La création des vidéos panoramiques de haute qualité pour des contenus immersifs en VR est généralement faite à l'aide d'un appareil doté de plusieurs caméras couvrant une scène cible. Malheureusement, cette configuration introduit à la fois des artefacts spatiaux et temporels dus à la différence entre les centres optiques et à la synchronisation imparfaite. Les mesures de qualité d'image traditionnelles ne peuvent pas être utilisées pour évaluer la qualité de ces vidéos, en raison de leur incapacité à capturer des distorsions géométriques. Dans cette thèse, nous proposons des méthodes pour l'évaluation objective des vidéos panoramiques basées sur le flux optique et la saillance visuelle. Nous validons cette métrique avec une étude centrée sur l'homme qui combine l'annotation d'erreurs percues et l'eye-tracking.Un défi important pour mesurer la qualité des vidéos panoramiques est le manque d'une vérité-terrain. Nous avons étudié l'utilisation des vidéos originales comme référence pour le panorama de sortie. Nous notons que cette approche n'est pas directement applicable, car chaque pixel du panorama final peut avoir une à $N$ sources correspondant à $N$ vidéos d'entrée avec des régions se chevauchant. Nous montrons que ce problème peut être résolu en calculant l'écart type des déplacements de tous les pixels sources à partir du déplacement du panorama en tant que mesure de la distorsion. Cela permet de comparer la différence de mouvement entre deux images données dans les vidéos originales et le mouvement dans le panorama final. Les cartes de saillance basées sur la perception humaine sont utilisées pour pondérer la carte de distorsion pour un filtrage plus précis.Cette méthode a été validée par une étude centrée sur l'homme utilisant une expérience empirique. L'expérience visait à déterminer si les humains et la métrique d'évaluation détectaient et mesuraient les mêmes erreurs, et à explorer quelles erreurs sont les plus importantes pour les humains lorsqu'ils regardent une vidéo panoramique.Les méthodes décrites ont été testées et validées et fournissent des résultats intéressants en ce qui concerne la perception humaine pour les mesures de qualité. Ils ouvrent également la voie à de nouvelles méthodes d'optimisation de l'assemblage vidéo, guidées par ces mesures de qualité. / High quality panoramic videos for immersive VR content are commonly created using a rig with multiple cameras covering a target scene. Unfortunately, this setup introduces both spatial and temporal artifacts due to the difference in optical centers as well as the imperfect synchronization. Traditional image quality metrics cannot be used to assess the quality of such videos, due to their inability to capture geometric distortions. In this thesis, we propose methods for the objective assessment of panoramic videos based on optical flow and visual salience. We validate this metric with a human-centered study that combines human error annotation and eye-tracking.An important challenge in measuring quality for panoramic videos is the lack of ground truth. We have investigated the use of the original videos as a reference for the output panorama. We note that this approach is not directly applicable, because each pixel in the final panorama can have one to N sources corresponding to N input videos with overlapping regions. We show that this problem can be solved by calculating the standard deviation of displacements of all source pixels from the displacement of the panorama as a measure of distortion. This makes it possible to compare the difference in motion between two given frames in the original videos and motion in the final panorama. Salience maps based on human perception are used to weight the distortion map for more accurate filtering.This method was validated with a human-centered study using an empirical experiment. The experiment was designed to investigate whether humans and the evaluation metric detect and measure the same errors, and to explore which errors are more salient to humans when watching a panoramic video.The methods described have been tested and validated and they provide interesting findings regarding human-based perception for quality metrics. They also open the way to new methods for optimizing video stitching guided by those quality metrics.

Dans cette thèse, nous avons proposé une approche analytique multi-échelle pour le word spotting dans les documents manuscrits. Le modèle proposé fonctionne selon deux niveaux différents. Un module de filtrage global permettant de définir plusieurs zones candidates de la requête dans le document testé. Ensuite, l’échelle de l’observation est modifiée à un niveau inférieur afin d’affiner les résultats et sélectionner uniquement ceux qui sont vraiment pertinents. Cette approche de word spotting est basée sur des familles généralisées de filtres de Haar qui s’adaptent à chaque requête pour procéder au processus de spotting et aussi sur un principe de vote qui permet de choisir l’emplacement spatial où les réponses générées par les filtres sont accumulées. Nous avons en plus proposé une autre approche pour l’extraction de texte du graphique dans les bandes dessinées. Cette approche se base essentiellement sur les caractéristiques pseudo-Haar qui sont générées par l’application des filtres généralisés de Haar sur l’image de bande dessinée. Cette approche est une approche analytique et ne nécessite aucun processus d’extraction ni des bulles ni d’autres composants. / The presented thesis follows two directions. The first one disposes a technique for text and graphic separation in comics. The second one points out a learning free segmentation free word spotting framework based on the query-by-string problem for manuscript documents. The two approaches are based on human perception characteristics. Indeed, they were inspired by several characteristics of human vision such as the Preattentive processing. These characteristics guide us to introduce two multi scale approaches for two different document analysis tasks which are text extraction from comics and word spotting in manuscript document. These two approaches are based on applying generalized Haar-like filters globally on each document image whatever its type. Describing and detailing the use of such features throughout this thesis, we offer the researches of document image analysis field a new line of research that has to be more explored in future. The two approaches are layout segmentation free and the generalized Haar-like filters are applied globally on the image. Moreover, no binarization step of the processed document is done in order to avoid losing data that may influence the accuracy of the two frameworks. Indeed, any learning step is performed. Thus, we avoid the process of extraction features a priori which will be performed automatically, taking into consideration the different characteristics of the documents.

This doctoral thesis is the result of a research effort performed in two fields of speech technology, i.e., speech recognition and mispronunciation detection. Although the two areas are clearly distinguishable, the proposed approaches share a common hypothesis based on psychoacoustic processing of speech signals. The conjecture implies that the human auditory periphery provides a relatively good separation of different sound classes. Hence, it is possible to use recent findings from psychoacoustic perception together with mathematical and computational tools to model the auditory sensitivities to small speech signal changes. The performance of an automatic speech recognition system strongly depends on the representation used for the front-end. If the extracted features do not include all relevant information, the performance of the classification stage is inherently suboptimal. The work described in Papers A, B and C is motivated by the fact that humans perform better at speech recognition than machines, particularly for noisy environments. The goal is to make use of knowledge of human perception in the selection and optimization of speech features for speech recognition. These papers show that maximizing the similarity of the Euclidean geometry of the features to the geometry of the perceptual domain is a powerful tool to select or optimize features. Experiments with a practical speech recognizer confirm the validity of the principle. It is also shown an approach to improve mel frequency cepstrum coefficients (MFCCs) through offline optimization. The method has three advantages: i) it is computationally inexpensive, ii) it does not use the auditory model directly, thus avoiding its computational cost, and iii) importantly, it provides better recognition performance than traditional MFCCs for both clean and noisy conditions. The second task concerns automatic pronunciation error detection. The research, described in Papers D, E and F, is motivated by the observation that almost all native speakers perceive, relatively easily, the acoustic characteristics of their own language when it is produced by speakers of the language. Small variations within a phoneme category, sometimes different for various phonemes, do not change significantly the perception of the language’s own sounds. Several methods are introduced based on similarity measures of the Euclidean space spanned by the acoustic representations of the speech signal and the Euclidean space spanned by an auditory model output, to identify the problematic phonemes for a given speaker. The methods are tested for groups of speakers from different languages and evaluated according to a theoretical linguistic study showing that they can capture many of the problematic phonemes that speakers from each language mispronounce. Finally, a listening test on the same dataset verifies the validity of these methods. / <p>QC 20120914</p> / European Union FP6-034362 research project ACORNS / Computer-Animated language Teachers (CALATea)

In this study, a P300 based Brain-Computer Interface (BCI) system design is realized by the implementation of the Spelling Paradigm. The main challenge in these systems is to improve the speed of the prediction mechanisms by the application of different signal processing and pattern classification techniques in BCI problems. The thesis study includes the design and implementation of a 10 channel Electroencephalographic (EEG) data acquisition system to be practically used in BCI applications. The electrical measurements are realized with active electrodes for continuous EEG recording. The data is transferred via USB so that the device can be operated by any computer. v Wiener filtering is applied to P300 Speller as a signal enhancement tool for the first time in the literature. With this method, the optimum temporal frequency bands for user specific P300 responses are determined. The classification of the responses is performed by using Support Vector Machines (SVM&rsquo / s) and Bayesian decision. These methods are independently applied to the row-column intensification groups of P300 speller to observe the differences in human perception to these two visual stimulation types. It is observed from the investigated datasets that the prediction accuracies in these two groups are different for each subject even for optimum classification parameters. Furthermore, in these datasets, the classification accuracy was improved when the signals are preprocessed with Wiener filtering. With this method, the test characters are predicted with 100% accuracy in 4 trial repetitions in P300 Speller dataset of BCI Competition II. Besides, only 8 trials are needed to predict the target character with the designed BCI system.

This diploma thesis deals with the issue of intermediation of abstract art in gallery education programs for primary school students. Its theoretical part outlines the process from figurative to non-figurative imaging through the theory and history of art. The term of abstraction is further expounded by general and developmental psychology. The final theoretical part deals with the intermediation of art and all its connections, including potential difficulties which can occur. The empirical part contains a research with gallery lecturers, which brings and analyzes their attitude to the process of gallery-educational programs providing intermediation of non-figurative art to pupils of younger school age.

Avec l’intégration des appareils photos dans les appareils mobiles, leur démocratisation et la réduction de la taille de l’imageur, de l’optique et de la taille pixels, les photos sont de plus en plus sujettes au flou de bougé dû aux tremblements de la main. À cette tendance s’ajoute un accroissement constaté dans l’exigence de qualité d’image de la part des utilisateurs. Pour réduire ce flou, des systèmes de stabilisation d’image ont été développés. Néanmoins ceux-ci ne permettent pas de garantir la qualité de netteté des images et souffrent parfois d’une intégration limitée. En réponse à ces limitations, ces travaux de recherche proposent, d’une part, un modèle de tremblement physiologique permettant de simuler de manière fidèle les flous de bougé et, d’autre part, une étude sur la perception visuelle du flou permettant le développement d’une métrique de qualité. Enfin des architectures de stabilisations, exploitant ces nouveaux outils, sont proposées. Ces nouvelles architectures permettent de réduire le nombre de composants externes ainsi que de garantir la netteté des images stabilisées. / With the integration of cameras in mobile devices, their democratization and the reduction of the imager’s size, the optical system dimensions and the pixels miniaturization, the photos become more and more subject to motion blur due to the hand tremor. In addition, the requirements in terms of image quality become higher and higher. Hence, in order to reduce this blur, several image stabilization systems have been developed. Nevertheless, they cannot guarantee the sharpness quality of resulting images and in some cases, they show integration difficulties. In order to overcome these limitations, the research work presented in this thesis proposes, first of all, a physiological tremor model that aims to simulate realistic camera shake and secondly, presents a study on visual perception of blur. This study enables the development of a quality metric. Finally, stabilization algorithms and architectures exploiting these new tools are presented. These new architectures reduce the number of external components and ensure sharp stabilized images.

Denna studie har genomförts i form av en undersökning med 25 respondenter. Arbetet harsyftat till att fackmän inom belysningsbranschen samt privatpersoner och intressenter ska fåen indikation på hur människan uppfattar ljuset från fasdimrade LED-ljuskällor i privata hem.Studien har studerat två frågeställningar gällande människans upplevelse av flimmer blandannat hur känslig människan är för flimmer hos en ljusreglerad LED-ljuskälla innan detuppfattas som obekvämt. Vetenskaplig litteratur har använts för att få en förståelse och djuparekunskap kring problematiken gällande människans hälsopåverkan vid exponering av flimmer,samt vilka riktlinjer och möjliga tillvägagångssätt som finns för att lösa denna problematik. Föratt ta reda på hur väl människan uppfattar flimmer samt hur upplevelsen av obekvämlighetenpåverkas har en enkät tagits fram till studien. Undersökningen konstaterar att människoruppfattar flimmer i olika mängd, trots frekvenser över 90Hertz (Hz). Genom data från digitalamätningar och inhämtad empiri från arbetets undersökningsstudie kan man konstatera att detinte enbart är frekvensnivån som påverkar människans uppfattning av flimmer, utan ävenljuskällans andel flimmerprocent. Ett samband mellan upplevd obekvämlighet och flimmer kananas, men inte fastslås som ett tydligt resultat. Arbetet avgränsades genom att undersökningenenbart utfördes under en kortare period med ett färre antal respondenter. / This study has been carried out in the form of a survey with 25 respondents. The work has aimedto give professionals and individuals an indication of how people perceive the light from phasedimmedLED light sources in private homes. The study has investigated two issues regardingthe human experience of flicker, and how sensitive human beings are to flicker from a phasedimmedLED light source before it will be perceived as inconvenient. Scientific literature hasbeen reviewed to gain an understanding and deeper knowledge about the impact of flicker onhuman health, and which guidelines and possible approaches there are to solve the problem. Asurvey has been developed to find out if flicker is perceived by humans, and how the perceptionof inconvenience is experienced. The study found out that people perceive flicker differently,despite measured frequencies above 90 Hertz (Hz). Through data from digital measurementsand acquired empirical evidence from the study’s results, it is not only the frequency level thataffects humans perception of flicker, but also the percentage of light emitted from the lightsource. A correlation between perceived inconvenience and flicker can be noted, but notdetermined as a clear result. The work was limited by only investigating a small population ofrespondents over a short period of time.

How can light positively influence and encourage pedestrians’ engagement and interaction with the urban environments at night? In this Master Thesis, I questioned how to develop nighttime urban planning from a socially-oriented approach. In order to answer this question, I studied different evidence such as two publications, three lighting designers’ and a lighting studio’ approach; three case studies analyses, two of them located in Colombia (Cartagena and Medellín) and one in Sweden (Stockholm);and my own qualitative observation and quantitative measurements studied between April and May at Norrmalmstorg and Biblioteksgatan in Stockholm, Sweden. From that review, I propose a Guideline consisted of three sections: (1) Main dimensions, (2) lighting attributes, and (3) lighting systems in urban planning. In general, this guideline is a framework to develop the analytical tools for various design stages in nighttime urban planning.

How can reimagining the way we light our cities prove to be the urban catalyst to meet targets set out in the Paris Climate Agreement? In this Master Thesis, I question how restructuring our urban lighting can provide enormous benefits for not only reducing our energy consumption and becoming more efficient societies, but also support social interaction in shaping our cities for the future. The aspects pertaining to urban lighting and its influence on our environment have been organised into economy, political and urban evolution (social) categories, as a way to dissect and appreciate lights associated role in shaping our societies. Seeking sustainable strategies to meet urgent energy targets and urban development initiatives, are no longer optional but necessary. However, even with this undisputed need for change, particularly within urban lighting, approaches taken are still very much sequestered from general understanding and lack cross-disciplinary cooperation. Extensive publications, works and teachings from lighting designers, as well as currently adopted policies from government bodies and international case studies have been considered, in order to understand current strategies adopted and their potential co-operation for a future holistic strategy in tackling climate change as well as apparent inequalities in our urban environments.

Failures in networks that reside in business environments cause harm to organizations depending on them. Every minute of inoperativety is hurtful and as a network adminstrator you want to minimize the rates of failures as well as the time of inoperation. Therefore, a fruitful network monitoring system is of great interest for such organizations. This bachelor’s thesis is the outcome of a pre-study performed on behalf of MIC Nordic and sought to advice them in the implementation of a new monitoring system. The aim of this study was to investigate how a Network Operation Center (NOC) can be implemented in an existing monitoring environment, to integrate current monitoring systems to a central point for monitoring. This study takes an holitstic approach to network management and the research can be divided into two main categories: Communication between network components and Presentation of information. Our process involves an analysis of the environment of MIC Nordic and an in depth inquiry on the current state of network monitoring and interface design. The study then culminates in the implementation of a prototype. The prototype serves in first hand as a research tool to collect experience and empirical evidence to increase the crediblity of our conclusions. It is also an attempt of demonstrating the complete process behind developing a NOC, that we believe can fill a gap among the previous research in the field. From our results you can expect a prototype with functionality for monitoring network components and a graphical user interface (GUI) for displaying information. The results are designed towards solving the specific network management problem that was given and the environment that it concerns. This pre-study suggests that the best solution for implementing a NOC in the given environment is to use SNMP for communication. From an investigation on how to present network management information in a effective way we propose to follow a user-centered approach and to utilize human perception theory in the design process. The authors recommend further research that maintain the holistic approach but applies more quantitative methods to broaden the scope. / Störningar i nätverk som används i företagsmiljöer skapar problem för organisationer som är beroende av dess funktion. Varje minut som nätverket är verkningslöst är ofördelaktigt och som nätverksadministratör så vill du minimera antal störningar och tiden då nätverket är verkningslöst. Ett effektivt nätverksövervaknings system är därf ör av stort intresse f ör organisationer beroende av ett funktionerande nätverk. Den här rapporten är resultatet av en förundersökning som utfördes på uppdrag av MIC Nordic, för att ge en rekommendation om hur ett nytt övervakningssystem för deras nätverk kan implementeras. Målet med studien var att undersöka hur ett Network Operation Center (NOC) kan implementeras i en existerande miljö för att integrera nuvarande övervakningssystem till en central punkt för övervakning. Den här studien tar ett holistiskt grepp om nätverksövervakning och undersökningen kan delas in i två primära kategorier: Kommunikation mellan nätverkskomponenter och Presentation av information. Vår process involverar en analys av MIC Nordics miljö och en djupgående utredning om nätverksövervakning samt gränssnitts design. Studien kulminerar i en implementaion av en prototyp. Prototypen är i f örsta hand ett undersökningsverktyg för att samla på oss erfarenhet och empiriska belägg för att öka trovärdigheten i våra slutsatser. Prototypen utgör även ett försök av författarna att demonstrera den kompletta proceduren av att utveckla en NOC, avsikten är att det kan fylla ett behov bland tidigare avhandligar i ämnet. Resultatet innehåller en prototyp med funktionalitet för att övervaka nätverkskomponenter samt ett grafiskt gränssnitt för att visa informationen. Resultaten är designade mot en lösning som är specifik för problemet som gavs av MIC Nordic och deras miljö. Denna förstudie proponerar att den bästa lösningen för att implementera en NOC i den givna miljön är att använda SNMP för kommunikation. Efter en granskning av hur man kan presentera information som rör nätverksövervakning på ett effektivt sätt så lanserar författarna en användarcentrerad metod som utnyttjar läran om hur människor uppfattar saker och ting. Författarna uppmuntrar vidare undersökningar som bibehåller det holistiska greppet men som applicerar mer kvantitativa metoder för att utöka undersökningens omfattning.