published_or_final_version / Politics and Public Administration / Master / Master of Public Administration

published_or_final_version / Community Medicine / Master / Master of Public Health

We present a novel approach for vehicle detection and tracking in highway surveillance videos. This method incorporates well-studied computer vision and machine learning techniques to form an unsupervised system, where vehicles are automatically "learned" from video sequences. First an enhanced adaptive background mixture model is used to identify positive and negative examples. Then a video-specific classifier is trained with these examples. Both the background model and the trained classifier are used in conjunction to detect vehicles in a frame. Tracking is achieved by a simplified multi-hypotheses approach. An over-complete set of tracks is created considering every observation within a time interval. As needed hypothesized detections are generated to force continuous tracks. Finally, a scoring function is used to separate the valid tracks in the over-complete set. The proposed detection and tracking algorithm is tested in a challenging application; vehicle counting. Our method achieved very accurate results in three traffic surveillance videos that are significantly different in terms of view-point, quality and clutter. / text

La digestion anaérobie implique un écosystème complexe qui dégrade progressivement la matière organique est la transforme en dioxyde de carbone et en (bio)méthane. Ce biogaz est une source d'énergie renouvelable dont l'utilisation s'inscrit dans une optique de développement durable. Le potentiel biogaz reste néanmoins sous-exploité à cause d'un manque d'outils adaptés pour garantir la pérennité des installations et mieux maîtriser la qualité du biogaz. Nous proposons des stratégies pour mieux valoriser les résidus organiques liquides, à travers deux approches. L'une propose des méthodes pour identifier un risque de déstabilisation et prévenir un mauvais fonctionnement du procédé, l'autre repose sur la régulation de la qualité du biogaz. Ces méthodes nécessitent des modèles dynamiques pour décrire avec précision l'évolution des variables du système. Nous proposons une modélisation fine des transferts liquide-gaz, qui conduit à un modèle où le coefficient de transfert (kLa) dépend du débit total de biogaz.<br />Ce lien implique une relation linéaire entre la qualité du biogaz et la concentration en CO2 dissous, qui sert à améliorer les modèles existants. Un indice de risque de déstabilisation du procédé est construit à partir d'une analyse de stabilité d'un modèle simplifié. Cette procédure, appliquée à un procédé pilote, peut détecter une éventuelle déstabilisation du fermenteur, plus tôt que les indicateurs usuels (pH, acides gras volatils). Enfin, nous introduisons une nouvelle stratégie de contrôle de la qualité du biogaz basée sur la régulation de l'alcalinité dans le digesteur. Différentes lois de commande sont proposées et validées expérimentalement sur un réacteur pilote.

Decisions for asset allocation and protection are predicated upon accurate knowledge of the current operating environment as well as correctly characterizing the evolution of the environment over time. The desired kinematic and kinetic states of objects in question cannot be measured directly in most cases and instead are inferred or estimated from available measurements using a filtering process. Often, nonlinear transformations between the measurement domain and desired state domain distort the state domain probability density function yielding a form which does not necessarily resemble the form assumed in the filtering algorithm. The distortion effect must be understood in greater detail and appropriately accounted for so that even if sensors, state estimation algorithms, and state propagation algorithms operate in different domains, they can all be effectively utilized without any information loss due to domain transformations. This research presents an analytical investigation into understanding how non-linear transformations of stochastic, but characterizable, processes affect state and uncertainty estimation with direct application to space object surveillance and space- craft attitude determination. Analysis is performed with attention to construction of the state domain probability density function since state uncertainty and correlation are derived from the statistical moments of the probability density function. Analytical characterization of the effect nonlinear transformations impart on the structure of state probability density functions has direct application to conventional non- linear filtering and propagation algorithms in three areas: (1) understanding how smoothing algorithms used to estimate indirectly observed states impact state uncertainty, (2) justification or refutation of assumed state uncertainty distribution for more realistic uncertainty quantification, and (3) analytic automation of initial state estimate and covariance in lieu of user tuning. A nonlinear filtering algorithm based upon Bayes’ Theorem is presented to ac- count for the impact nonlinear domain transformations impart on probability density functions during the measurement update and propagation phases. The algorithm is able to accommodate different combinations of sensors for state estimation which can also be used to hypothesize system parameters or unknown states from available measurements because information is able to appropriately accounted for.

Generally  cameras  are  used  for  area  based  surveillance,  but  with  advancement  in  technology, cameras  can  be  employed  for wide  area  space  surveillance  (volumetric space surveillance). The volumetric systems require a number of cameras in order to cover a large area. Single panoramic cameras  do  not  provide  required  functionality  for  wide  area  space  surveillance  and Pan-Tile-Zoom (PTZ) cameras do not either cover wide areas.   To  meet  this challenge,  we  propose  a  cost  effective smart volumetric  surveillance system  which utilizes smaller number of cameras in order to provide a coverage of 360 degree. A test case used for this system includes detection of birds in the wind farms. The proposed approach provides a solution  for recording/preventing collision  of  birds  with  wind turbine. To  realize  the  system, a model is developed with two camera nodes. This model provides a 360 degree conical volumetric space coverage. The  system is  utilizing  a  servo  motor  with  capability  of  changing  speed  and arduino  embedded  platform for  controlling  different  functionality. After  triggering  the  camera nodes from arduino, software perform the image analysis.  This  cost  effective  and  wide  area  surveillance  system  consumes at  least half lower  power as compared  to traditional  linear  systems with  eight  cameras. It  is  worth  noting  that  the system provides  conical  shaped  area coverage  of  200  square  meters  with  height ranging  from approximately 40 to 100 meters. The system would capture images in raw format in local storage and  the test  results  show  that  saving  images  with  PNG  format  requires a small  memory  size  as compared  to BMP  and TIFF.    Experiment  results  indicate  that  surveillance  system  can  be effectively employed for detecting birds in opening places.

La surveillance est un concept central en soins infirmiers et ce, particulièrement à l’unité des soins intensifs. S’assurer que la condition du patient demeure stable ou s’améliore tout en assurant une surveillance continue est en fait la principale raison d’admission dans ce type d’unité. L’aménagement physique d’une unité de soins intensifs se compose typiquement d’un poste central et de lits ou chambres disposés autour de celui-ci. Au poste central se trouvent des moniteurs permettant aux infirmières de surveiller à distance la condition clinique des patients. La technologie, bien qu’étant considérée comme un outil essentiel dans la prestation des soins, peut donc limiter les contacts directs avec les patients. En effet, il est facile pour une infirmière de se fier aux données en évidence sur les moniteurs pour évaluer la condition d’un patient et ce, sans se rendre au chevet de celui-ci. Nous avons réalisé une ethnographie compte tenu de l’utilité de cette approche dans l’examen des pratiques d’une population donnée, à savoir le personnel infirmier exerçant aux soins intensifs. Elle nous a permis de saisir comment les infirmières exerçant dans ce milieu de soins incorporent la technologie dans leurs activités quotidiennes de surveillance des patients. Nous avons ensuite examiné dans quelle mesure ces activités contribuent à un processus d’objectivation de la personne, notamment lorsque celle-ci est inconsciente. Ce projet de recherche s’appuie sur le plan théorique sur les travaux de Giorgio Agamben, développés à partir de l’expérience humaine en camps de concentration. Les concepts d’état d’exception et de vie nue, en particulier, retiennent notre attention afin de théoriser certains processus infirmiers en lien avec la technologie aux soins intensifs. Ce choix théorique pour le moins radical repose sur la possibilité de comprendre une unité de soins intensifs en tant que version atténuée d’un espace au cœur duquel l’état d’exception peut être compris. Les résultats de cette étude montrent que les infirmières à l’unité des soins intensifs sont dépendantes de la haute technologie à leur disposition lorsqu’elles soignent leurs patients. Cependant, elles font valoir l’importance de leur rôle au chevet de la clientèle souffrant de conditions de santé instables. Les infirmières se montrent toutefois hésitantes quant à leur implication auprès des patients en fin de vie, notamment dans le processus de prise de décision. Selon notre cadre théorique, les infirmières affirment respecter la dimension bios de la personne malade, c'est-à-dire la façon dont celle-ci décide de vivre, guidée par ses valeurs, ses opinions, ses droits et sa liberté. Cela entraîne toutefois certains conflits avec des membres de l’équipe médicale qui ont des représentations autres des patients, représentations qui s’accordent davantage avec la dimension zoé dont discute Agamben. En dépit de ces différences, il convient de noter que la technologie avancée joue un rôle primordial dans la manière dont ces représentations se développent et influencent les dynamiques entre le personnel soignant, les patients et les familles.

New and emerging diseases in human and animal populations appear to be predominately associated with generalist pathogens that are able to infect multiple hosts. Carnivores are susceptible to a wide range of these pathogens and can act as effective samplers of their vertebrate prey, which are important reservoirs of many emerging diseases. This thesis evaluates the utility of carnivores as sentinels for pathogens present in their prey by exploration of four selected pathogen-prey-sentinel combinations in three rural study sites of varying habitat in northern England and Scotland over a twenty-two month period (2007-2009). Selected pathogens were Coxiella burnetii, Leptospira spp., Encephalitozoon cuniculi, and rabbit haemorrhagic disease virus (RHDV), selected prey species were wild rodents and rabbits, and selected carnivores were foxes, domestic cats and corvids. Seroprevalence to C.burnetii, Leptospira spp and E.cuniculi was assessed using adapted or novel test methodologies to enable their use for multiple mammalian species, however these were not applicable to corvids. RHDV seroprevalence was not assessed due to low acquisition of rabbit samples. Overall, seroprevalence to all three pathogens was significantly higher in predators than prey, at 24.2% and 12.4 % for C.burnetii, 22.73% and 1.95% for Leptospira spp and 39.06% and 5.31% for E.cuniculi in predator and prey species respectively. A similar pattern was found in all study areas and was consistent irrespective of individual prey or predator species, although serological evidence of exposure to E.cuniculi was not detected in domestic cats in any area. A semi-quantitative assessment of the time and financial costs of the study approach and application to hypothetical examples indicates that sampling carnivores is a much more costeffective approach to pathogen detection than sampling prey. The results indicate that carnivores can act as useful sentinels for broad-scale detection of pathogen presence and relative levels of prevalence in prey and predator populations. Careful selection of predator species and methods of sample acquisition are necessary to maximise their utility, and issues associated with diagnostic test performance and validation must also be acknowledged. Suggestions are made as to how this principle might be applied to future surveillance programmes. In addition, the study is the first report on the seroprevalence of C.burnetii, Leptospira spp and E.cuniculi in multiple wildlife species (field voles, bank voles, wood mice, foxes), the first detection of antibodies to C. burnetii in wildlife and cats, the first detection of antibodies to L mini, L hardjo prajitno and L hardjo bovis in wild rodents, and to L mini in cats, and the first detection of antibodies to E.cuniculi in wild rodents and foxes in the UK.

The RNA binding proteins Nrd1 and Nab3 function in transcription termination by RNA Pol II, acting via interactions with the CTD of the largest polymerase subunit, particularly on snRNA and snoRNA genes. They also participate in nuclear RNA surveillance and ncRNA degradation, functioning together with the exosome and the Trf-Air-Mtr4 polyadenylation (TRAMP) complexes. To better understand the signals for surveillance and ncRNA degradation, I applied an RNA-protein crosslinking approach in combination with Solexa sequencing. This approach identified in vivo binding sites for Nrd1, Nab3 and Trf4. Several million sequences were recovered and mapped to the yeast genome. This identified three classes of substrates: 1) Expected targets, including snRNAs, snoRNAs and characterized short ncRNAs. 2) Unknown but anticipated substrates, including several hundred previously uncharacterized ncRNAs that lie antisense to protein coding genes (asRNAs). 3) Unexpected targets, including many Pol III transcribed precursor RNAs. Bioinformatics analyses of the high-throughput sequencing data revealed that known binding motifs for Nrd1 and Nab3 were frequently recovered. Many recovered RNAs contained non-templated oligo(A) tails with an average of 2-5 nt length. This clearly distinguishes targets for surveillance machinery from polyadenylated mRNAs that get stabilized by polyadenylation (A70-90 in yeast). For a few selected, predicted asRNAs I was able to validate the crosslinking data by demonstrating that corresponding long RNAs are both detectable and increased by loss of Nrd1, Nab3, Trf4 or the exosome component Rrp6. Interestingly, loss of Nrd1 or Nab3 led to transcriptional read through on long asRNA transcripts. In addition, I have identified pre-TLC1 (telomerase RNA) as a target for the surveillance machinery. Processing of this long ncRNA was only poorly characterized in yeast but I could demonstrate that its transcription termination and maturation is mainly dependent on actions of the Nrd1-Nab3-Sen1, TRAMP4 and exosome complexes. It was previously reported that Nrd1-Nab3 acts only on short RNAs, due to the association with Ser5 phosphorylated CTD. My findings suggest that action of Nrd1- Nab3 is not exclusively on Ser5 phosphorylated form of the CTD. Unexpectedly the Pol II associated factors Nrd1 and Nab3 bound Pol III precursor transcripts. Surveillance of Pol III transcripts was dependent on Nrd1 and Nab3 since depletion of Nrd1 or Nab3 led to accumulation of pre-tRNAs. In addition, I could demonstrate that pre-RNase P RNA is oligoadenylated in vivo, which was dependent on Nrd1, Nab3 and Trf4. Together, my findings suggest a revised model of nuclear RNA surveillance in which Nrd1-Nab3 not only acts in co-transcriptional RNA recognition on Pol II transcripts but also post-transcriptionally on Pol III RNAs. The TRAMP complex is recruited to the defective RNA by the Nrd1-Nab3 complex, which remains associated with the RNA through the process of polyadenylation, until the exosome degrades the aberrant transcript.

This thesis examines how cinema engages with changing surveillance practices, and the hypothesised paradigm shift from discipline to control. The first part of the thesis outlines those changes in terms of three crucial areas in any discussion of surveillance – the organisation of the body, space and time. Since its publication in the 1970s, Michel Foucault’s Discipline and Punish has been a continuous influence on much social theory. However, recent developments in surveillance practices suggest that the Foucauldian model of discipline may no longer be the most appropriate theoretical framework within which to discuss contemporary modes of surveillance. In Postscript on Control Societies, written in 1990, Gilles Deleuze offers a possible new paradigm (the control society) through which to explore emerging trends in surveillance practices, often linked to the increasing use of digital technologies. While the paradigm of control does not simply replace that of discipline, it does help us to understand the development and amelioration of disciplinary structures. The second part of the thesis offers an original perspective on ongoing debates in surveillance studies concerning discipline and control, by investigating how this shift is articulated and reflected upon in a diverse range of films (from mainstream productions such as Enemy of the State to avant-garde ‘essays’ such as Harun Farocki’s Ich glaubte Gefangene zu sehen) which explicitly engage with changes in surveillance practice. It focuses specifically on the cinematic representation of the body, space and time in the context of the hypothesised transition from discipline to control, and addresses a series of important questions for cinema’s engagement with surveillance: can cinema, with its reliance on the visual image, address the emerging surveillance society which is increasingly invisible and, if so, what strategies does cinema use to achieve this; and, what is the implication of such strategies for the cinematic spectator? In conclusion, the thesis reflects on how cinema shapes our understanding of the emerging surveillance society.