Showing papers in "Computer Vision and Image Understanding in 2003"

TL;DR: An algorithm--the TPS-RPM algorithm--with the thin-plate spline (TPS) as the parameterization of the non-rigid spatial mapping and the softassign for the correspondence is developed.

TL;DR: This work introduces and test different Bayesian network classifiers for classifying expressions from video, focusing on changes in distribution assumptions, and feature dependency structures, and proposes a new architecture of hidden Markov models (HMMs) for automatically segmenting and recognizing human facial expression from video sequences.

TL;DR: It is able to show that the FastICA algorithm configured according to ICA architecture II yields the highest performance for identifying faces, while the InfoMax algorithm configurations is better for recognizing facial actions.

TL;DR: By comparative evaluations, it has been shown that the efficiency of the proposed algorithm is superior to those of the conventional algorithms.

TL;DR: A new model-based moving feature extraction analysis is presented that automatically extracts and describes human gait for recognition and is shown to be able to handle high levels of occlusion, which is of especial importance in gait as the human body is self-occluding.

TL;DR: A component-based method and two global methods for face recognition and evaluate them with respect to robustness against pose changes are presented and the component system clearly outperformed both global systems.

TL;DR: Recognition of human faces using a gallery of still or video images and a probe set of videos is systematically investigated using a probabilistic framework and a computationally efficient sequential importance sampling (SIS) algorithm is developed to estimate the posterior distribution.

TL;DR: The notion of iconic feature based (IFB) algorithms, which lie between geometrical and standard intensity based algorithms, are introduced, and a new registration energy for IFB registration that generalizes some of the existing techniques is presented.

TL;DR: This paper presents a system that performs automatic annotation of the principal highlights in soccer video, suited for both production and posterity logging.

TL;DR: Analysis reveals that under many circumstances, using thermal infrared imagery yields higher performance, while in other cases performance in both modalities is equivalent, and provides a partial explanation for the multiple contradictory claims in the literature regarding performance of various algorithms on visible data sets.

TL;DR: An original approach in order to improve the results of color image segmentation by pixel classification by classify pixels represented in the hybrid color space which is specifically designed to yield the best discrimination between the pixel classes.

TL;DR: A novel content-based 3D motion retrieval algorithm is presented that partition the motion library and construct a motion index tree based on a hierarchical motion description and serves as a classifier to determine the sub-library that contains the promising similar motions to the query sample.

TL;DR: A rigid-invariant variational framework that can deal as well with local non-rigid transformations and arbitrary topological changes is proposed, and a simple and robust shape representation is used for global-to-local alignment.

TL;DR: The findings show that the asymmetry measures of automatically selected facial regions capture individual differences that are relatively stable to facial expression variations, and a synergy is achieved by combining facial asymmetry information with conventional EigenFace and FisherFace methods.

TL;DR: An accurate and robust evaluator that dynamically removes contributions of backgrounds and analyzes voting patterns around peaks in the accumulator space is proposed that can be applied to any HT that keeps the basic characteristics of the voting process.

TL;DR: A method for globally optimal partial line segment matching under bounded or Gaussian error, point matching under aGaussian error model with subpixel accuracy and precise orientation models, and a simple and robust technique for finding multiple distinct objcct instances are presented.

TL;DR: In the proposed scheme, video objects are first segmented, followed by a video object abstraction algorithm for identifying key frames to reduce data redundancy and provide reliable feature data for next stage of the algorithm.

TL;DR: Progress in matching shots which are images of the same 3D location in a film, and analogues of local spatial consistency, cross-correlation, and epipolar geometry for these tracks are described.

TL;DR: This paper presents a set of multimodal point similarity measures based on joint intensity distribution and test them as an integral part of non-rigid multi-modal registration system, showing that segmentation-based measure, which models the joint distribution as a sum of intensity classes, performs best.

TL;DR: A method for characterizing brain shift that is based on non-rigid surface registration, and can be combined with a constitutively realistic finite element approach for volumetric displacement estimation is proposed.

TL;DR: A model-based approach to the automatic detection and reconstruction of buildings from aerial imagery that is capable of reconstructing a variety of building types including peaked, flat, multi-level flat, and curved surfaces is presented.

TL;DR: The main advantage of the proposed 3D reconstruction algorithm over others is that it is able to retain the specific features of the face in the video sequence even when these features are different from those of the generic model and it does so even as the quality of the input video varies.

TL;DR: It is shown that there are only two topologies in Z2 and five topologiesin Z3 whose connected sets are connected in the intuitive sense, and two of the fiveTopologies for Z3 are products of the topologies known from Z1 and Z2.

TL;DR: A novel and extensive evaluation is carried out to demonstrate and compare the usefulness of these algorithms for video retrieval tasks using a carefully created test collection of over 6000 still images, where performance is measured against relevance judgements based on human image annotations.

TL;DR: A region-based approach to nonrigid motion tracking that provides robustness to occlusions, wrinkles, shadows, and specular highlights and is tailored to take advantage of texture mapping hard-ware available in many workstations, PCs, and game consoles.

TL;DR: A spline-based elastic registration scheme which is based on a well-defined minimizing functional for which the solution can be stated analytically and which can preserve the shape of rigid structures embedded in an otherwise elastic material is introduced.

TL;DR: A new image-matching mathematical model that combines region matching and segmentation by formulation of the energy minimization problem with free boundary conditions could enable some improvements of mammogram comparisons and tumor detection system performances.

TL;DR: This paper introduces parameterized component analysis to learn a subspace that is invariant to affine (or higher order) geometric transformations in person-specific facial appearance models (PSFAM), which use modular PCA to model complex intra-person appearance changes.

TL;DR: This paper casts the problem of 2D point-set alignment and correspondence matching into a unified framework to constrain the recovery of pose parameters using relational constraints provided by the structural arrangement of the points.

TL;DR: A new approach for brain surface matching by determining the correspondence of 3D point sets between pairs of surfaces using a combination of geodesic distance and surface curvature is presented.