Networks that solve specific visual tasks, such as the evaluation of spatial relations with hyperacuity precision, can be eastily synthesized from a small set of examples. This may have significant implications for the interpretation of many psychophysical results in terms of neuronal models.

The interactions between illusory and real contours have been inves- tigated under monocular, binocular and dichoptic conditions. Results show that under all three presentation conditions, periodic alternations, generally called rivalry, occur during the perception of cognitive (or illusory) triangles, while earlier research had failed to find such rivalry (Bradley & Dumais, 1975). With line triangles, rivalry is experienced only under dichoptic conditions. A model is proposed to account for the observed phenomena, and the results of simulations are presented.

Purpose: To examine the relationship between visual acuity (VA) and radial deformation acuity (RDA) in children 6 to 12 years of age with amblyopia. Methods: RDA was measured in 35 participants with the Manchester RDA charts. VA was measured with the Early Treatment Diabetic Retinopathy Research Study (ETDRS) chart. Results: Median VA in non-amblyopic and amblyopic eyes was 0.04 logMAR (IQF -0.06 – 0.12) and 0.24 (IQF 0.12 – 0.04), respectively (Wilcoxon Signed Ranks test, z = -5.07, p < 0.001). Median RDA in non-amblyopic and amblyopic eyes was 2.73 log (IQF 2.53 – 2.87) and 2.63 log (IQF 2.53 – 2.77), respectively (Wilcoxon, z = -2.56, p < 0.05). Spearman correlation suggested that the amblyopic deficits in VA and RDA were related, r = -0.42, p < 0.05. Conclusion: A deficit in RDA was present in most children with amblyopia. A moderate relationship was noted between the amblyopic deficits found in VA and RDA.

No description available.

No description available.

Inspirés par les propriétés optiques des yeux composés de la mouche et par l’observation de micromouvements périodiques de sa rétine, différents capteurs visuels ont montré qu’il était possible de localiser un contraste très précisément. Ce fut les premières démonstrations d’hyperacuité visuelle de l’œil de la mouche.Dans cette thèse, un œil composé artificiel, doté d'un large champ de vision, est utilisé. Grâce à un nouvel algorithme de fusion des données visuelles, cet œil peut mesurer le déplacement d’un robot et lui permettre de se stabiliser au-dessus d’un environnement texturé. Localiser un contraste de manière linéaire sur l'ensemble du champ visuel demeure toutefois difficile. Un deuxième algorithme permet d’améliorer la localisation d’une barre grâce à un étalonnage, mais montre une certaine dépendance aux variations de contraste et de luminosité ambiante.Afin d'éviter ce processus d’étalonnage, un troisième algorithme qui s’appuie sur les travaux d'Heiligenberg et Baldi, a été proposé pour localiser deux contrastes. Ces auteurs ont montré que la somme pondérée de plusieurs capteurs ayant un champ récepteur gaussien pouvait fournir une estimation linéaire de la position d'un stimulus. Nous avons, pour la première fois, appliqué une variante de ce principe à un œil composé artificiel. Cet œil, une fois monté sur un robot permet de suivre une cible à une distance constante.Finalement, un œil composé artificiel dont la résolution intrinsèque est faible, peut être doté d’une hyperacuité visuelle et permettre de suivre une cible avec précision. Ces travaux ont ainsi conduit à proposer des stratégies bio-inspirées pour la localisation et la poursuite de cible. / Inspired by the optical properties of the fly compound eyes and the observation of its retinal periodic micro-movements, several visual sensors established that the localization of a contrast can be made very precisely. It was the first demonstration of the visual hyperacuity of the fly compound eye.In this thesis, an artificial compound eye with a wide field of view was used. Thanks to a novel algorithm fusing the visual signals, the sensor embedded onboard an aerial robot measures its displacement and enables the robot to hover above a textured environment.The localization of a contrast precisely over the whole field of view is still difficult. A second algorithm improved the localization of a bar thanks to a calibration. But it has a dependency to the contrast and the illuminance variations.In order to avoid a calibration process, a third algorithm was proposed to localize two contrasts. It is based on the work of Heiligenberg and Baldi, which showed that an array of Gaussian receptive field can provide a linear estimation of a stimulus position. For the first time, we applied a modified version of their estimation to an artificial compound eye. This sensor mounted on a rover allows following a target precisely at a constant distance.Finally, an artificial compound eye with a coarse spatial resolution can be endowed with hyperacuity and enables a robot to follow a target with precision. A step forward has been made toward bio-inspired target localization and pursuit.

Processing of visual stimuli in the vertebrate retina is complex and diverse. The retinal output to the higher centres of the nervous system, mediated by ganglion cells, consists of several different channels. Neurons in these channels can have very distinct response properties, which originate in different retinal pathways. In this work, the retinal origins and possible functional implications of the segregation of visual pathways will be investigated with a detailed, biologically realistic computational model of the retina. This investigation will focus on the two main retino-cortical pathways in the mammalian retina, the parvocellular and magnocellular systems, which are crucial for conscious visual perception. These pathways differ in two important aspects. The parvocellular system has a high spatial, but low temporal resolution. Conversely, the magnocellular system has a high temporal fidelity, spatial sampling however is less dense than for parvocellular cells. Additionally, the responses of magnocellular ganglion cells can show pronounced nonlinearities, while the parvocellular system is essentially linear. The origin of magnocellular nonlinearities is unknown and will be investigated in the first part of this work. As their main source, the results suggest specific properties of the photoreceptor response and a specialised amacrine cell circuit in the inner retina. The results further show that their effect combines in a multiplicative way. The model is then used to examine the influence of nonlinearities on the responses of ganglion cells in the presence of involuntary fixational eye movements. Two different stimulus conditions will be considered: visual hyperacuity and motion induced illusions. In both cases, it is possible to directly compare properties of the ganglion cell population response with psychophysical data, which allows for an analysis of the influence of different components of the retinal circuitry. The simulation results suggest an important role for nonlinearities in the magnocellular stream for visual perception in both cases. First, it will be shown how nonlinearities, triggered by fixational eye movements, can strongly enhance the spatial precision of magnocellular ganglion cells. As a result, their performance in a hyperacuity task can be equal to or even surpass that of the parvocellular system. Second, the simulations imply that the origin of some of the illusory percepts elicited by fixational eye movements could be traced back to the nonlinear properties of magnocellular ganglion cells. As these activity patterns strongly differ from those in the parvocellular system, it appears that the magnocellular system can strongly dominate visual perception in certain conditions. Taken together, the results of this theoretical study suggest that retinal nonlinearities may be important for and strongly influence visual perception. The model makes several experimentally verifiable predictions to further test and quantify these findings. Furthermore, models investigating higher visual processing stages may benefit from this work, which could provide the basis to produce realistic afferent input.

Les insectes sont capables de prouesses remarquables lorsqu’il s’agit d’éviter des obstacles,voler en environnement perturbé ou poursuivre une cible. Cela laisse penser que leurs capacités de traitement, aussi minimalistes soient-elles, sont parfaitement optimisées pour le vol. A cela s’ajoute des mécanismes raffinés, comme la stabilisation de la vision par rapport au corps, permettant d’améliorer encore plus leurs capacités de vol.Ces travaux de thèse présentent l’élaboration d’un micro drone de type quadrirotor, qui ressemble fortement à un insecte sur le plan perceptif (vibration rétinienne) et reprend des points structurels clés, tels que le découplage mécanique entre le corps et le système visuel. La conception du quadrirotor (de type open-source), son pilotage automatique et son système occulo-moteur sont minutieusement détaillés.Des traitements adaptés permettent, malgré un très faible nombre de pixels (24 pixels seulement), de poursuivre finement du regard une cible en mouvement. A partir de là, nous avons élaboré des stratégies basées sur le pilotage par le regard, pour stabiliser le robot en vol stationnaire, à l’aplomb d’une cible et asservir sa position ; et ce, en se passant d’une partie des capteurs habituellement utilisés en aéronautique tels que les magnétomètres et les accéléromètres. Le quadrirotor décolle, se déplace et atterrit de façon autonome en utilisant seulement ses gyromètres, son système visuel original mimant l’oeil d’un insecte et une mesure de son altitude. Toutes les expérimentations ont été validées dans une arène de vol, équipée de caméras VICON.Enfin, nous décrivons une nouvelle toolbox qui permet d’exécuter en temps réel des modèles Matlab/Simulink sur des calculateurs Linux embarqués de façon complètement automatisée (http://www.gipsalab.fr/projet/RT-MaG/). Cette solution permet d’écrire les modèles, de les simuler, d’élaborer des lois de contrôle pour enfin, piloter en temps réel, le robot sous l’environnement Simulink. Cela réduit considérablement le "time-to-flight" et offre une grande flexibilité (possibilité de superviser l’ensemble des données de vol, de modifier en temps réel les paramètres des contrôleurs, etc.). / Insects, like hoverflies are able of outstanding performances to avoid obstacles, reject disturbances and hover or track a target with great accuracy. These means that fast sensory motor reflexes are at work, even if they are minimalist, they are perfectly optimized for the flapping flight at insect scale. Additional refined mechanisms, like gaze stabilization relative to the body, allow to increase their flight capacity.In this PhD thesis, we present the design of a quadrotor, which is highly similar to an insect in terms of perception (visual system) and implements a bio-inspired gaze control system through the mechanical decoupling between the body and the visual system. The design of the quadrotor (open-source), itspilot and its decoupled eye are thoroughly detailed. New visual processing algorithms make it possible to faithfully track a moving target, in spite of a very limited number of pixels (only 24 pixels). Using this efficient gaze stabilization, we developed new strategies to stabilize the robot above a target and finely control its position relative to the target. These new strategies do not need classical aeronautic sensors like accelerometers and magnetometers. As a result, the quadrotor is able to take off, move and land automatically using only its embedded rate-gyros, its insect-like eye, and an altitude measurement. All these experiments were validated in a flying arena equipped with a VICON system. Finally, we describe a new toolbox, called RT-MaG toolbox, which generate automatically a real-time standalone application for Linux systems from a Matlab/Simulink model (http://www.gipsalab.fr/projet/RT-MaG/). These make it possible to simulate, design control laws and monitor the robot’s flight in real-time using only Matlab/Simulink. As a result, the "time-to-flight" is considerably reduced and the final application is highly reconfigurable (real-time monitoring, parameter tuning, etc.).