In general, stereovision can be defined as a two part problem. The first is the correspondence problem. This involves determining the image point in each image of a set of images that correspond to the same physical point P. We will call this set of image points, N. The second problem is the reconstruction problem. Once a set of image points, N, that correspond to point P has been determined, N is then used to extract three dimensional information about point P. This master's thesis presents three novel solutions to the reconstruction problem. Two of the techniques presented are for detecting the location of a 3-D point and one for detecting a line expressed in a three dimensional coordinate system. These techniques are tested and validated using a unique 3-D finger detection algorithm. The techniques presented are unique because of their simplicity and because they do not require the cameras to be placed in specific locations, orientations or have specific alignments. On the contrary, it will be shown that the techniques presented in this thesis allow the two cameras used to assume almost any relative pose provided that the object of interest is within their field of view. The relative pose of the cameras at a given instant in time, along with basic equations from the perspective image model are used to form a system of equations that when solved, reveal the 3-D coordinates of a particular fixed point of interest or the three dimensional equation of a fixed line of interest. Finally, it will be shown that a single moving camera can successfully perform the same line and point detection accomplished by two cameras by altering the pose of the camera. The results presented in this work are beneficial to any typical stereovision application because of the computational ease in comparison to other point and line reconstruction techniques. But more importantly, this work allows for a single moving camera to perceive three-dimensional position information, which effectively removes the two camera constraint for a stereo vision system. When used with other monocular cues such as texture or color, the work presented in this thesis could be as accurate as binocular stereo vision at interpreting three dimensional information. Thus, this work could potentially increase the three dimensional perception of a robot that normally uses one camera, such as an eye-in-hand robot or a snake like robot.

Thèse doctorat : Images et Systèmes : Villeurbanne, INSA : 2006. / Thèse rédigée en anglais. Titre provenant de l'écran-titre. Bibliogr. p. 161-172.

Miniature cameras play a key role in numerous imaging applications ranging from endoscopy and metrology inspection devices to smartphones and head-mount acquisition systems. However, due to the physical constraints, the imaging conditions, and the low quality of small optics, their imaging capabilities are limited in terms of the delivered resolution, the acquired depth of field, and the captured dynamic range. Computational imaging jointly addresses the imaging system and the reconstructing algorithms to bypass the traditional limits of optical systems and deliver better restorations for various applications. The scene is encoded into a set of efficient measurements which could then be computationally decoded to output a richer estimate of the scene as compared with the raw images captured by conventional imagers. In this dissertation, three task-based computational imaging techniques are developed to make low-quality miniature cameras capable of delivering realistic high-resolution reconstructions, providing full-focus imaging, and acquiring depth information for high dynamic range objects. For the superresolution task, a non-regularized direct superresolution algorithm is developed to achieve realistic restorations without being penalized by improper assumptions (e.g., optimizers, priors, and regularizers) made in the inverse problem. An adaptive frequency-based filtering scheme is introduced to upper bound the reconstruction errors while still producing more fine details as compared with previous methods under realistic imaging conditions. For the full-focus imaging task, a computational depth-based deconvolution technique is proposed to bring a scene captured by an ordinary fixed-focus camera to a full-focus based on a depth-variant point spread function prior. The ringing artifacts are suppressed on three levels: block tiling to eliminate boundary artifacts, adaptive reference maps to reduce ringing initiated by sharp edges, and block-wise deconvolution or depth-based masking to suppress artifacts initiated by neighboring depth-transition surfaces. Finally for the depth acquisition task, a multi-polarization fringe projection imaging technique is introduced to eliminate saturated points and enhance the fringe contrast by selecting the proper polarized channel measurements. The developed technique can be easily extended to include measurements captured under different exposure times to obtain more accurate shape rendering for very high dynamic range objects.

<p>Diffusion tensor imaging (DTI) has emerged as a unique method to characterize brain tissue microstructure non-invasively. DTI typically provides the ability to study white matter structure with a standard voxel resolution of 8&mu;L over imaging field-of-views of the extent of the human brain. As such, it has long been recognized as a promising tool not only in clinical research for the diagnostic and monitoring of white matter diseases, but also for investigating the fundamental biological principles underlying the organization of long and short-range cortical networks. However, the complexity of brain structure within an MRI voxel makes it difficult to dissociate the tissue origins of the measured anisotropy. The tensor characterization is a composite result of proton pools in different tissue and cell structures with diverse diffusion properties. As such, partial volume effects introduce a strong bias which can lead to spurious measurements, especially in regions with a complex tissue structure such as interdigitating crossing fibers or in convoluted cortical folds near the grey/white matter interface.</p><p>This dissertation focuses on the design and development of acquisition and image reconstruction strategies to improve the spatial resolution of diffusion imaging. After a brief review of the theory of diffusion MRI and of the basic principles of streamline tractography in the human brain, the main challenges to increasing the spatial resolution are discussed. A comprehensive characterization of artifacts due to motion and field inhomogeneities is provided and novel corrective methods are proposed to enable the acquisition of diffusion weighted data with 2D mulitslice imaging techniques with full brain coverage, increased SNR and high spatial resolutions of 1.25&times;1.25&times;1.25 mm<super>3</super> within an acceptable scan time. The method is extended to a multishot k_{_z}-encoded 3D multislab spiral DTI and evaluated in normal human volunteers.</p><p>To demonstrate the increased SNR and enhanced resolution capability of the proposed methods and more generally to assess the value of high-spatial resolution in diffusion imaging, a study of cortical depth-dependence of fractional anisotropy was performed at an unprecedented <italic>in-vivo</italic> inplane resolution of 0.390&times;0.390&mu;m<super>2</super> and an investigation of the trade-offs between spatial resolution and cortical specificity was conducted within the connectome framework.</p> / Dissertation

This thesis presents a new system that reconstructs the 3D representation of dental casts. To maintain the integrity of the 3D representation, a standard model is built to cover the blind spots that the camera cannot reach. The standard model is obtained by scanning a real human mouth model with a laser scanner. Then the model is simplified by an algorithm which is based on iterative contraction of vertex pairs. The simplified standard model uses a local parametrization method to obtain the curvature information. The system uses a digital camera and a square tube mirror in front of the camera to capture multi-view images. The mirror is made of stainless steel in order to avoid double reflections. The reflected areas of the image are considered as images taken by the virtual cameras. Only one camera calibration is needed since the virtual cameras have the same intrinsic parameters as the real camera. Depth is computed by a simple and accurate geometry based method once the corresponding points are identified. Correspondences are selected using a feature point based stereo matching process, including fast normalized cross-correlation and simulated annealing.

In this thesis, the applications of image stitching and object insertion are considered and two gradient based approaches offering solutions are proposed. An essential part of the proposed methods is obtaining an image from a given gradient data set. This is done using an existing Haar wavelet based reconstruction technique, which consists of two main steps. First, the Haar wavelet decomposition of the image to be reconstructed is obtained directly from a given gradient. Second, the image is obtained using Haar wavelet synthesis. In both stitching and object insertion applications considered, the gradient from which the image must be reconstructed is a non-conservative vector field and this requires adding an iterative Poisson solver at each resolution level, during the synthesis step of the reconstruction technique. The performance of the reconstruction algorithm is evaluated by comparing it with other existing techniques, in terms of solution accuracy and computation speed. The proposed image stitching technique consists of three main parts: registering the images to be combined, blending their gradients over a region of interest and obtaining a composite image from a gradient. The object insertion technique considers the images registered and has two main stages: gradient blending of images in a region of interest and recovering an image from the gradient. The performance of the stitching algorithm is evaluated visually, by presenting the results produced to combine images with varying orientation, scales, illumination, and color conditions. Experimental results illustrate both the stitching and the insertion techniques proposed, and indicate that they yield seamless composite images. / Graduate

Random heterogeneous, scale-dependent structures can be observed from many image sources, especially from remote sensing and scientific imaging. Examples include slices of porous media data showing pores of various sizes, and a remote sensing image including small and large sea-ice blocks. Meanwhile, rather than the images of phenomena themselves, there are many image processing and analysis problems requiring to deal with \emph{discrete-state} fields according to a labeled underlying property, such as mineral porosity extracted from microscope images, or an ice type map estimated from a sea-ice image. In many cases, if discrete-state problems are associated with heterogeneous, scale-dependent spatial structures, we will have to deal with complex discrete state fields. Although scale-dependent image modeling methods are common for continuous-state problems, models for discrete-state cases have not been well studied in the literature. Therefore, a fundamental difficulty will arise which is how to represent such complex discrete-state fields. Considering the success of hidden field methods in representing heterogenous behaviours and the capability of hierarchical field methods in modeling scale-dependent spatial features, we propose a Hidden Hierarchical Markov Field (HHMF) approach, which combines the idea of hierarchical fields with hidden fields, for dealing with the discrete field modeling challenge. However, to define a general HHMF modeling structure to cover all possible situations is difficult. In this research, we use two image application problems to describe the proposed modeling methods: one for scientific image (porous media image) reconstruction and the other for remote-sensing image synthesis. For modeling discrete-state fields with a spatially separable complex behaviour, such as porous media images with nonoverlapped heterogeneous pores, we propose a Parallel HHMF model, which can decomposes a complex behaviour into a set of separated, simple behaviours over scale, and then represents each of these with a hierarchical field. Alternatively, discrete fields with a highly heterogeneous behaviour, such as a sea-ice image with multiple types of ice at various scales, which are not spatially separable but arranged more as a partition tree, leads to the proposed Tree-Structured HHMF model. According to the proposed approach, a complex, multi-label field can be repeatedly partitioned into a set of binary/ternary fields, each of which can be further handled by a hierarchical field.

The resolution of the image is one of the main measures of image quality. Higher resolution is desired and often required in most of the applications, because higher resolution means more details in the image. The use of better image sensors and optics is an expensive and also limiting way of increasing pixel density within the image. The use of image processing methods, to obtain a high resolution image from low resolution images is a cheap and effective solution. This kind of image enhancement is called super resolution image reconstruction. This thesis focuses on the definition, implementation and analysis on well-known techniques of super resolution. The comparison and analysis are the main concerns to understand the improvements of the super resolution methods over single frame interpolation techniques. In addition, the comparison also gives us an insight to the practical uses of super resolution methods. As a result of the analysis, the critical examination of the techniques and their performance evaluation are achieved.

Subband/Wavelet filter analysis-synthesis filters are a major component in many compression algorithms. Such compression algorithms have been applied to images, voice, and video. These algorithms have achieved high performance. Typically, the configuration for such compression algorithms involves a bank of analysis filters whose coefficients have been designed in advance to enable high quality reconstruction. The analysis system is then followed by subband quantization and decoding on the synthesis side. Decoding is performed using a corresponding set of synthesis filters and the subbands are merged together. For many years, there has been interest in improving the analysis-synthesis filters in order to achieve better coding quality. Adaptive filter banks have been explored by a number of authors where by the analysis filters and synthesis filters coefficients are changed dynamically in response to the input. A degree of performance improvement has been reported but this approach does require that the analysis system dynamically maintain synchronization with the synthesis system in order to perform reconstruction. In this thesis, we explore a variant of the adaptive filter bank idea. We will refer to this approach as fixed-analysis adaptive-synthesis filter banks. Unlike the adaptive filter banks proposed previously, there is no analysis synthesis synchronization issue involved. This implies less coder complexity and more coder flexibility. Such an approach can be compatible with existing subband wavelet encoders. The design methodology and a performance analysis are presented.

Thesis (Ph. D.)--Ohio State University, 2007. / Title from first page of PDF file. Includes bibliographical references (p. 118-124).