This study explores the human and technical limitations of looking and seeing. It proposes a model for design that expands technical sight toward harmony with our human notion. It proposes a model for design that expands technical sight toward harmony with our human notion. This study is guided by the phenomenological experience of being expressed primarily by Heidegger as well as neuro-physiological research on the mind and body relationship by Ramachandran, Sacks Nicolelis and Damasio. It examines, in two paths, the technical developments that seek to alter or enhance our ways of looking and seeing. The first path is an assessment of ways of looking with optics-based cameras that includes how cameras might be set to look, how they behave in looking and how they translate that look into an image on display. The second path is an assessment of the image in varying states of readiness which include the capture state, state of rendering (for view) and state of display. The study uncovers the various ways that images are translated to be seen, and how sight and ocular vision might be detached in the process of imprinting what is seen in the imagination. It includes key examples of modern image device capabilities, makes suggestions about how the framework of this study can be applied in specific cases and predicts the state of image devices in the future.

Avec l'apparition récente des caméras 3D, des perspectives nouvelles pour différentes applications de l'interprétation de scène se sont ouvertes. Cependant, ces caméras ont des limites qui affectent la précision de leurs mesures. En particulier pour les caméras Temps-de-Vol, deux types d'erreur peuvent être distingués : le bruit statistique de la caméra et la distorsion de la mesure de profondeur. Dans les travaux de la littérature des caméras Temps-de-Vol, le bruit est peu étudié et les modèles de distorsion de la mesure de profondeur sont généralement difficiles à mettre en œuvre et ne garantissent pas la précision requise pour certaines applications. L'objectif de cette thèse est donc d'étudier, modéliser et proposer un étalonnage précis et facile à mettre en œuvre de ces 2 types d'erreur des caméras Temps-de-Vol. Pour la modélisation du bruit comme pour la distorsion de la mesure de profondeur, deux solutions sont proposées présentant chacune une solution à un problème différent. La première vise à fournir un modèle précis alors que le second favorise la simplicité de la mise en œuvre. Ainsi, pour le bruit, alors que la majorité des modèles reposent uniquement sur l'information d'amplitude, nous proposons un premier modèle qui intègre aussi la position du pixel dans l'image. Pour encore une meilleure précision, nous proposons un modèle où l'amplitude est remplacée par la profondeur de l'objet et le temps d'intégration. S'agissant de la distorsion de la mesure de profondeur, nous proposons une première solution basée sur un modèle non-paramétrique garantissant une meilleure précision. Ensuite, pour fournir une solution plus facile à mettre en œuvre que la précédente et que celles de l'état de l'art, nous nous basons sur la connaissance à priori de la géométrie planaire de la scène observée. / 3D cameras open new possibilities in different fields such as 3D reconstruction, Augmented Reality and video-surveillance since they provide depth information at high frame-rates. However, they have limitations that affect the accuracy of their measures. In particular for TOF cameras, two types of error can be distinguished : the stochastic camera noise and the depth distortion. In state of the art of TOF cameras, the noise is not well studied and the depth distortion models are difficult to use and don't guarantee the accuracy required for some applications. The objective of this thesis is to study, to model and to propose a calibration method of these two errors of TOF cameras which is accurate and easy to set up. Both for the noise and for the depth distortion, two solutions are proposed. Each of them gives a solution for a different problem. The former aims to obtain an accurate model. The latter, promotes the simplicity of the set up. Thereby, for the noise, while the majority of the proposed models are only based on the amplitude information, we propose a first model which integrate also the pixel position in the image. For a better accuracy, we propose a second model where we replace the amplitude by the depth and the integration time. Regarding the depth distortion, we propose a first solution based on a non-parametric model which guarantee a better accuracy. Then, we use the prior knowledge of the planar geometry of the observed scene to provide a solution which is easier to use compared to the previous one and to those of the litterature.

A research report submitted to the faculty of Engineering and the built environment, University of the Witwatersrand, Johannesburg, in partial fulfilment of the requirements for the degree of Master of Science in Engineering. Johannesburg, August 2016 / There has been much interest in using the Microsoft Xbox 360 Kinect cameras for visual servo control applications. It is a relatively cheap device with expected shortcomings. This work contributes to the practical considerations of using the Kinect for visual servo control applications. A comprehensive characterisation of the Kinect is synthesised from existing literature and results from a nonlinear calibration procedure. The Kinect reduces computational overhead on image processing stages, such as pose estimation or depth estimation. It is limited by its 0.8m to 3.5m practical depth range and quadratic depth resolution of 1.8mm to 35mm, respectively. Since the Kinect uses an infra-red (IR) projector, a class one laser, it should not be used outdoors, due to IR saturation, and objects belonging to classes of non- IR-friendly surfaces should be avoided, due to IR refraction, absorption, or specular reflection. Problems of task stability due to invalid depth measurements in Kinect depth maps and practical depth range limitations can be reduced by using depth map preprocessing and activating classical visual servoing techniques when Kinect-based approaches are near task failure. / MT2017

L’étude et la compréhension du comportement humain est devenue l’une des problématiques majeures dans différents secteurs d’activités. Ce besoin de comprendre les habitudes des individus a conduit plusieurs entreprises vers l’utilisation de vidéos pour l’analyse et l’interprétation des comportements. Ces raisons ont conduit à l’émergence de travaux de recherches qui ont pour objectif l’automatisation de ces procédures. De ce fait, l’étude du comportement humain est devenue l’un des principaux sujets de recherche dans le domaine de la vision par ordinateur, et de nombreuses solutions d’analyse du comportements basées sur l’utilisation de l’intelligence artificielle ont émergé.Dans ce travail, notre objectif est le développement d’un système qui va permettre de suivre simultanément plusieurs individus dans un réseau multi-caméras dans le contexte de l’analyse comportementale. Pour cela, nous avons proposé un système de suivi qui est composé de trois modules principaux et d’un module de gestion. Le premier est un module de comptage pour mesurer les entrées. Le deuxième module, basé sur l’utilisation de filtres à particules, est un système de suivi mono-caméra destiné à suivre les individus indépendamment dans chacune des caméras. Le troisième module, basé sur la sélection des régions saillantes de chaque individu, sert à la ré-identification et permet d’associer les individus détectés dans les différentes caméras. Enfin, le module de gestion est conçu pour créer des trajectoires sémantiques à partir des trajectoires brutes obtenues précédemment. / The study and the understanding of human behavior has become one of the major concerns for various reasons in different sectors of activity. This need to understand the habits of people led several big firms towards the use of videos surveillance for analyzing and interpreting behaviors. These reasons led to the emergence of research aimed at automating these procedures. As a result, the study of human behavior has become the main subject of several researches in the field of computer vision. Thus, a variety of behavior analysis solutions based on artificial intelligence emerged.In this work, our objective is the proposal of a solution that enable the simultaneous track of several individuals in a multi-camera network in order to reconstruct their trajectories in the context of behavioral analysis. For this, we have proposed a system that is made of three main modules and a management module. The first module is a counting module to measure entries. The second module is a mono-camera tracking system that is based on the use of particle filtering to track individuals independently in each camera. The third module is a re-identification module which is based on the selection of salient regions for each individual. It enables the association of the individuals that are detected in the different cameras. The last module which is the management module is based on the use of ontologies for interpreting trajectories. This module is designed to create semantic trajectories from raw trajectories obtained previously.

The purpose of this project is to implement, analyze and compare visual odometry approaches to help the localization task in autonomous vehicles. The stereo visual odometry algorithm Libviso2 is compared with a proposed omnidirectional multi-camera approach. The proposed method consists of performing monocular visual odometry on all cameras individually and selecting the best estimate through a voting scheme involving all cameras. The omnidirectionality of the vision system allows the part of the surroundings richest in features to be used in the relative pose estimation. Experiments are carried out using cameras Bumblebee XB3 and Ladybug 2, fixed on the roof of a vehicle. The voting process of the proposed omnidirectional multi-camera method leads to some improvements relatively to the individual monocular estimates. However, stereo visual odometry provides considerably more accurate results. / O objetivo deste mestrado é implementar, analisar e comparar abordagens de odometria visual, de forma a contribuir para a localização de um veículo autônomo. O algoritmo de odometria visual estéreo Libviso2 é comparado com um método proposto, que usa um sistema multi-câmera omnidirecional. De acordo com este método, odometria visual monocular é calculada para cada câmera individualmente e, seguidamente, a melhor estimativa é selecionada através de um processo de votação que involve todas as câmeras. O fato de o sistema de visão ser omnidirecional faz com que a parte dos arredores mais rica em características possa sempre ser usada para estimar a pose relativa do veículo. Nas experiências são utilizadas as câmeras Bumblebee XB3 e Ladybug 2, fixadas no teto de um veículo. O processo de votação do método multi-câmera omnidirecional proposto apresenta melhorias relativamente às estimativas monoculares individuais. No entanto, a odometria visual estéreo fornece resultados mais precisos.

"Patient motion, which causes artifacts in reconstructed images, can be a serious problem in Single Photon Emission Computed Tomography (SPECT) imaging. If patient motion can be detected and quantified, the reconstruction algorithm can compensate for the motion. A real-time multi-threaded Visual Tracking System (VTS) using optical cameras, which will be suitable for deployment in clinical trials, is under development. The VTS tracks patients using multiple video images and image processing techniques, calculating patient motion in three-dimensional space. This research aimed to develop and implement an algorithm for feature matching and stereo location computation using multiple cameras. Feature matching is done based on the epipolar geometry constraints for a pair of images and extended to the multiple view case with an iterative algorithm. Stereo locations of the matches are then computed using sum of squared distances from the projected 3D lines in SPECT coordinates as the error metric. This information from the VTS, when coupled with motion assessment from the emission data itself, can provide a robust compensation for patient motion as part of reconstruction."

Sistemas de Interação Humano-Computador baseados em toque são uma tecnologia disseminada em tablets, smartphones e notebooks. Trata-se de um grande avanço que aumenta a facilidade de comunicação e, ao mesmo tempo, diminui a necessidade de interfaces como mouse e teclado. Entretanto, a superfície de interação utilizada por esses sistemas normalmente é equipada com sensores para a captação dos movimentos realizados pelo usuário, o que impossibilita transformar uma superfície planar qualquer (uma mesa, por exemplo) em uma superfície de interação. Por outro lado, a popularização de sensores de profundidade a partir do lançamento do Microsoft Kinect propiciou o desenvolvimento de sistemas que adotam objetos do dia a dia como superfícies de interação. Nesta dissertação é proposta uma interface natural para interação com superfícies planares utilizando uma câmera RGB-D em posição descendente. Inicialmente, o plano de interação é localizado na nuvem de pontos 3D através de uma variação do algoritmo RANSAC com coerência temporal. Objetos acima do plano são segmentados a partir da transformada watershed baseada em uma função de energia que combina cor, profundidade e informação de confiança. A cor de pele é utilizada para isolar as mãos, e os dedos que interagem com o plano são identificados por um novo processo de esqueletonização 2D. Finalmente, as pontas dos dedos são rastreadas com o uso do algoritmo Húngaro, e o filtro de Kalman é usado para produzir trajetórias mais suaves. Para demonstrar a utilidade da técnica, foi desenvolvido um protótipo que permite ao usuário desenhar em uma superfície de forma natural e intuitiva. / Touch-based Human-Computer Interfaces (HCIs) are a widespread technology present in tablets, smartphones, and notebooks. This is a breakthrough which increases the ease of communication and at the same time reduces the need for interfaces such as mouse and keyboard. However, the interaction surface used by these systems is usually equipped with sensors to capture the movements made by the user, making it impossible to substitute this surface by any other such as a table, for example. On the other hand, the progress of commercial 3D depth sensing technologies in the past five years, having as a keystone Microsoft’s Kinect sensor, has increased the interest in 3D hand gesture recognition using depth data. In this dissertation, we present a natural Human-Computer Interface (HCI) for interaction with planar surfaces using a topdown RGB-D camera. Initially, the interaction plane is located in the 3D point cloud by using a variation of RANSAC with temporal coherence. Off-plane objects are segmented using the watershed transform based on an energy function that combines color, depth and confidence information. Skin color information is used to isolate the hand(s), and a novel 2D skeletonization process identifies the interaction fingers. Finally, the fingertips are tracked using the Hungarian algorithm, and a Kalman filter is applied to produce smoother trajectories. To demonstrate the usefulness of the technique, we also developed a prototype in which the user can draw on the surface using lines and sprays in a natural way.

O principal intento deste trabalho é demonstrar como a combinação de resultados de modelos radiométricos e de análises do limite de resolução de imagens produzem impactos positivos não apenas no projeto óptico de retinógrafos de alta resolução, mas que também geram ferramentas que auxiliam na definição da arquitetura de todo o equipamento, considerando, inclusive, os limites impostos por normas internacionais para a segurança do paciente em relação a radiação emitida na realização dos exames. Particularmente ao projeto óptico, foram desenvolvidos os sistemas de iluminação e imageamento da retina, onde se destacam duas propostas de metodologias: a primeira para a implementação da técnica de eliminação das bem conhecidas retro-reflexões geradas no conjunto óptico que é comum à iluminação e ao imageamento da retina, e a segunda para a obtenção de soluções efetivas em sistemas imageadores, que prima pela diminuição da sensibilidade dos componentes aos erros que são introduzidos pelos processos de fabricação e montagem. Analisando os resultados obtidos com o protótipo, foram validados os sistemas e modelos propostos, que se tornaram parte integrante de um produto oftalmológico aprovado para a comercialização, o retinógrafo Kiron. / The fundamental objective of this work is to demonstrate how radiometric models and analysis of the limit of resolution can provide not only positive impacts to the optical project of high resolution fundus cameras, but also creates tools to perform an optimization of all the equipment, considering the limits imposed by international standards to ensure security to the patient regarding the equipment radiation emission. In the optical project in particular, the imaging and illuminating systems of the retina were developed, in which stands out two proposed methodologies: first, to eliminate the well known retro-reflections in the common optical assembly of the imaging and illuminating systems, and a second one that helps the optical designer to find effective solutions in imaging systems, focused in the reduction of the component sensitivity to the fabrication and assembly processes errors. Analyzing the results from the prototypes the systems and the models proposed were validated, becoming integrant part of an ophthalmic product approved for commercialization, the Kiron fundus camera.

The first part of this thesis presents a convenient and flexible calibration method to estimate the relative rotation and translation among multiple cameras. A simple planar pattern is used for accurate calibration and is not required to be simultaneously observed by all cameras. Thus the method is especially suitable for widely spaced camera array. In order to fairly evaluate the calibration results for different camera setups, a novel accuracy metric is introduced based on the deflection angles of projection rays, which is insensitive to a number of setup factors. / The objective of this thesis is to develop a multiview system that can synthesize photorealistic novel views of the scene captured by sparse cameras distributed in a wide area. The system cost is largely reduced due to the small number of required cameras, and the image capture is greatly facilitated because the cameras are allowed to be widely spaced and flexibly placed. The key techniques to achieve this goal are investigated in this thesis. / Cui, Chunhui. / "November 2010." / Adviser: Ngan King Ngi. / Source: Dissertation Abstracts International, Volume: 73-04, Section: B, page: . / Thesis (Ph.D.)--Chinese University of Hong Kong, 2011. / Includes bibliographical references (leaves 140-155). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. [Ann Arbor, MI] : ProQuest Information and Learning, [201-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstract also in Chinese.

Thesis (M.S.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 1979. / MICROFICHE COPY AVAILABLE IN ARCHIVES AND ENGINEERING. / Bibliography: leaves 177-181. / by Elazer Reuven Edelman. / M.S.