Bayesian scene analysis for multi-camera 3D tracking and camera positioning

Los sistemas de seguimiento mono-cámara han demostrado su notable capacidad para el análisis de trajectorias de objectos móviles y para monitorización de escenas de interés; sin embargo, tanto su robustez como sus posibilidades en cuanto a comprensión semántica de la escena están fuertemente limitadas por su naturaleza local y monocular, lo que los hace insuficientes para aplicaciones realistas de videovigilancia. El objetivo de esta tesis es la extensión de las posibilidades de los sistemas de seguimiento de objetos móviles para lograr un mayor grado de robustez y comprensión de la escena. La extensión propuesta se divide en dos direcciones separadas. La primera puede considerarse local, ya que está orientada a la mejora y enriquecimiento de las posiciones estimadas para los objetos móviles observados directamente por las cámaras del sistema; dicha extensión se logra mediante el desarrollo de un sistema multi-cámara de seguimiento 3D, capaz de proporcionar consistentemente las posiciones 3D de múltiples objetos a partir de las observaciones capturadas por un conjunto de sensores calibrados y con campos de visión solapados. La segunda extensión puede considerarse global, dado que su objetivo consiste en proporcionar un contexto global para relacionar las observaciones locales realizadas por una cámara con una escena de mucho mayor tamaño; para ello se propone un sistema automático de localización de cámaras basado en las trayectorias observadas de varios objetos móviles y en un mapa esquemático de la escena global monitorizada. Ambas líneas de investigación se tratan utilizando, como marco común, técnicas de estimación bayesiana: esta elección está justificada por la versatilidad y flexibilidad proporcionada por dicho marco estadístico, que permite la combinación natural de múltiples fuentes de información sobre los parámetros a estimar, así como un tratamiento riguroso de la incertidumbre asociada a las mismas mediante la inclusión de modelos de observación específicamente diseñados. Además, el marco seleccionado abre grandes posibilidades operacionales, puesto que permite la creación de diferentes métodos numéricos adaptados a las necesidades y características específicas de distintos problemas tratados. El sistema de seguimiento 3D con múltiples cámaras propuesto está específicamente diseñado para permitir descripciones esquemáticas de las medidas realizadas individualmente por cada una de las cámaras del sistema: esta elección de diseño, por tanto, no asume ningún algoritmo específico de detección o seguimiento 2D en ninguno de los sensores de la red, y hace que el sistema propuesto sea aplicable a redes reales de vigilancia con capacidades limitadas tanto en términos de procesamiento como de transmision. La combinación robusta de las observaciones capturadas individualmente por las cámaras, ruidosas, incompletas y probablemente contaminadas por falsas detecciones, se basa en un metodo de asociación bayesiana basado en geometría y color: los resultados de dicha asociación permiten el seguimiento 3D de los objetos de la escena mediante el uso de un filtro de partículas. El sistema de fusión de observaciones propuesto tiene, como principales características, una gran precisión en términos de localización 3D de objetos, y una destacable capacidad de recuperación tras eventuales errores debidos a un número insuficiente de datos de entrada. El sistema automático de localización de cámaras se basa en la observación de múltiples objetos móviles y un mapa esquemático de las áreas transitables del entorno monitorizado para inferir la posición absoluta de dicho sensor. Para este propósito, se propone un novedoso marco bayesiano que combina modelos dinámicos inducidos por el mapa en los objetos móviles presentes en la escena con las trayectorias observadas por la cámara, lo que representa un enfoque nunca utilizado en la literatura existente. El sistema de localización se divide en dos sub-tareas diferenciadas, debido a que cada una de estas tareas requiere del diseño de algoritmos específicos de muestreo para explotar en profundidad las características del marco desarrollado: por un lado, análisis de la ambigüedad del caso específicamente tratado y estimación aproximada de la localización de la cámara, y por otro, refinado de la localización de la cámara. El sistema completo, diseñado y probado para el caso específico de localización de cámaras en entornos de tráfico urbano, podría tener aplicación también en otros entornos y sensores de diferentes modalidades tras ciertas adaptaciones. ABSTRACT Mono-camera tracking systems have proved their capabilities for moving object trajectory analysis and scene monitoring, but their robustness and semantic possibilities are strongly limited by their local and monocular nature and are often insufficient for realistic surveillance applications. This thesis is aimed at extending the possibilities of moving object tracking systems to a higher level of scene understanding. The proposed extension comprises two separate directions. The first one is local, since is aimed at enriching the inferred positions of the moving objects within the area of the monitored scene directly covered by the cameras of the system; this task is achieved through the development of a multi-camera system for robust 3D tracking, able to provide 3D tracking information of multiple simultaneous moving objects from the observations reported by a set of calibrated cameras with semi-overlapping fields of view. The second extension is global, as is aimed at providing local observations performed within the field of view of one camera with a global context relating them to a much larger scene; to this end, an automatic camera positioning system relying only on observed object trajectories and a scene map is designed. The two lines of research in this thesis are addressed using Bayesian estimation as a general unifying framework. Its suitability for these two applications is justified by the flexibility and versatility of that stochastic framework, which allows the combination of multiple sources of information about the parameters to estimate in a natural and elegant way, addressing at the same time the uncertainty associated to those sources through the inclusion of models designed to this end. In addition, it opens multiple possibilities for the creation of different numerical methods for achieving satisfactory and efficient practical solutions to each addressed application. The proposed multi-camera 3D tracking method is specifically designed to work on schematic descriptions of the observations performed by each camera of the system: this choice allows the use of unspecific off-the-shelf 2D detection and/or tracking subsystems running independently at each sensor, and makes the proposal suitable for real surveillance networks with moderate computational and transmission capabilities. The robust combination of such noisy, incomplete and possibly unreliable schematic descriptors relies on a Bayesian association method, based on geometry and color, whose results allow the tracking of the targets in the scene with a particle filter. The main features exhibited by the proposal are, first, a remarkable accuracy in terms of target 3D positioning, and second, a great recovery ability after tracking losses due to insufficient input data. The proposed system for visual-based camera self-positioning uses the observations of moving objects and a schematic map of the passable areas of the environment to infer the absolute sensor position. To this end, a new Bayesian framework combining trajectory observations and map-induced dynamic models for moving objects is designed, which represents an approach to camera positioning never addressed before in the literature. This task is divided into two different sub-tasks, setting ambiguity analysis and approximate position estimation, on the one hand, and position refining, on the other, since they require the design of specific sampling algorithms to correctly exploit the discriminative features of the developed framework. This system, designed for camera positioning and demonstrated in urban traffic environments, can also be applied to different environments and sensors of other modalities after certain required adaptations.

Dust tracking techniques applied to the STARDUST facility: First results

An important issue related to future nuclear fusion reactors fueled with deuterium and tritium is the creation of large amounts of dust due to several mechanisms (disruptions, ELMs and VDEs). The dust size expected in nuclear fusion experiments (such as ITER) is in the order of microns (between 0.1 and 1000 μm). Almost the total amount of this dust remains in the vacuum vessel (VV). This radiological dust can re-suspend in case of LOVA (loss of vacuum accident) and these phenomena can cause explosions and serious damages to the health of the operators and to the integrity of the device. The authors have developed a facility, STARDUST, in order to reproduce the thermo fluid-dynamic conditions comparable to those expected inside the VV of the next generation of experiments such as ITER in case of LOVA. The dust used inside the STARDUST facility presents particle sizes and physical characteristics comparable with those that created inside the VV of nuclear fusion experiments. In this facility an experimental campaign has been conducted with the purpose of tracking the dust re-suspended at low pressurization rates (comparable to those expected in case of LOVA in ITER and suggested by the General Safety and Security Report ITER-GSSR) using a fast camera with a frame rate from 1000 to 10,000 images per second. The velocity fields of the mobilized dust are derived from the imaging of a two-dimensional slice of the flow illuminated by optically adapted laser beam. The aim of this work is to demonstrate the possibility of dust tracking by means of image processing with the objective of determining the velocity field values of dust re-suspended during a LOVA.

Bayesian scene analysis for multi-camera 3D tracking and camera positioning

Los sistemas de seguimiento mono-cámara han demostrado su notable capacidad para el análisis de trajectorias de objectos móviles y para monitorización de escenas de interés; sin embargo, tanto su robustez como sus posibilidades en cuanto a comprensión semántica de la escena están fuertemente limitadas por su naturaleza local y monocular, lo que los hace insuficientes para aplicaciones realistas de videovigilancia. El objetivo de esta tesis es la extensión de las posibilidades de los sistemas de seguimiento de objetos móviles para lograr un mayor grado de robustez y comprensión de la escena. La extensión propuesta se divide en dos direcciones separadas. La primera puede considerarse local, ya que está orientada a la mejora y enriquecimiento de las posiciones estimadas para los objetos móviles observados directamente por las cámaras del sistema; dicha extensión se logra mediante el desarrollo de un sistema multi-cámara de seguimiento 3D, capaz de proporcionar consistentemente las posiciones 3D de múltiples objetos a partir de las observaciones capturadas por un conjunto de sensores calibrados y con campos de visión solapados. La segunda extensión puede considerarse global, dado que su objetivo consiste en proporcionar un contexto global para relacionar las observaciones locales realizadas por una cámara con una escena de mucho mayor tamaño; para ello se propone un sistema automático de localización de cámaras basado en las trayectorias observadas de varios objetos móviles y en un mapa esquemático de la escena global monitorizada. Ambas líneas de investigación se tratan utilizando, como marco común, técnicas de estimación bayesiana: esta elección está justificada por la versatilidad y flexibilidad proporcionada por dicho marco estadístico, que permite la combinación natural de múltiples fuentes de información sobre los parámetros a estimar, así como un tratamiento riguroso de la incertidumbre asociada a las mismas mediante la inclusión de modelos de observación específicamente diseñados. Además, el marco seleccionado abre grandes posibilidades operacionales, puesto que permite la creación de diferentes métodos numéricos adaptados a las necesidades y características específicas de distintos problemas tratados. El sistema de seguimiento 3D con múltiples cámaras propuesto está específicamente diseñado para permitir descripciones esquemáticas de las medidas realizadas individualmente por cada una de las cámaras del sistema: esta elección de diseño, por tanto, no asume ningún algoritmo específico de detección o seguimiento 2D en ninguno de los sensores de la red, y hace que el sistema propuesto sea aplicable a redes reales de vigilancia con capacidades limitadas tanto en términos de procesamiento como de transmision. La combinación robusta de las observaciones capturadas individualmente por las cámaras, ruidosas, incompletas y probablemente contaminadas por falsas detecciones, se basa en un metodo de asociación bayesiana basado en geometría y color: los resultados de dicha asociación permiten el seguimiento 3D de los objetos de la escena mediante el uso de un filtro de partículas. El sistema de fusión de observaciones propuesto tiene, como principales características, una gran precisión en términos de localización 3D de objetos, y una destacable capacidad de recuperación tras eventuales errores debidos a un número insuficiente de datos de entrada. El sistema automático de localización de cámaras se basa en la observación de múltiples objetos móviles y un mapa esquemático de las áreas transitables del entorno monitorizado para inferir la posición absoluta de dicho sensor. Para este propósito, se propone un novedoso marco bayesiano que combina modelos dinámicos inducidos por el mapa en los objetos móviles presentes en la escena con las trayectorias observadas por la cámara, lo que representa un enfoque nunca utilizado en la literatura existente. El sistema de localización se divide en dos sub-tareas diferenciadas, debido a que cada una de estas tareas requiere del diseño de algoritmos específicos de muestreo para explotar en profundidad las características del marco desarrollado: por un lado, análisis de la ambigüedad del caso específicamente tratado y estimación aproximada de la localización de la cámara, y por otro, refinado de la localización de la cámara. El sistema completo, diseñado y probado para el caso específico de localización de cámaras en entornos de tráfico urbano, podría tener aplicación también en otros entornos y sensores de diferentes modalidades tras ciertas adaptaciones. ABSTRACT Mono-camera tracking systems have proved their capabilities for moving object trajectory analysis and scene monitoring, but their robustness and semantic possibilities are strongly limited by their local and monocular nature and are often insufficient for realistic surveillance applications. This thesis is aimed at extending the possibilities of moving object tracking systems to a higher level of scene understanding. The proposed extension comprises two separate directions. The first one is local, since is aimed at enriching the inferred positions of the moving objects within the area of the monitored scene directly covered by the cameras of the system; this task is achieved through the development of a multi-camera system for robust 3D tracking, able to provide 3D tracking information of multiple simultaneous moving objects from the observations reported by a set of calibrated cameras with semi-overlapping fields of view. The second extension is global, as is aimed at providing local observations performed within the field of view of one camera with a global context relating them to a much larger scene; to this end, an automatic camera positioning system relying only on observed object trajectories and a scene map is designed. The two lines of research in this thesis are addressed using Bayesian estimation as a general unifying framework. Its suitability for these two applications is justified by the flexibility and versatility of that stochastic framework, which allows the combination of multiple sources of information about the parameters to estimate in a natural and elegant way, addressing at the same time the uncertainty associated to those sources through the inclusion of models designed to this end. In addition, it opens multiple possibilities for the creation of different numerical methods for achieving satisfactory and efficient practical solutions to each addressed application. The proposed multi-camera 3D tracking method is specifically designed to work on schematic descriptions of the observations performed by each camera of the system: this choice allows the use of unspecific off-the-shelf 2D detection and/or tracking subsystems running independently at each sensor, and makes the proposal suitable for real surveillance networks with moderate computational and transmission capabilities. The robust combination of such noisy, incomplete and possibly unreliable schematic descriptors relies on a Bayesian association method, based on geometry and color, whose results allow the tracking of the targets in the scene with a particle filter. The main features exhibited by the proposal are, first, a remarkable accuracy in terms of target 3D positioning, and second, a great recovery ability after tracking losses due to insufficient input data. The proposed system for visual-based camera self-positioning uses the observations of moving objects and a schematic map of the passable areas of the environment to infer the absolute sensor position. To this end, a new Bayesian framework combining trajectory observations and map-induced dynamic models for moving objects is designed, which represents an approach to camera positioning never addressed before in the literature. This task is divided into two different sub-tasks, setting ambiguity analysis and approximate position estimation, on the one hand, and position refining, on the other, since they require the design of specific sampling algorithms to correctly exploit the discriminative features of the developed framework. This system, designed for camera positioning and demonstrated in urban traffic environments, can also be applied to different environments and sensors of other modalities after certain required adaptations.

Emergency management in the highways of the future. A performance-based multilayered ITS architecture design proposal

Emergency management is one of the key aspects within the day-to-day operation procedures in a highway. Efficiency in the overall response in case of an incident is paramount in reducing the consequences of any incident. However, the approach of highway operators to the issue of incident management is still usually far from a systematic, standardized way. This paper attempts to address the issue and provide several hints on why this happens, and a proposal on how the situation could be overcome. An introduction to a performance based approach to a general system specification will be described, and then applied to a particular road emergency management task. A real testbed has been implemented to show the validity of the proposed approach. Ad-hoc sensors (one camera and one laser scanner) were efficiently deployed to acquire data, and advanced fusion techniques applied at the processing stage to reach the specific user requirements in terms of functionality, flexibility and accuracy.

Methodology of image analysis for study of the vertisols moisture content

The main problem to study vertical drainage from the moisture distribution, on a vertisol profile, is searching for suitable methods using these procedures. Our aim was to design a digital image processing methodology and its analysis to characterize the moisture content distribution of a vertisol profile. In this research, twelve soil pits were excavated on a ba re Mazic Pellic Vertisols ix of them in May 13/2011 and the rest in May 19 /2011 after a moderate rainfall event. Digital RGB images were taken from each vertisol pit using a Kodak? camera selecting a size of 1600x945 pixels. Each soil image was processed to homogenized brightness and then a spatial filter with several window sizes was applied to select the optimum one. The RGB image obtained were divided in each matrix color selecting the best thresholds for each one, maximum and minimum, to be applied and get a digital binary pattern. This one was analyzed by estimating two fractal scaling exponents box counting dimension D BC) and interface fractal dimension (D) In addition, three pre-fractal scaling coefficients were determinate at maximum resolution: total number of boxes intercepting the foreground pattern (A), fractal lacunarity (?1) and Shannon entropy S1). For all the images processed the spatial filter 9x9 was the optimum based on entropy, cluster and histogram criteria. Thresholds for each color were selected based on bimodal histograms.

A Ground-Truth Video Dataset for the Development and Evaluation of Vision-based Sense-and-Avoid systems

The importance of vision-based systems for Sense-and-Avoid is increasing nowadays as remotely piloted and autonomous UAVs become part of the non-segregated airspace. The development and evaluation of these systems demand flight scenario images which are expensive and risky to obtain. Currently Augmented Reality techniques allow the compositing of real flight scenario images with 3D aircraft models to produce useful realistic images for system development and benchmarking purposes at a much lower cost and risk. With the techniques presented in this paper, 3D aircraft models are positioned firstly in a simulated 3D scene with controlled illumination and rendering parameters. Realistic simulated images are then obtained using an image processing algorithm which fuses the images obtained from the 3D scene with images from real UAV flights taking into account on board camera vibrations. Since the intruder and camera poses are user-defined, ground truth data is available. These ground truth annotations allow to develop and quantitatively evaluate aircraft detection and tracking algorithms. This paper presents the software developed to create a public dataset of 24 videos together with their annotations and some tracking application results.

Using the Cross-Entropy Method for Control Optimization: A Case Study of See-and-Avoid on Unmanned Aerial Vehicles

This paper presents an adaptation of the Cross-Entropy (CE) method to optimize fuzzy logic controllers. The CE is a recently developed optimization method based on a general Monte-Carlo approach to combinatorial and continuous multi-extremal optimization and importance sampling. This work shows the application of this optimization method to optimize the inputs gains, the location and size of the different membership functions' sets of each variable, as well as the weight of each rule from the rule's base of a fuzzy logic controller (FLC). The control system approach presented in this work was designed to command the orientation of an unmanned aerial vehicle (UAV) to modify its trajectory for avoiding collisions. An onboard looking forward camera was used to sense the environment of the UAV. The information extracted by the image processing algorithm is the only input of the fuzzy control approach to avoid the collision with a predefined object. Real tests with a quadrotor have been done to corroborate the improved behavior of the optimized controllers at different stages of the optimization process.

Statistical moving object detection for mobile devices with camera

A novel and high-quality system for moving object detection in sequences recorded with moving cameras is proposed. This system is based on the collaboration between an automatic homography estimation module for image alignment, and a robust moving object detection using an efficient spatiotemporal nonparametric background modeling.

Observations of Comet 9P/Tempel 1 around the Deep Impact event by the OSIRIS cameras onboard Rosetta

The OSIRIS cameras on the Rosetta spacecraft observed Comet 9P/Tempel 1 from 5 days before to 10 days after it was hit by the Deep Impact projectile. The Narrow Angle Camera (NAC) monitored the cometary dust in 5 different filters. The Wide Angle Camera (WAC) observed through filters sensitive to emissions from OH, CN, Na, and OI together with the associated continuum. Before and after the impact the comet showed regular variations in intensity. The period of the brightness changes is consistent with the rotation period of Tempel 1. The overall brightness of Tempel 1 decreased by about 10% during the OSIRIS observations. The analysis of the impact ejecta shows that no new permanent coma structures were created by the impact. Most of the material moved with View the MathML source∼200ms−1. Much of it left the comet in the form of icy grains which sublimated and fragmented within the first hour after the impact. The light curve of the comet after the impact and the amount of material leaving the comet (View the MathML source4.5–9×106kg of water ice and a presumably larger amount of dust) suggest that the impact ejecta were quickly accelerated by collisions with gas molecules. Therefore, the motion of the bulk of the ejecta cannot be described by ballistic trajectories, and the validity of determinations of the density and tensile strength of the nucleus of Tempel 1 with models using ballistic ejection of particles is uncertain.

Osiris— The optical, spectroscopic and infrared remote imaging system for the Rosetta orbiter

The scientific objectives, design, and implementation of the Optical, Spectroscopic, and Infrared Remote Imaging System (OSIRIS) for the International Rosetta Mission are described. The instrument comprises two camera systems with a common electronics box. A narrow angle camera will provide high resolution images of the structure and morphology of the nucleus of a comet. A wide angle camera with high straylight rejection and dynamic range will be used to investigate the innermost coma and the emission process at the surface of the comet. An infrared imaging system, which dramatically enhances the scientific return has been included in the narrow angle camera at little extra cost.

Use of infrared thermography as a tool to monitor skin temperature along the recovery process of an anterior cruciate ligament surgery

Infrared thermography (IRT) is a safe and non-invasive tool used for examining physiological functions based on skin temperature (Tsk) control. Thermograms from 25 anterior cruciate ligament (ACL) surgically operated patients (2 female, 23 male) were taken with a FLIR infrared camera according to the protocol established by the International Academy of Clinical Thermology (IACT). This work consists of 4 studies. Studies 1 and 3 were related to establish the probable thermal difference among different moments of an ACL rupture after surgery: before starting the rehabilitation (P0), at the end of rehabilitation (P1) and 18 months from the end of rehabilitation (P2). For this purpose, on the other hand, studies 2 and 4 were related to establish the skin thermal difference (Tsk) between the injured and the non-injured leg in P0, P1 and P2. Results of the first study showed significant temperature increases in the posterior thigh area between P0 and P1 probably due to a compensatory mechanism. According to this, we can conclude that temperature of the posterior area of the injured and noninjured leg has increased from the first to the last day of the rehabilitation process. In the second study we found significant temperature differences between the injured and non-injured leg in both stages of rehabilitation (p<.01). On the one hand, the temperature of the injured leg is higher in the anterior view and the temperature of the non-injured leg is higher in the posterior view. By the time the patients had recovered from the reconstruction, thermal imbalances should have not been shown between symmetrical parts, but differences seemed to be still latent.. Study 3 shows that temperatures seem to be higher after a year and a half (P2) than in P1. Study 4 shows how thermal values 18 months later seemed to be normalized between both legs. No significant differences were found between the injured leg and the noninjured leg after one year and a half of the rehabilitation process. Considering results from Study 3 and 4 we can conclude that patients seemed to have recovered from a thermal point of view. The temperature in P2 was higher but symmetrical. RESUMEN La termografía infrarroja (IRT) es una herramienta segura y no invasiva utilizada para examinar funciones fisiológicas que se basan en el control de temperatura de la piel (Tsk). Termogramas de 25 pacientes intervenidos quirúrgicamente del ligamento cruzado anterior (LCA) (2 mujeres, 23 hombres) fueron tomadas con una cámara de infrarrojos FLIR de acuerdo con el protocolo establecido por la Academia Internacional de Termología Clínica (IACT). Este trabajo consiste en 4 estudios. Los estudios 1 y 3 describen la diferencia térmica entre los diferentes momentos tras la operación del ligamento cruzado anterior: antes de comenzar la rehabilitación (P0), al final de la rehabilitación (P1) y 18 meses tras finalizar la rehabilitación (P2). Por otra parte, los estudios 2 y 4 describen la diferencia de temperatura de la piel (Tsk) entre la pierna lesionada y la pierna no lesionada en P0, P1 y P2. Los resultados del primer estudio mostraron aumentos significativos de temperatura en la zona posterior de los muslos entre P0 y P1, probablemente debido a un mecanismo de compensación. De acuerdo con esto, se puede concluir que la temperatura de la zona posterior de la pierna lesionada y no lesionada se ha incrementado desde el primero hasta el último día del proceso de rehabilitación. En el segundo estudio se encontraron diferencias significativas de temperatura entre la pierna lesionada y no lesionada en ambas etapas de la rehabilitación (p<.01). Por un lado, la temperatura de la pierna lesionada es mayor en la vista anterior. Por otro lado, la temperatura de la pierna no lesionada es mayor en la vista posterior. Una vez que los pacientes se han recuperado de todo el proceso, no deberían existir desequilibrios térmicos entre partes simétricas del cuerpo, pero las diferencias todavía estaban latentes. El tercer estudio muestra que la temperatura es más alta en P2 que en P1. El cuarto estudio muestra cómo los valores térmicos entre ambas piernas en P2 se han normalizado entre ambas piernas. No se encontraron diferencias significativas entre la pierna lesionada y la pierna no lesionada después de 18 meses tras el proceso de rehabilitación. Considerando los resultados del studio 3 y 4, podemos concluir que se ha llegado a la recuperación total desde un punto de vista térmico. La temperatura es más elevada en P2 pero simétrica.