Showing papers by "Luc Van Gool published in 1998"

Multi Viewpoint Stereo from Uncalibrated Video Sequences

Abstract: This contribution describes an automatic 3D surface modeling system that extracts dense metric 3D surfaces from an uncalibrated video sequence. A static 3D scene is observed from multiple viewpoints by freely moving a video camera around the object. No restrictions on camera movement and internal camera parameters like zoom are imposed, as the camera pose and intrinsic parameters are calibrated from the sequence.

Metric 3D surface reconstruction from uncalibrated image sequences

Abstract: Modeling of 3D objects from image sequences is one of the challenging problems in computer vision and has been a research topic for many years. Important theoretical and algorithmic results were achieved that allow to extract even complex 3D scene models from images. One recent effort has been to reduce the amount of calibration and to avoid restrictions on the camera motion. In this contribution an approach is described which achieves this goal by combining state-of-the-art algorithms for uncalibrated projective reconstruction, self-calibration and dense correspondence matching.

Automatic Modeling and 3D Reconstruction of Urban House Roofs from High Resolution Aerial Imagery

Abstract: Many tasks in modern urban planning require 3-dimensional (3D) spatial information, preferably in the form of 3D city models Constructing such models requires automatic methods for reliable 3D building reconstruction House roofs encountered in residential areas in European cities exhibit a wide variety in their shapes This limits the use of predefined roof models for their reconstruction The strategy put forward in this paper is, first, to construct a polyhedral model of the roof structure, which captures the topology of the roof, but which might not be very accurate in a metric sense; and then, in a second step, to improve the metric accuracy by fitting this model to the data This decoupling of topology extraction from metric reconstruction allows a more efficient roof modelling involving less criteria And, restricting the processing, at all stages, to one or just a few roof structures, by using a colour-based segmentation of the images, allows to use constraints that are not very tight The approach has been tested on a state-of-the-art dataset of aerial images of residential areas in Brussels

3D Structure from Multiple Images of Large-Scale Environments

Abstract: Invited Presentations.- Cumuli, Panorama, and Vanguard Project Overview.- Dualizing Scene Reconstruction Algorithms.- Multiview Relations and Correspondence Search.- Geometry of Multiple Affine Views.- Tensor Embedding of the Fundamental Matrix.- Optimal Estimation of Matching Constraints.- Matching and Reconstruction from Widely Separated Views.- Improving Block-Based Disparity Estimation by Considering the Non-uniform Distribution of the Estimation Error.- 3D Structure from Multiple Images.- Beyond the Epipolar Constraint: Integrating 3D Motion and Structure Estimation.- Multi-Camera Acquisitions for High-Accuracy 3D Reconstruction.- Metric 3D Surface Reconstruction from Uncalibrated Image Sequences.- Automatic 3D Model Construction for Turn-Table Sequences.- Calibration and Reconstruction Using Scene Constraints.- Geometrically Constrained Structure from Motion: Points on Planes.- Euclidean and Affine Structure/Motion for Uncalibrated Cameras from Affine Shape and Subsidiary Information.- From Ordinal to Euclidean Reconstruction with Partial Scene Calibration.- Imposing Euclidean Constraints During Self-Calibration Processes.- Interactive 3D Modeling from Multiple Images Using Scene Regularities.- Range Integration and Augmented Reality Applications.- Integration of Multiple Range Maps through Consistency Processing.- Fitting Geometrical Deformable Models to Registered Range Images.- The Use of Reality Models in Augmented Reality Applications.- Applying Augmented Reality Techniques in the Field of Interactive Collaborative Design.- A Guided Tour Through Multiview Relations.

Getting facial features and gestures in 3D

Abstract: In the study of face perception and face animation, there is a growing interest in using 3-D data. The advent of dedicated laser scanners has proved instrumental in this regard. The availability of 3D face models makes it easier to change viewpoints and illumination conditions in perception experiments or to build animated likenesses in graphics. Here we propose to go one step further and also capture face dynamics (visemes, epressions) in 3-D. To that end, an active 3-D acquisition system is proposed, that yields 3-D, textured snapshots from a single image. By the fact that data are captured from a single shot, the face may move during the operation. Alternatively, images can be taken at video rate and for each frame a 3-D reconstruction (still textured if required) can be made. Such 3-D movies can then be used to analyse facial expressions in 3-D, through the tracking of points or features on the face. Preliminary experiments in that direction are presented. The reported work is part of an effort to model facial expressions without taking recourse to the modeling of the underlying physiology.

Projectice subgroups for grouping

Abstract: The history of grouping in computer vision stretches a long way back. One strand could be called geometry–based and focuses on shapes with special regularities such as symmetry. From a mathematical point of view, the subparts of such shapes are related by special transformations. The presented work proposes to systematically classify such ‘special’ transformations by studying projective subgroups that come with fixed structures. These are geometric entities such as points or lines that remain fixed under the projectivities in the subgroups. As subgroups have their own invariants, these can then be used to guide the search for the corresponding groupings.