Showing papers in "Computer Vision and Image Understanding in 2002"

TL;DR: In this paper, the authors present a systematic approach to the problem of photorealistic 3D model acquisition from the combination of range and image sensing, where the output is a true texture-mapped geometric model of the scene.

TL;DR: The theoretical framework of FDT in continuous space is extended to digital cubic spaces and it is shown that for any fuzzy digital object, fuzzy distance is a metric for the support of the object.

TL;DR: The proposed system enhances the adaptability to cope with the random and dynamic environment of a road scene and leads to a reliable lane-departure warning system.

TL;DR: Four different methods for selecting thresholds that work on very different principles for image differencing are described.

TL;DR: This work proposes a method by which these tasks can be solved simultaneously: correction of measurement errors, registration of data shapes, and integrating them as a unified shape representation.

TL;DR: Through the development of a prototype gesture-controlled panoramic map browser, it is demonstrated that a vocabulary of predefined hand gestures can be used to interact successfully with applications running on an off-the-shelf personal computer equipped with a home video camera.

TL;DR: This paper reports on an evolutionary registration algorithm which does not require initial prealignment and has a very broad basin of convergence and has available to it a selection of evolutionary techniques to avoid local minima which plague both ICP and its variants.

TL;DR: The development of this automatic crude registration has allowed us to create a system that can generate complex 3D models from a set of partial reconstructions without any user intervention or prior knowledge of relative positions.

TL;DR: A novel version of the basic motion constraint equation that can be evaluated directly on the sensor grid is given and a regularization scheme to compute dense full flow fields from the sparse TLS solutions is proposed.

TL;DR: This special issue contains nine high quality papers representative of the state-of-the art technologies used to acquire and process range image data.

TL;DR: A fuzzy Markovian method for brain tissue segmentation from magnetic resonance images that aims at calculating a fuzzy membership in each voxel to indicate the partial volume degree, which is statistically modeled.

TL;DR: This paper qualitatively evaluates the output of several iterative closest point (ICP) variants on real-world data and expands to include additional ICP variants and an implementation of Chen and Medioni's point-to-plane minimizing algorithm.

TL;DR: This paper provides new results on the semi-continuity behavior of connectivity openings, studies the reconstruction operator in a complete lattice framework, and extends some known binary results regarding reconstruction to this framework.

TL;DR: This paper presents a relevance feedback technique that uses decision trees to learn a common thread among instances marked relevant in a preexisting content-based image retrieval (CBIR) system that is used to access high resolution computed tomographic images of the human lung.

TL;DR: Nine 3D differential operators based on only first order partial derivatives of an image yield a larger number of corresponding points than the other operators, and their performance on the basis of the statistical measures is better, and the separability of the detections is better so that a suitably chosen threshold can significantly decrease the number of false detections.

TL;DR: A new global registration technique to distribute registration errors evenly across all views, and a statistically sound thresholding to improve ICP robustness against noise and nonoverlapping data is introduced.

TL;DR: The empirical evaluation shows that feature extraction based on physicians' perceptual categories achieves significantly higher retrieval precision than the traditional scattershot approach, and the use of perceptually based features gives the system the ability to provide an explanation for its retrieval decisions, thereby instilling more confidence in its users.

TL;DR: An image thresholding approach based on the index of nonfuzziness maximization of the 2-D grayscale histogram is proposed, which performs well and effectively but also is more robust when applied to noisy images.

TL;DR: A new framework to perform motion compression for time-dependent 3D geometric data using the iterative closest point (ICP) algorithm to achieve significant compression, thereby leading to efficient storage and transmission of large volumes of 3D data.

TL;DR: This work presents results from optimization for 3D images when using from one to six local distances, all in the 5 ×5 × 5 neighborhood of a voxel.

TL;DR: A hybrid registration algorithm is proposed that combines advantages of working with small amounts of interest points, estimates the Euclidean transform matching both surfaces, and uses all available points and the iterative closest reciprocal point algorithm to refine the estimate and finally match surfaces.

TL;DR: An algorithm is shown that prealigns the front- and the backviews of rotationally symmetric objects for the registration of the two 3D-surfaces without using corresponding points.

TL;DR: This paper finds the best transformation of two range images to align each measured point and reconstruct 3D total object shape by taking such properties of the measurement error into account.

TL;DR: This study develops models for illumination and surface reflectance for use in outdoor color vision, and in particular for predicting the color of surfaces under outdoor conditions, by predicting the colors of surfaces in real images.

TL;DR: In experimental evaluations, the proposed method retrieved at least one similar shape in the top three retrieved items 99-100% of the time, and is suitable for fast, approximate shape retrieval in comparison with more accurate but more costly structural matching.

TL;DR: An efficient, robust means of obtaining reliable surface descriptions, suitable for object recognition, at multiple scales from range data, and an effective recognition system for free form objects based on attributed graphs constructed from the surface segmentation is presented.

TL;DR: A new regularization formulation for inverse problems in computer vision and image processing is introduced, which allows one to reconstruct second order piece-wise smooth images, that is, images consisting of an assembly of regions with almost constant value, almost constant slope, or almost constant curvature.

TL;DR: This work presents a simple but powerful computational model and associated algorithms based on the use of perceptual organizational principles, such as temporal coherence and spatial proximity, for motion segmentation, which can easily handle drastic illumination changes, occlusion events, and multiple moving objects, without theUse of training and specific object or illumination models.

TL;DR: Although spatial information was not used and its default internal parameters were used, MNS outperformed most compared methods, and showed good and fast performance under severe scale, viewpoint, occlusion, and background change using a single image for object modeling.

TL;DR: Experimental results showed that the proposed approach is very promising for automatic generation of3-D models which can be used for rendering a 3-D scene in a virtual reality system.