E-MHT a montré des performances supérieures par rapport aux méthodes standards, avec une capacité remarquable à supporter des conditions d'imagerie très dégradées. Nous avons appliqué les méthodes de suivi proposées dans le cadre de plusieurs projets biologiques, ce qui a conduit à des résultats biologiques originaux. La flexibilité et la robustesse de notre méthode nous a notamment permis de suivre des prions infectant des cellules, de caractériser le transport de protéines lors du développement de l'ovocyte de la drosophile, ainsi que d'étudier la trafic d'ARN messager dans l'ovocyte de drosophile.

Odour localisation is the problem of finding the source of an odour or other volatile chemical. It promises many valuable practical and humanitarian applications. Most localisation methods require a robot to reactively track an odour plume along its entire length. This approach is time consuming and may be not be possible in a cluttered indoor environment, where airflow tends to form sectors of circulating airflow. Such environments may be encountered in crawl-ways under floors, roof cavities, mines, caves, tree-canopies, air-ducts, sewers or tunnel systems. Operation in these places is important for such applications as search and rescue and locating the sources of toxic chemicals in an industrial setting. This thesis addresses odour localisation in this class of environments. The solution consists of a sense-map-plan-act style control scheme (and low level behaviour based controller) with two main stages. Firstly, the airflow in the environment is modelled using naive physics rules which are encapsulated into an algorithm named a Naive Reasoning Machine. It was used in preference to conventional methods as it is fast, does not require boundary conditions, and most importantly, provides approximate solutions to the degree of accuracy required for the task, with analogical data structures that are readily useful to a reasoning algorithm. Secondly, a reasoning algorithm navigates the robot to specific target locations that are determined with a physical map, the airflow map, and knowledge of odour dispersal. Sensor measurements at the target positions provide information regarding the likelihood that odour was emitted from potential odour source locations. The target positions and their traversal are determined so that all the potential odour source sites are accounted for. The core method provides values corresponding to the confidence that the odour source is located in a given region. A second search stage exploiting vision is then used to locate the specific location of the odour source within the predicted region. This comprises the second part of a bi-modal, two-stage search, with each stage exploiting complementary sensing modalities. Single hypothesis airflow modelling faces limitations due to the fact that large differences between airflow topologies are predicted for only small variations in a physical map. This is due to uncertainties in the map and approximations in the modelling process. Furthermore, there are uncertainties regarding the flow direction through inlet/outlet ducts. A method is presented for dealing with these uncertainties, by generating multiple airflow hypotheses. As the robot performs odour localisation, airflow in the environment is measured and used to adjust the confidences of the hypotheses using Bayesian inference. The best hypothesis is then selected, which allows the completion of the localisation task. This method improves the robustness of odour localisation in the presence of uncertainties, making it possible where the single hypothesis method would fail. It also demonstrates the potential for integrating naive physics into a statistical framework. Extensive experimental results are presented to support the methods described above.

QC 20100621

A cruise missile is a guided missile that uses a lifting wing and a jet propulsion system to allow sustained flight. Cruise missiles are, in essence, unmanned aircraft and they are generally designed to carry a large conventional or nuclear warhead many hundreds of miles with excellent accuracy. In this study, navigation and control studies on cruise missiles are performed. Due to the variety and complexity of the subsystems of the cruise missiles, the main concern is limited with the navigation system. Navigation system determines the position, velocity, attitude and time solutions of the missile. Therefore, it can be concluded that an accurate self-contained navigation system directly influences the success of the missile. In the study, modern radar data association algorithms are implemented as new Terrain Aided Navigation (TAN) algorithms which can be used with low-cost Inertial Measurement Units (IMU&rsquo / s). In order to perform the study, first a thorough survey of the literature on mid-course navigation of cruise missiles is performed. Then, study on modern radar data association algorithms and their implementations to TAN are done with simple simulations. At the case study part, a six degree of freedom (6 DOF) flight simulation tool is developed which includes the aerodynamic and dynamic model of the cruise missile model including error model of the navigation system. Finally, the performances of the designed navigation systems with the implemented TAN algorithms are examined in detail with the help of the simulations performed.

In this study, an expert system based EA and tracking system is developed and the performances of these systems are optimized. Tracking system consists of a monopulse tracking radar and a Multiple Hypothesis Tracking (MHT) algorithm. MHT is modelled as a measurement&amp / #8208 / oriented approach, which is capable of initiating tracks. As each measurement is received, probabilities are calculated for the hypotheses and target states are estimated using a Kalman filter. Range Gate Pull-Off (RGPO) is selected as an EA technique to be developed because it is accepted to be the primary deception technique employed against tracking radar. Two modes of RGPO technique / linear and parabolic, according to time delay controller are modelled. Genetic Algorithm (GA) Toolbox of MATLAB is used for the optimization of these systems over some predetermined scenarios. It is observed that the performance of the tracking radar system is improved significantly and successful tracking is achieved over all given scenarios, even for closely spaced targets. RGPO models are developed against this improved tracking performance and deception of tracking radar is succeeded for all given target models.

In this study, primary steps of a visual surveillance system are presented: moving object detection and tracking of these moving objects. Background subtraction has been performed to detect the moving objects in the video, which has been taken from a static camera. Four methods, frame differencing, running (moving) average, eigenbackground subtraction and mixture of Gaussians, have been used in the background subtraction process. After background subtraction, using some additional operations, such as morphological operations and connected component analysis, the objects to be tracked have been acquired. While tracking the moving objects, active contour models (snakes) has been used as one of the approaches. In addition to this method / Kalman tracker and mean-shift tracker are other approaches which have been utilized. A new approach has been proposed for the problem of tracking multiple targets. We have implemented this method for single and multiple camera configurations. Multiple cameras have been used to augment the measurements. Homography matrix has been calculated to find the correspondence between cameras. Then, measurements and tracks have been associated by the new tracking method.