An eye for an eye: robust motion segmentation by applying the latest in human vision research

Automatically extracting interesting objects from videos is a very challenging task and is applicable to many research areas such robotics, medical imaging, content based indexing and visual surveillance. Automated visual surveillance is a major research area in computational vision and a commonly applied technique in an attempt to extract objects of interest is that of motion segmentation. Motion segmentation relies on the temporal changes that occur in video sequences to detect objects, but as a technique it presents many challenges that researchers have yet to surmount. Changes in real-time video sequences not only include interesting objects, environmental conditions such as wind, cloud cover, rain and snow may be present, in addition to rapid lighting changes, poor footage quality, moving shadows and reflections. The list provides only a sample of the challenges present. This thesis explores the use of motion segmentation as part of a computational vision system and provides solutions for a practical, generic approach with robust performance, using current neuro-biological, physiological and psychological research in primate vision as inspiration.

Reconstruction from image correspondences

A single picture provides a largely incomplete representation of the scene one is looking at. Usually it reproduces only a limited spatial portion of the scene according to the standpoint and the viewing angle, besides it contains only instantaneous information. Thus very little can be understood on the geometrical structure of the scene, the position and orientation of the observer with respect to it remaining also hard to guess. When multiple views, taken from different positions in space and time, observe the same scene, then a much deeper knowledge is potentially achievable. Understanding inter-views relations enables construction of a collective representation by fusing the information contained in every single image. Visual reconstruction methods confront with the formidable, and still unanswered, challenge of delivering a comprehensive representation of structure, motion and appearance of a scene from visual information. Multi-view visual reconstruction deals with the inference of relations among multiple views and the exploitation of revealed connections to attain the best possible representation. This thesis investigates novel methods and applications in the field of visual reconstruction from multiple views. Three main threads of research have been pursued: dense geometric reconstruction, camera pose reconstruction, sparse geometric reconstruction of deformable surfaces. Dense geometric reconstruction aims at delivering the appearance of a scene at every single point. The construction of a large panoramic image from a set of traditional pictures has been extensively studied in the context of image mosaicing techniques. An original algorithm for sequential registration suitable for real-time applications has been conceived. The integration of the algorithm into a visual surveillance system has lead to robust and efficient motion detection with Pan-Tilt-Zoom cameras. Moreover, an evaluation methodology for quantitatively assessing and comparing image mosaicing algorithms has been devised and made available to the community. Camera pose reconstruction deals with the recovery of the camera trajectory across an image sequence. A novel mosaic-based pose reconstruction algorithm has been conceived that exploit image-mosaics and traditional pose estimation algorithms to deliver more accurate estimates. An innovative markerless vision-based human-machine interface has also been proposed, so as to allow a user to interact with a gaming applications by moving a hand held consumer grade camera in unstructured environments. Finally, sparse geometric reconstruction refers to the computation of the coarse geometry of an object at few preset points. In this thesis, an innovative shape reconstruction algorithm for deformable objects has been designed. A cooperation with the Solar Impulse project allowed to deploy the algorithm in a very challenging real-world scenario, i.e. the accurate measurements of airplane wings deformations.

MultiPARTES: Multicore Virtualization for Mixed-Criticality Systems

Modern embedded applications typically integrate a multitude of functionalities with potentially different criticality levels into a single system. Without appropriate preconditions, the integration of mixed-criticality subsystems can lead to a significant and potentially unacceptable increase of engineering and certification costs. A promising solution is to incorporate mechanisms that establish multiple partitions with strict temporal and spatial separation between the individual partitions. In this approach, subsystems with different levels of criticality can be placed in different partitions and can be verified and validated in isolation. The MultiPARTES FP7 project aims at supporting mixed- criticality integration for embedded systems based on virtualization techniques for heterogeneous multicore processors. A major outcome of the project is the MultiPARTES XtratuM, an open source hypervisor designed as a generic virtualization layer for heterogeneous multicore. MultiPARTES evaluates the developed technology through selected use cases from the offshore wind power, space, visual surveillance, and automotive domains. The impact of MultiPARTES on the targeted domains will be also discussed. In a number of ongoing research initiatives (e.g., RECOMP, ARAMIS, MultiPARTES, CERTAINTY) mixed-criticality integration is considered in multicore processors. Key challenges are the combination of software virtualization and hardware segregation and the extension of partitioning mechanisms to jointly address significant non-functional requirements (e.g., time, energy and power budgets, adaptivity, reliability, safety, security, volume, weight, etc.) along with development and certification methodology.

Fast and robust face finding via local context

[EN]In visual surveillance face detection can be an important cue for initializing tracking algorithms. Recent work in psychophics hints at the importance of the local context of a face for robust detection, such as head contours and torso. This paper describes a detector that actively utilizes the idea of local context. The promise is to gain robustness that goes beyond the capabilities of traditional face detection making it particularly interesting for surveillance. The performance of the proposed detector in terms of accuracy and speed is evaluated on data sets from PETS 2000 and PETS 2003 and compared to the object-centered approach. Particular attention is paid to the role of available image resolution.

A three-dimensional representation method for noisy point clouds based on growing self-organizing maps accelerated on GPUs

The research described in this thesis was motivated by the need of a robust model capable of representing 3D data obtained with 3D sensors, which are inherently noisy. In addition, time constraints have to be considered as these sensors are capable of providing a 3D data stream in real time. This thesis proposed the use of Self-Organizing Maps (SOMs) as a 3D representation model. In particular, we proposed the use of the Growing Neural Gas (GNG) network, which has been successfully used for clustering, pattern recognition and topology representation of multi-dimensional data. Until now, Self-Organizing Maps have been primarily computed offline and their application in 3D data has mainly focused on free noise models, without considering time constraints. It is proposed a hardware implementation leveraging the computing power of modern GPUs, which takes advantage of a new paradigm coined as General-Purpose Computing on Graphics Processing Units (GPGPU). The proposed methods were applied to different problem and applications in the area of computer vision such as the recognition and localization of objects, visual surveillance or 3D reconstruction.

Inconsistencies in Guidelines for Visual Health Surveillance of VDT Workers

Objectives: In Europe, 25% of workers use video display terminals (VDTs). Occupational health surveillance has been considered a key element in the protection of these workers. Nevertheless, it is unclear if guidelines available for this purpose, based on EU standards and available evidence, meet currently accepted quality criteria. The aim of this study was to appraise three sets of European VDT guidelines (UK, France, Spain) in which regulatory and evidence-based approaches for visual health have been formulated and recommendations for practice made. Methods: Three independent appraisers used an adapted AGREE instrument with seven domains to appraise the guidelines. A modified nominal group technique approach was used in two consecutive phases: first, individual evaluation of the three guidelines simultaneously, and second, a face-to-face meeting of appraisers to discuss scoring. Analysis of ratings obtained in each domain and variability among appraisers was undertaken (correlation and kappa coefficients). Results: All guidelines had low domain scores. The domain evaluated most highly was Scope and purpose, while Applicability was scored minimally. The UK guidelines had the highest overall score, and the Spanish ones had the lowest. The analysis of reliability and differences between scores in each domain showed a high level of agreement. Conclusions: These results suggest current guidelines used in these countries need an update. The formulation of evidence-base European guidelines on VDT could help to reduce the significant variation of national guidelines, which may have an impact on practical application.

Burden Of Focal Cryoablation Versus Brachytherapy Versus Active Surveillance In The Treatment Of Very Low-risk Prostate Cancer: A Preliminary Head-to-head Comprehensive Assessment.

Focal cryoablation (FC), brachytherapy (B) and active surveillance (AS) were offered to patients diagnosed with very low-risk prostate cancer (VLRPC) in an equal access protocol. Comprehensive validated self-report questionnaires accessed patients' erectile (IIEF-5) and voiding (IPSS) functions, Beck scales measured anxiety (BAI), hopelessness (BHS) and depression (BDI), SF-36 reflected patients' quality of life added to the emotional thermometers including five visual analogue scales (distress, anxiety, depression, anger and need for help). Kruskal-Wallis or ANOVA tests and Spearman's correlations were obtained among groups and studied variables. Thirty patients were included, median follow-up 18 months (15-21). Those on AS (n = 11) were older, presented higher hopelessness (BHS) and lower general health perceptions (SF-36) scores than patients opting for FC (n = 10) and B (n = 9), P = 0.0014, P = 0.0268 and P = 0.0168 respectively. Patients on B had higher IPSS scores compared to those under FC and AC, P = 0.0223. For all 30 included patients, Spearman's correlation (rs ) was very strong between BHS and general health perceptions (rs  = -0.800, P < 0.0001), and weak/moderate between age and BHS (rs  = 0.405, P = 0.026) and age and general health perceptions (rs  = -0.564, P = 0.001). The sample power was >60%. To be considered in patients' counselling and care, current study supports the hypothesis that even VLRPC when untreated undermines psychosocial domains.

Copwatching et perception publique de la police. L'intervention policière comme performance sous surveillance

Envisagée comme une situation sociale de rassemblement en rue, l'intervention policière se caractérise par sa nature publique et observable. Le « copwatching », en tant que pratique amateur de captation vidéo des interventions de police, problématise la visibilité des agents sur la voie publique en les soumettant à une surveillance médiatique. M'appuyant sur des exemples vidéos, j'explore les dimensions visuelles et symboliques du travail policier. De la performance policière en rue à la figuration médiatique du policier, l'étude des médiations entre action et image permet d'interroger la perception publique de la police engagée dans les rassemblements. Abstract Cop-watching and the public perception of the police. Police intervention as performance under surveillance. This paper deals with police work as a public performance on the street. Cop-watching, a way of publicly observing and documenting police activities, reveals the complex relationship between public perceptions of law enforcement and the visual nature of much police work. Three methodological approaches to cop-watching videos allow us to highlight the visual dimensions of police work : the study of mass-media images of cops, the study of street performance by real cops, and the study of the ways people make sense of the visual symbolic environment present in everyday police work.

Temporalité et surveillance dans l'art de David Rokeby

Pour respecter les droits d'auteur, la version électronique de ce mémoire à été dépouillée de ses documents visuels et audio-visuels. La version intégrale du mémoire a été déposée au Service de la gestion des documents et des archives de l'Université de Montréal.