Les récents développements des technologies de l’imagerie 3D et en particulier la stéréoscopie ont ouvert de nouveaux horizons dans de nombreux domaines d’application tels que la TV 3D, le cinéma 3D, les jeux vidéo et la vidéoconférence. Ces avancées technologiques ont soulevé plusieurs défis aussi bien sur le plan théorique que pratique et en particulier dans le domaine du codage des données 3D. En effet, l’énorme quantité d’information issue des systèmes d’acquisition requiert des solutions efficaces pour la coder et la transmettre. L’objectif de cette thèse est le développement de méthodes pour optimiser les principales étapes de la chaine de traitement et transmission d’images stéréoscopiques. Nous nous limitons dans ce travail au rehaussement de contraste, le codage et l’évaluation de la qualité d’images stéréoscopiques. La première partie de ce travail traite les problèmes d’évaluation et d’amélioration de la qualité d’images stéréoscopiques. Nous nous intéressons d’abord au rehaussement de contraste en s’inspirant des méthodes 2D et en intégrant quelques éléments liés à la perception visuelle. Nous proposons ainsi une première méthode de rehaussement de contraste local basée sur la carte de saillance visuelle. L’aspect qualité est aussi traité selon une approche basée sur les protocoles et méthodes conues pour le cas des images 2D et 3D. Cette méthode exploite les caractéristiques et propriétés connues du système visuel humain (SVH) telles que la non-linéarité, la sensibilité au contraste, la sélectivité directionnelle et fréquentielle ainsi que le seuil de discrimination binoculaire. Nous avons aussi d´eveloppé une méthode de prédiction de la qualité d’images stéréoscopiques sans référence. Cette dernière est basée sur des descripteurs 3D statistiques issus de la scène naturelle afin identifier les distorsions. Ces descripteurs 3D statistiques correspondent aux attributs extraits à partir de la paire stéréo naturelle et de la carte de disparité. L’extraction de ces descripteurs se fait au moyen de l’analyse en ondelettes des images stéréoscopiques. La deuxième partie de cette thèse traite les problèmes de compression d’images stéréoscopiques. Nous avons commencé par l’exploitation de la transformée en cosinus discret unidirectionnel et unidimensionnel pour encoder l’image résiduelle issue de la compensation de disparité. Ensuite, en se basant sur la transformée en ondelettes, nous avons étudié deux techniques pour optimiser le calcul de l’image résiduelle. Enfin, nous avons proposé des méthodes d’allocation de débit pour la compression des images stéréoscopiques. En général, le problème d’allocation de bits est résolu d’une manière empirique en cherchant le débit optimale qui minimise une certaine distorsion. Cependant cette stratégie est complexe. Pour cela, nous avons proposé des méthodes d’allocation de débits, rapides et efficaces appropriées pour le codage en boucle ouverte et en boucle fermée. Cette thèse ouvre des perspectives dans les trois thématiques abordées, à savoir le rehaussement de contraste, le codage et l’évaluation de la qualité d’images stéréoscopiques. / Recent developments in 3D stereoscopic technology have opened new horizons in many application fields such as 3DTV, 3D cinema, video games and videoconferencing and at the same time raised a number of challenges related to the processing and coding of 3D data. Today, stereoscopic imaging technology is becoming widely used in many fields. There are still some problems related to the physical limitations of image acquisition systems, e.g. transmission and storage requirements. The objective of this thesis is the development of methods for improving the main steps of stereoscopic imaging pipeline such as enhancement, coding and quality assessment. The first part of this work addresses quality issues including contrast enhancement and quality assessment of stereoscopic images. Three algorithms have been proposed. The first algorithm deals with the contrast enhancement aiming at promoting the local contrast guided by calculated/estimated object importance map in the visual scene. The second and the third algorithms aim at predicting the distortion severity of stereo images. In the second one, we have proposed a fullreference metric that requires the reference image and is based on some 2D and 3D findings such as amplitude non-linearity, contrast sensitivity, frequency and directional selectivity, and binocular just noticeable difference model. While in the third algorithm, we have proposed a no-reference metric which needs only the stereo pair to predict its quality. The latter is based on Natural Scene statistics to identify the distortion affecting the stereo image. The statistic 3D features consist in combining features extracted from the natural stereo pair and those from the estimate disparity map. To this end, a joint wavelet transform, inspired from the vector lifting concept is first employed. Then, the features are extracted from the obtained subbands. The second part of this dissertation addresses stereoscopic image compression issues. We started by investigating a one-dimensional directional discrete cosine transform to encode the disparity compensated residual image. Afterwards, and based on the wavelet transform, we investigated two techniques for optimizing the computation of the residual image. Finally, we present efficient bit allocation methods for stereo image coding purpose. Generally, the bit allocation problem is solved in an empirical manner by looking for the optimal rates leading to the minimum distortion value. Thanks to recently published work on approximations of the entropy and distortion functions, we proposed accurate and fast bit allocation schemes appropriate for the open-loop and closed-loop based stereo coding structures.

ITC/USA 2013 Conference Proceedings / The Forty-Ninth Annual International Telemetering Conference and Technical Exhibition / October 21-24, 2013 / Bally's Hotel & Convention Center, Las Vegas, NV / This paper described the details of subjective validation for visually lossless compression of stereoscopic 3 dimensional (3D) images. The subjective testing method employed in this work is adapted from methods used previously for visually lossless compression of 2 dimensional (2D) images. Confidence intervals on the correct response rate obtained from the subjective validation of compressed stereo pairs provide reliable evidence to indicate that the compressed stereo pairs are visually lossless.

ITC/USA 2012 Conference Proceedings / The Forty-Eighth Annual International Telemetering Conference and Technical Exhibition / October 22-25, 2012 / Town and Country Resort & Convention Center, San Diego, California / This paper proposes a method of measuring visibility thresholds for quantization distortion in JPEG2000 for compression of stereoscopic 3D images. The crosstalk effect is carefully considered to ensure that quantization errors in each channel of stereoscopic images are imperceptible to both eyes. A model for visibility thresholds is developed to reduce the daunting number of measurements required for subjective experiments.

This thesis describes an automatic evaluation approach for estimating the quality of stereo displays and vision systems using image features. The method is inspired by the human visual system. Display of stereo images is widely used to enhance the viewing experience of three-dimensional (3D) visual displays and communication systems. Applications are numerous and range from entertainment to more specialized applications such as: 3D visualization and broadcasting, robot tele-operation, object recognition, body exploration, 3D teleconferencing, and therapeutic purposes. Consequently, perceived image quality is important for assessing the performance of 3D imaging applications. There is no doubt that subjective testing (i.e., asking human viewers to rank the quality of stereo images) is the most accurate method for quality evaluation. It reflects true human perception. However, these assessments are time consuming and expensive. Furthermore, they cannot be done in real time. Therefore, the goal of this research is to develop an objective quality evaluation methods computational models that can automatically predict perceived image quality) correlating well with subjective predictions that are required in the field of quality assessment. I believe that the perceived distortion and disparity of any stereoscopic display are strongly dependent on local features, such as edge (non-uniform) and non-edge (uniform) areas. Therefore, in this research, I propose a No-Reference (NR) objective quality assessment for coded stereoscopic images based on segmented local features of artifacts and disparity. Local feature information such as edge and non-edge area based relative disparity estimation, as well as the blockiness, blur, and the zero-crossing within the block of images, are evaluated in this method. A block-based edge dissimilarity approach is used for disparity estimation. I use the Toyama stereo images database to evaluate the performance and to compare it with other approaches both qualitatively and quantitatively.

This thesis describes an automatic evaluation approach for estimating the quality of stereo displays and vision systems using image features. The method is inspired by the human visual system. Display of stereo images is widely used to enhance the viewing experience of three-dimensional (3D) visual displays and communication systems. Applications are numerous and range from entertainment to more specialized applications such as: 3D visualization and broadcasting, robot tele-operation, object recognition, body exploration, 3D teleconferencing, and therapeutic purposes. Consequently, perceived image quality is important for assessing the performance of 3D imaging applications. There is no doubt that subjective testing (i.e., asking human viewers to rank the quality of stereo images) is the most accurate method for quality evaluation. It reflects true human perception. However, these assessments are time consuming and expensive. Furthermore, they cannot be done in real time. Therefore, the goal of this research is to develop an objective quality evaluation methods computational models that can automatically predict perceived image quality) correlating well with subjective predictions that are required in the field of quality assessment. I believe that the perceived distortion and disparity of any stereoscopic display are strongly dependent on local features, such as edge (non-uniform) and non-edge (uniform) areas. Therefore, in this research, I propose a No-Reference (NR) objective quality assessment for coded stereoscopic images based on segmented local features of artifacts and disparity. Local feature information such as edge and non-edge area based relative disparity estimation, as well as the blockiness, blur, and the zero-crossing within the block of images, are evaluated in this method. A block-based edge dissimilarity approach is used for disparity estimation. I use the Toyama stereo images database to evaluate the performance and to compare it with other approaches both qualitatively and quantitatively.

This thesis analyzes the problem of acquiring stereoscopic images in all gazing directions around a reference viewpoint in space with the purpose of creating stereoscopic panoramas of non-static scenes. The generation of immersive stereoscopic imagery suitable to stimulate human stereopsis requires images from two distinct viewpoints with horizontal parallax in all gazing directions, or to be able to simulate this situation in the generated imagery. The available techniques to produce omnistereoscopic imagery for human viewing are not suitable to capture dynamic scenes stereoscopically. This is a not trivial problem when considering acquiring the entire scene at once while avoiding self-occlusion between multiple cameras. In this thesis, the term omnidirectional refers to all possible gazing directions in azimuth and a limited set of directions in elevation. The acquisition of dynamic scenes restricts the problem to those techniques suitable for collecting in one simultaneous exposure all the necessary visual information to recreate stereoscopic imagery in arbitrary gazing directions. The analysis of the problem starts by defining an omnistereoscopic viewing model for the physical magnitude to be measured by a panoramic image sensor intended to produce stereoscopic imagery for human viewing. Based on this model, a novel acquisition model is proposed, which is suitable to describe the omnistereoscopic techniques based on horizontal stereo. From this acquisition model, an acquisition method based on multiple cameras combined with the rendering by mosaicking of partially overlapped stereoscopic images is identified as a good candidate to produce omnistereoscopic imagery of dynamic scenes. Experimental acquisition and rendering tests were performed for different multiple-camera configurations. Furthermore, a mosaicking criterion between partially overlapped stereoscopic images based on the continuity of the perceived depth and the prediction of the location and magnitude of unwanted vertical disparities in the final stereoscopic panorama are two main contributions of this thesis. In addition, two novel omnistereoscopic acquisition and rendering techniques were introduced. The main contributions to this field are to propose a general model for the acquisition of omnistereoscopic imagery, to devise novel methods to produce omnistereoscopic imagery, and more importantly, to contribute to the awareness of the problem of acquiring dynamic scenes within the scope of omnistereoscopic research.

This thesis analyzes the problem of acquiring stereoscopic images in all gazing directions around a reference viewpoint in space with the purpose of creating stereoscopic panoramas of non-static scenes. The generation of immersive stereoscopic imagery suitable to stimulate human stereopsis requires images from two distinct viewpoints with horizontal parallax in all gazing directions, or to be able to simulate this situation in the generated imagery. The available techniques to produce omnistereoscopic imagery for human viewing are not suitable to capture dynamic scenes stereoscopically. This is a not trivial problem when considering acquiring the entire scene at once while avoiding self-occlusion between multiple cameras. In this thesis, the term omnidirectional refers to all possible gazing directions in azimuth and a limited set of directions in elevation. The acquisition of dynamic scenes restricts the problem to those techniques suitable for collecting in one simultaneous exposure all the necessary visual information to recreate stereoscopic imagery in arbitrary gazing directions. The analysis of the problem starts by defining an omnistereoscopic viewing model for the physical magnitude to be measured by a panoramic image sensor intended to produce stereoscopic imagery for human viewing. Based on this model, a novel acquisition model is proposed, which is suitable to describe the omnistereoscopic techniques based on horizontal stereo. From this acquisition model, an acquisition method based on multiple cameras combined with the rendering by mosaicking of partially overlapped stereoscopic images is identified as a good candidate to produce omnistereoscopic imagery of dynamic scenes. Experimental acquisition and rendering tests were performed for different multiple-camera configurations. Furthermore, a mosaicking criterion between partially overlapped stereoscopic images based on the continuity of the perceived depth and the prediction of the location and magnitude of unwanted vertical disparities in the final stereoscopic panorama are two main contributions of this thesis. In addition, two novel omnistereoscopic acquisition and rendering techniques were introduced. The main contributions to this field are to propose a general model for the acquisition of omnistereoscopic imagery, to devise novel methods to produce omnistereoscopic imagery, and more importantly, to contribute to the awareness of the problem of acquiring dynamic scenes within the scope of omnistereoscopic research.