Showing papers in "Computer Vision and Image Understanding in 1998"

TL;DR: A grouping process is used to generate 2-D roof hypotheses from fragmented linear features extracted from the input image to detect buildings and construct 3-D shape descriptions of buildings from a monocular aerial image with a general viewpoint.

TL;DR: A technique called distance-ordered homotopic thinning (DOHT) for skeletonizing 3D binary images is presented and two variations of DOHT are presented that arise from using different rules for preserving points.

TL;DR: X Vision as discussed by the authors is a programming environment for real-time vision which provides high performance on standard workstations outfitted with a simple digitizer and consists of a small set of image-level tracking primitives, and a framework for combining them to form complex tracking systems.

TL;DR: This work presents a paradigm based on experimental psychology and statistics, in which humans rate the output of low level vision algorithms, and investigates whether there is a statistically significant difference in edge detector outputs as perceived by humans when considering an object recognition task.

TL;DR: A survey of methods developed by researchers to access and manipulate document images without the need for complete and accurate conversion is provided.

TL;DR: It is confirmed that the proposed method of page segmentation based on the approximated area Voronoi diagram is effective for extraction of body text regions, and it is as efficient as other methods based on connected component analysis.

TL;DR: This paper proposes techniques and formulations to match and cluster video shots of similar visual contents, taking into account the visual characteristics and temporal dynamics of video, and extends the Scene Transition Graphrepresentation for the analysis of temporal structures extracted from video.

TL;DR: The proposed rate pruning paradigm is a framework in which it is easy to analyze, compare, and tailor new pruning methods, and is supported by simulation results of the various pruned methods.

TL;DR: This paper addresses many implementation issues that have often been neglected in previous research, including temporal filtering of the output stream, algorithms' flexibility, and robustness to noise, subsampling, etc.

TL;DR: Core-related geometric properties and image object representations are laid out which, together with the aforementioned insensitivities, allow the core to be used effectively for a variety of image analysis objectives.

TL;DR: A model-based approach to automated 3D extraction of buildings from aerial images using a tight coupling of 2D image and 3D object modeling and processing, ending up in complex 3D building estimations of shape and location.

TL;DR: Experiments on both 2-D and 3-D data sets show that convergence is possible even for very rough initial positionings, and that the final registration accuracy typically approaches less than one quarter of the interpoint sampling resolution of the images.

TL;DR: The thesis of this paper is that the change induced by human activity can be inferred from changes in the organization among the visual features, and four measures are proposed to quantify the global statistical properties of the individual features and the relationships among them.

TL;DR: A deformable super quadric model based on a superquadric fit followed by a free-form deformation (FFD) to fit unstructured 3D points to help solve computer vision and image understanding problems.

TL;DR: It is demonstrated that proper modeling of degeneracy in the presence of outliers enables the detection of mismatches which would otherwise be missed and is a generalization of the robust estimator RANSAC.

TL;DR: This paper presents an overview of existing work on articulated and elastic motion, motivated by problems relating to the motion of the human body and of an animal heart, respectively.

TL;DR: The problems of corner detection and blob detection are treated in detail, and a combined framework for feature tracking is presented, which overcomes some of the inherent limitations of exposing fixed-scale tracking methods to image sequences in which the size variations are large.

TL;DR: A novel concept of instantour is proposed which enables us to instantly look around a visualized space of a dynamic real world without significant time delay and is also useful for another type of telepresence which uses recorded omnidirectional video streams.

TL;DR: This paper reviews the two classes of models of retino?cortical data reduction used in hardware implementations and reproduces the retina to cortex mapping based on conformal mapping functions and the overlap feature of receptive fields of retinal ganglion cells.

TL;DR: This paper addresses the problem of the recovery of a realistic textured model of a scene from a sequence of images, without any prior knowledge either about the parameters of the cameras or about their motion.

TL;DR: A novel method is derived to classify the dominant camera motions in video shots by analyzing the optical flow in a decomposed manner and is efficient and effective because only some mean values and standard deviations are used.

TL;DR: Both the original version of David Lowe's influential and classic algorithm for tracking known objects and a reformulation of it implemented by Ishiiet al. rely on (different) approximated imaging models, but removing their simplifying assumptions yields a fully projective solution.

TL;DR: A new method for the segmentation of images of document pages having both traditional and complex layouts is introduced, to efficiently produce a flexible description of the background space which surrounds the printed regions in the page image under all the above conditions.

TL;DR: The Ascender system acquires, extends, and refines 3D geometric site models from calibrated aerial imagery to add previously unseen buildings and to improve the geometric accuracy of the existing 3D building models.

TL;DR: Methods for building detection and reconstruction from aerial imagery intended for the analysis of urban and suburban areas and applied to images of different resolutions are discussed, illustrating that mid-resolution methods cannot be directly applied to high-resolution images.

TL;DR: ARUBA, a general framework for automated 3-D building reconstruction from multiple color aerial images, is presented and it is demonstrated that color is a very important cue in reconstructing a general class of objects, it is crucial to retain all information during the entire processing chain and a mutual interaction between 2-D and3-D processing is important.

TL;DR: This paper introduces a general purpose algorithm for reliable integration of sets of surface measurements into a single 3D model which constructs a single continuous implicit surface representation which is the zero-set of a scalar field function.

TL;DR: The segmentation of planar surfaces from the DSM is supported by existing ground plans and this 2D building information is used to derive hypotheses on the possible roof shapes in order to obtain a 3D boundary representation based on the segmented planes.

TL;DR: It is found that this CNN chip with a simple 3 × 3 CNN kernel can reliably classify four textures and it is believed that more textures can be separated and adequate texture segmentation (< 1% error) can be achieved.

TL;DR: It is shown that a normalization procedure based on affine transformations whose anchor points are the locations of the eyes and mouth substantially increases the effectiveness of general purpose classification techniques in face recognition.