Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks365 from public data
- 20,865
- 3,862
- Multiple view geometry in computer visiond2,863
The book covers the geometric principles and how to represent objects algebraically so they can be computed and applied and explains how to apply the methods and implement the algorithms directly.
- 2,734
- 2,080
- In defense of the eight-point algorithm1,931
This paper shows that by preceding the eight-point algorithm with a very simple normalization (translation and scaling) of the coordinates of the matched points, results are obtained comparable with the best iterative algorithms.
- 829
- 818
- Estimation of relative camera positions for uncalibrated cameras705
A non-iterative algorithm is given for determining the focal lengths of the two cameras, as well as their relative placement, assuming all other internal camera parameters to be known.
- Subexpression sharing in filters using canonic signed digit multipliers661
This paper examines methods for optimizing the design of CSD multipliers, and in particular the gains that can be made by sharing subexpressions, and it is shown that sharing the two most common sub expressions can be expected to lead to a 33% saving of the number of additions.
- Rotation Averaging650
This paper is conceived as a tutorial on rotation averaging, summarizing the research that has been carried out in this area and contains many new results, which were developed to fill gaps in knowledge, answering fundamental questions such as radius of convergence of the algorithms, and existence of local minima.
- 645
- 564
- 537
- Linear pushbroom cameras497
The linear pushbroom model introduced in this paper has the advantage of computational simplicity while at the same time giving very accurate results compared with the full orbiting push broom model.
- Euclidean reconstruction from uncalibrated views478
A practical algorithm for Euclidean reconstruction from several views with the same camera is given and is shown to behave very robustly in the presence of noise giving excellent calibration and reconstruction results.
- Theory and Practice of Projective Rectification473
A new method for image rectification, the process of resampling pairs of stereo images taken from widely differing viewpoints in order to produce a pair of “matched epipolar projections”, based on an examination of the fundamental matrix of Longuet-Higgins which describes the epipolar geometry of the image pair.
- Multiframe Motion Segmentation with Missing Data Using PowerFactorization and GPCA446
This algorithm involves projecting all point trajectories onto a 5-dimensional subspace using the SVD, the PowerFactorization method, or RANSAC, and fitting multiple linear subspaces representing different rigid-body motions to the points in ℝ5 using GPCA.
- Learning to Estimate Hidden Motions with Global Motion Aggregation436
This paper argues that the occlusion problem can be better solved in the two-frame case by modelling image self-similarities, and introduces a global motion aggregation module, a transformer-based approach to find long-range dependencies between pixels in the first image, and perform global aggregation on the corresponding motion features.
- 418
- Self-calibration from multiple views with a rotating camera413
There is no epipolar structure since all images are taken from the same point in space and determination of point matches is considerably easier than for images taken with a moving camera, since problems of occlusion or change of aspect or illumination do not occur.
- 378
- Bringing a Blurry Frame Alive at High Frame-Rate With an Event Camera341
This paper proposes a simple and effective approach, the Event-based Double Integral (EDI) model, to reconstruct a high frame-rate, sharp video from a single blurry frame and its event data, based on solving a simple non-convex optimization problem in a single scalar variable.
- High-quality Image-based Interactive Exploration of Real-World Environments328
This paper enables users to virtually explore large environments from the comforts of their sofa with a system that marries the realism of high quality imagery acquired from real world locations together with the 3D interactivity typical of computer games.
- Kernel Methods on the Riemannian Manifold of Symmetric Positive Definite Matrices316
To encode the geometry of the manifold in the mapping, a family of provably positive definite kernels on the Riemannian manifold of SPD matrices is introduced, derived from the Gaussian kernel, but exploit different metrics on the manifold.
- Self-Calibration of Stationary Cameras298
A new practical method is given for the self-calibration of a camera where at least three images are taken from the same point in space with different orientations of the camera and calibration is computed from an analysis of point matches between the images.
- 292
- 282
- Lines and Points in Three Views and the Trifocal Tensor280
It is shown in this paper, that the trifocal tensor is essentially identical to a set of coefficients introduced by Shashua to effect point transfer in the three view case, which means that the 13-line algorithm may be extended to allow for the computation of the Trifocal Tensor given any mixture of sufficiently many line and point correspondences.
- Kernel Methods on Riemannian Manifolds with Gaussian RBF Kernels270
It is shown that many popular algorithms designed for Euclidean spaces, such as support vector machines, discriminant analysis and principal component analysis can be generalized to Riemannian manifolds with the help of such positive definite Gaussian kernels.
- Global Optimization through Rotation Space Search260
A method is developed for the estimation of the essential matrix, giving the first guaranteed optimal algorithm for estimating the relative pose using a cost function based on reprojection errors.
- Projective reconstruction and invariants from multiple images254
It is proven that once the epipolar geometry is known, the configurations of many geometric structures are determined up to a collineation of projective 3-space /spl Pscrsup 3/ by their projection in two independent images.
- 243
- Face Super-Resolution Guided by Facial Component Heatmaps239
This paper proposes a method that explicitly incorporates structural information of faces into the face super-resolution process by using a multi-task convolutional neural network (CNN) and achieves superior face hallucination results and outperforms the state-of-the-art.
- Multiple-View Geometry Under the {$L_\infty$}-Norm233
It is shown that a variety of structure and motion problems, for example, triangulation, camera resectioning and homography estimation can be recast as quasi-convex optimization problems within this framework.
- Dimensionality Reduction on SPD Manifolds: The Emergence of Geometry-Aware Methods228
This paper proposes to model the mapping from the high-dimensional SPD manifold to the low-dimensional one with an orthonormal projection and shows that learning can be expressed as an optimization problem on a Grassmann manifold and discusses fast solutions for special cases.
- 222
- Kruppa's equations derived from the fundamental matrix216
Kruppa's equations can be written explicitly in terms of the singular value decomposition (SVD) of the fundamental matrix.
- From Manifold to Manifold: Geometry-Aware Dimensionality Reduction for SPD Matrices214
An approach that lets us handle high-dimensional SPD matrices by constructing a lower-dimensional, more discriminative SPD manifold is introduced and leads to a significant accuracy gain over state-of-the-art methods.
- Sparse Coding and Dictionary Learning for Symmetric Positive Definite Matrices: A Kernel Approach212
Experiments show that the proposed sparse coding approach achieves notable improvements in discrimination accuracy, in comparison to state-of-the-art methods such as tensor sparse coding, Riemannian locality preserving projection, and symmetry-driven accumulation of local features.
- Deep Unsupervised Saliency Detection: A Multiple Noisy Labeling Perspective207
This work presents a novel perspective to unsupervised saliency detection through learning from multiple noisy labeling generated by "weak" and "noisy" unsuper supervised handcrafted saliency methods.
- 205
- 186
- Self-calibration from image triplets182
It is shown that affine calibration is recovered uniquely, and metric calibration up to a two fold ambiguity.
- Monocular Template-based Reconstruction of Inextensible Surfaces178
A monocular 3D reconstruction algorithm for inextensible deformable surfaces that uses point correspondences between a single image of the deformed surface taken by a camera with known intrinsic parameters and a template to recover the 3D surface shape as seen in the image.
- 176
- 175
- 146
- 144
- Digit-serial processing techniques139
It is shown that digit-serial computation gives rise to particularly efficient chip designs, and that choice of digit-size allows the user to match throughput requirements to specifications.
- 138
- 135
- Digit-Serial Computation133
This document describes the development of Digit-Serial Architecture, a system for modeling and modeling the behaviour of discrete-time circuits, and some of the techniques used in that process.
- 129
- Removing outliers using the L∞ Norm127
It is proved that the set of measurements with greatest residual must contain at least one outlier, and it is shown that even simple strategies for throwing out these maximum residual measurements are effective in removing outliers.
- Perspective Nonrigid Shape and Motion Recovery123
It is shown that, by exploiting algebraic dependencies among the entries of the projection matrices, one can upgrade the projective reconstruction to determine the affine configuration of the points in �”3, and the motion of the camerarelative to their centroid.
- 117
- Optimal Algorithms in Multiview Geometry114
This is a survey paper summarizing recent research aimed at finding guaranteed optimal algorithms for solving problems in Multiview Geometry in terms of minimizing residual imageplane error.
- PowerFactorization : 3D reconstruction with missing or uncertain data114
The method for incomplete data is applied to several problems of 3D reconstruction, generalizing the Tomasi-Kanade method for orthographic cameras and the Sturm-Triggs method for projective cameras to missing and uncertain data.
- 108
- 108
- 106
- Real-Time Simulation of Medical Ultrasound from CT Images105
A simple acoustic model of the ultrasound for the run-time phase is developed, which produces realistic ultrasound images in real-time when combined with the previously computed scattering image.
- Expanding the Family of Grassmannian Kernels: An Embedding Perspective101
This work introduces several positive definite Grassmannian kernels, including universal ones, and demonstrates their superiority over previously-known kernels in various tasks, such as classification, clustering, sparse coding and hashing.
- 98
- 96
- 94
- Extrinsic Methods for Coding and Dictionary Learning on Grassmann Manifolds93
The aim of building a bridge between the two realms is to address the problem of sparse coding and dictionary learning in Grassmann manifolds, i.e., the space of linear subspaces by embedding into higher dimensional Hilbert spaces by an isometric mapping.
- An overview of the e‐book marketplace93
An overview of the e‐book marketplace players and their services against a context in which e‐books are becoming an increasingly significant category of digital resource is provided.
- Parallel computation of mutual information on the GPU with application to real-time registration of 3D medical images92
This work proposes a novel method named sort and count for efficient parallelization of mutual information (MI) computation designed for massively multi-processing architectures that achieves real-time (less than 1s) rigid registration of 3D medical images using a commodity graphics processing unit (GPU).
- 92
- 90
- 89
- Learning Optical Flow from a Few Matches87
This paper proposes an alternative displacement representation, named Sparse Correlation Volume, which is constructed directly by computing the k closest matches in one feature map for each feature vector in the other feature map and stored in a sparse data structure.
- Action Anticipation with RBF Kernelized Feature Mapping RNN87
A novel Recurrent Neural Network-based algorithm for future video feature generation and action anticipation called feature mapping RNN, which uses only some of the earliest frames of a video to generate future features with a fraction of the parameters needed in traditional RNN.
- Single Image Optical Flow Estimation With an Event Camera85
Experimental results on both synthetic and real data (with blurred and non-blurred images) show the superiority of the model in comparison to state-of-the-art approaches.
- 85
- Chirality81
It is shown that the scene is determined from two views up to a more restricted class of mappings known as quasi-affine transformations, which are precisely those projectivities that preserve the convex hull of an object of interest.
- 79
- Sparse Coding on Symmetric Positive Definite Manifolds Using Bregman Divergences78
This paper proposes to seek sparse coding by embedding the space of SPD matrices into the Hilbert spaces through two types of the Bregman matrix divergences, which not only leads to an efficient way of performing sparse coding but also an online and iterative scheme for dictionary learning.
- Monocular Template-Based Reconstruction of Smooth and Inextensible Surfaces75
The idea is to say that the 2D-3D map must be everywhere a local isometry, which induces conditions on the Jacobian matrix of the map which are included in a least-squares minimization problem.
- 75
- 74
- Phase-Only Image Based Kernel Estimation for Single Image Blind Deblurring73
It is shown that the auto-correlation of the absolute phase-only image 1 can provide faithful information about the motion that caused the blur, leading to a new and efficient blur kernel estimation approach.
- Multi-Target Tracking With Time-Varying Clutter Rate and Detection Profile: Application to Time-Lapse Cell Microscopy Sequences73
A bootstrap filter composed of an estimator and a tracker based on the random finite set Bayesian filtering framework that can outperform state-of-the-art particle trackers on both synthetic and real data in this regime.
- 73
- 72
- 72
- 71
- Joint dimensionality reduction and metric learning: a geometric take69
A Riemannian framework is developed to jointly learn a mapping performing dimensionality reduction and a metric in the induced space and it is shown that this approach yields competitive runtimes with and higher accuracy than state-of-the-art metric learning algorithms.
- 68
- Reconstruction from Projections Using Grassmann Tensors68
It is shown that similar methods will apply in a much more general context, and hence may be applied to projections from �‘n to ℘m, which have been used in the analysis of dynamic scenes, and in radial distortion correction.
- Minimizing algebraic error68
This paper gives an efficient method of minimizing algebraic distance while taking account of the constraints and provides new algorithms for the problems of resectioning a pinhole camera, computing the fundamental matrix, and computing the tri–fsfocal tensor.
- 67
- Encoding Cortical Surface by Spherical Harmonics66
The weighted spherical harmonic representation is a surface smoothing technique given explicitly as a weighted linear combination of spherical harmonics that is used to parameterize cortical surfaces, establish the hemispheric correspondence, and normalize cortex surfaces in a unified mathematical framework.
- 65
- 64
- 64
- 63
- 62
- 61
- Critical Configurations for Projective Reconstruction from Multiple Views60
This paper investigates a classical problem in computer vision: Given corresponding points in multiple images, when is there a unique projective reconstruction of the 3D geometry of the scene points and the camera positions, and shows that in most cases, the ambiguity extends to any number of cameras.
- Shape from non-homogeneous, non-stationary, anisotropic, perspective texture60
The key idea is that the frontal texture is estimated, and a correct estimation leads to the most consistent surface, and in addition to surface shape, a frontal view of the texture is also recovered.
- A Six Point Solution for Structure and Motion60
An m view n ≥ 6 point robust reconstruction algorithm which uses the 6 point method as a search engine and extends the successful RANSAC based algorithms for 2-views and 3-views to m views.
- BAGM: A Backdoor Attack for Manipulating Text-to-Image Generative Models59
A Backdoor Attack on text-to-image Generative Models (BAGM), which upon triggering, infuses the generated images with manipulative details that are naturally blended in the content.
- Computation of the quadrifocal tensor59
It is shown by experiments with synthetic data that the iterative methods, though minimizing algebraic error, rather than more correctly geometric error measured in the image, give almost optimal results.
- Behavioral to structural translation in a bit-serial silicon compiler59
The bit-serial language (BSL) is defined, and the details of the behavioral-to-structural translation are given.
- A Fast Optimal Algorithm for L 2 Triangulation58
This paper presents a practical method for obtaining the global minimum to the least-squares (L2) triangulation problem and proposes a simpler branch-and-bound algorithm to approach the global estimate.
- Dual Pixel Exploration: Simultaneous Depth Estimation and Image Restoration57
This paper proposes an end-to-end DDDNet (DP-based Depth and Deblur Network) to jointly estimate the depth and restore the image, and defines a re-blur loss, which reflects the relationship of the DP image formation process with depth information, to regularise the depth estimate in training.
- 57
- 57
- 56
- The Cubic Rational Polynomial Camera Model55
It is empirically demonstrated that a SAR sensor is very accurately approximated by a cubic camera, but not by any linear camera model, and an algorithm for estimating the parameters of the cubic camera is outlined, given a set of image to world correspondences.
- 52
- 51
- Motion Anything: Any to Motion Generation50
Motion Anything is proposed, a multimodal motion generation framework that introduces an Attention-based Mask Modeling approach, enabling fine-grained spatial and temporal control over key frames and actions, and introduces Text-Music-Dance (TMD), a new motion dataset consisting of 2,153 pairs of text, music, and dance.
- Visual Odometry for Non-overlapping Views Using Second-Order Cone Programming50
A solution for motion estimation for a set of cameras which are firmly mounted on a head unit and do not have overlapping views in each image is presented, which can solve visual odometry by using SOCP as well.
- L/sub X/ minimization in geometric reconstruction problems50
It is shown that the optimal solution was reliably achieved on problems of small dimension and may be recast as a constrained minimization problem and solved using commonly available software.
- 49
- Monocular template-based 3D surface reconstruction: Convex inextensible and nonconvex isometric methods48
Experimental results on simulated and real data show that the convex point-wise method and the nonconvex method outperform respectively current initialization and refinement methods in 3D reconstructed surface accuracy.
- Pyramid Center-Symmetric Local Binary/Trinary Patterns for Effective Pedestrian Detection48
Experiments show that the proposed pyramid center-symmetric local binary/ternary patterns (pyramid CS-LBP/LTP) feature outperforms the histograms of oriented gradients feature and is comparable with the start-of-the-art pyramid HOG (PHOG) feature when using the intersection kernel support vector machines (HIKSVMs).
- EpO-Net: Exploiting Geometric Constraints on Dense Trajectories for Motion Saliency47
It is shown that by combining epipolar distances with optical flow, a powerful motion network can be learned and a robust fusion mechanism between motion and appearance-based features is proposed, which outperforms the previous methods on DAVIS-2016, 2017 and Segtrackv2 dataset.
- Sensitivity of Calibration to Principal Point Position47
It is demonstrated that the determination of the focallength of the camera is tied up very closely with the estimate of the principal point, and the relative uncertainty in the focal length is inversely proportional to the distance of thePrincipal point to the epipolar line.
- 47
- 46
- 46
- Non-linear Temporal Subspace Representations for Activity Recognition45
A novel pooling method is proposed, kernelized rank pooling, that represents a given sequence compactly as the pre-image of the parameters of a hyperplane in a reproducing kernel Hilbert space, projections of data onto which captures their temporal order.
- The e-book management framework: The management of e-books in academic libraries and its challenges45
The resultant e-book management framework identifies the processes associated with the management of e-books, and also offers insights into the challenges and issues associated with each stage.
- Choosing e‐books: a perspective from academic libraries45
Research is reported on into the criteria and processes that academic libraries use to choose e‐books, including business models, licence, price, platform, interface, subject coverage, and match to reading lists.
- Ambiguous Configurations for 3-View Projective Reconstruction44
It is shown that for any placement of three given cameras there always exists a critical set consisting of a fourth-degree curve such that a configuration of any number of cameras placed on this curve is critical for the set of points.
- Hallucinating Unaligned Face Images by Multiscale Transformative Discriminative Networks43
An end-to-end multiscale transformative discriminative neural network devised for super-resolving unaligned and very small face images of different resolutions ranging from 16 to 32 pixels in a unified framework that significantly outperforms the state-of-the-art face hallucination networks.
- Scalable Deep k-Subspace Clustering41
This paper introduces a method that simultaneously learns an embedding space along subspaces within a deep neural network to minimize a notion of reconstruction error, thus addressing the problem of subspace clustering in an end-to-end learning paradigm.
- Globally Optimal Inlier Set Maximization with Unknown Rotation and Focal Length41
This paper proposes a novel globally optimal approach based on branch-and-bound that computes the rotation and the focal length maximizing the number of inlier correspondences and considers the reprojection error in the image space.
- Optimizing over Radial Kernels on Compact Manifolds41
It is shown that positive definite radial kernels on the unit n-sphere, the Grassmann manifold and Kendall's shape manifold can be expressed in a simple form whose parameters can be automatically optimized within a support vector machine framework.
- 41
- 41
- Three-View Geometry41
This work returns to two- and three-view geometry but now within a more general framework which naturally extends to four- and n -views, and covers the computation of a reconstruction from multiple views.
- Localisation using an image-map40
A voting technique for a 3d structure and a run-time efficiency of kd-tree are improved to allow a quick finding of similar images and a quick localisation of a visual map using images from a typical digital camera.
- 40
- Hyperspectral video restoration using optical flow and sparse coding39
An algorithm for recovering dense hyperspectral video of dynamic scenes from a few measured multispectral bands per frame using optical flow and sparse coding is presented.
- Epipolar Geometry and the Fundamental Matrix38
The epipolar geometry is the intrinsic projective geometry between two views, and only depends on the cameras' internal parameters and relative pose, and the fundamental matrix F encapsulates this intrinsic geometry, a 3 × 3 matrix of rank 2.
- Relationship between photogrammmetry and computer vision37
The relationship between photogrammetry and computer vision is examined and the shared goals as well as distinct approaches are identified and improved dialogue is essential.
- Tracking by Detection for Interactive Image Augmentation in Laparoscopy34
Because laparoscopic images are known to be extremely difficult to match, it is proposed to use Shape-from-Shading and conformal flattening to cancel out much of the variation in appearance due to perspective foreshortening, and to apply robust matching to the flattened surfaces.
- 34
- Face Reconstructions using Flesh Deformation Modes32
A 3D computer model of the face and head is built using computer 3D graphics techniques to model the variation of the facial shape and Principal Component Analysis is used as a means of exploring the space of the major facial deformations.
- 32
- N-View Geometry32
This work returns to two- and three-view geometry but now within a more general framework which naturally extends to four- and n -views, and covers the computation of a reconstruction from multiple views.
- Reconstruction from two views using approximate calibration32
The cause of this problem is investigated, an algorithm is suggested that more reliably leads to a reconstruction, and Euclidean reconstruction that suffers only a very small degradation in residual point-reprojection error is suggested.
- High precision X-ray stereo for automated 3D CAD-based inspection32
This paper presents a new method that achieves fast, accurate, internal 3D geometry measurement based on 3D reconstruction from a few X-ray views of a part, and describes a novel application of the system to CAD-based verification of drilled hole positioning.
- Learning Saliency From Single Noisy Labelling: A Robust Model Fitting Perspective31
This paper calls on the theory of robust model fitting and formulate deep saliency prediction from a single noisy labelling as robust network learning and exploit model consistency across iterations to identify inliers and outliers.
- 31
- 31
- InfiniMotion: Mamba Boosts Memory in Transformer for Arbitrary Long Motion Generation30
The Motion Memory Transformer with Bidirectional Mamba Memory is introduced, enhancing the transformer's memory to process long motion sequences effectively without overwhelming computational resources, showcasing significant advancements in long motion generation.
- CullNet: Calibrated and Pose Aware Confidence Scores for Object Pose Estimation30
The problem of culling false positives among several pose proposal estimates is addressed and a new network called CullNet, solving this task is presented, found to be significantly more reliable for accurate object pose estimation.
- 30
- 30
- 30
- 29
- Rotation averaging and weak convexity28
The concept of geodesic convexity in the Special Orthogonal Group SO(3) is generalized and strong and new theorems about the location of global minima of the rotation averaging cost function are derived.
- Visual localization and loop-back detection with a high resolution omnidirectional camera28
An automatic scheme to find correspondences from a video sequence to create the map and a vote transfer technique that uses prior knowledge about correspondences within the map to localize a query image instead of a majority vote is used.
- IMPUS: Image Morphing with Perceptually-Uniform Sampling Using Diffusion Models27
This work proposes a heuristic bottleneck constraint based on a novel relative perceptual path diversity score that automatically controls the bottleneck size and balances the diversity along the path with its directness and proposes a perceptually-uniform sampling technique that enables visually smooth changes between the interpolated images.
- 27
- 26
- 26
- 26
- Where's the weetbix?25
Through the study this new retrieval problem proves itself to be a challenging task and an instant application of it is to help the customer find what they want without physically wandering around the shelves but a wide range of potential applications could be expected.
- Automatic Dry Eye Detection24
A new method for the automated detection of dry areas in videos taken after instilling fluorescein in the tear film is developed, consisting of a multi-step algorithm to first locate the iris in each image, then align the images and finally analyze the aligned sequence in order to find the regions of interest.
- Quantifying Bias in Text-to-Image Generative Models23
This work proposes an evaluation methodology to quantify general bias in T2I generative models without any preconceived notion, and introduces a suite of three metrics; namely, distribution bias, Jaccard hallucination and generative miss-rate, to extensively appraise general model bias.
- Manifold Learning Benefits GANs23
This paper improves Generative Adversarial Net-works by incorporating a manifold learning step into the discriminator, and finds that locality-constrained non-linear manifolds outperform linear mani-folds due to their non-uniform density and smoothness.
- Optimizing pipelined networks of associative and commutative operators23
A method of tree-height minimization of networks of commutative and associative operators is described, which aims at minimizing latency and shimming delays in a synchronous data flow architecture such as that used in bit/digit-serial computation.
- A new simultaneous circuit partitioning and chip placement approach based on simulated annealing23
This work proposes a new approach in which the circuit is both partitioned and placed simultaneously by a simulated annealing based algorithm, seen to yield excellent results in reasonable run times.
- 22
- Proximal Mean-Field for Neural Network Quantization22
This work designs an efficient iterative optimization procedure that involves stochastic gradient descent followed by a projection and proves that this simple projected gradient descent approach is, in fact, equivalent to a proximal version of the well-known mean-field method.
- 22
- LDP: Language-driven Dual-Pixel Image Defocus Deblurring Network21
This paper proposes the first framework that introduces the contrastive language-image pre-training framework (CLIP) to accurately estimate the blur map from a DP pair unsu-pervisedly, and achieves state-of-the-art performance in extensive experiments.
- 21
- Hippocampal shape analysis for Alzheimer's disease using an efficient hypothesis test and regularized discriminative deformation21
A framework to perform statistical shape analysis for segmented hippocampi is presented, including an efficient permutation test for detecting subtle class differences, and a regularized discriminative direction method for visualizing the shape discrepancy.
- 21
- Camera Calibration Using Line Correspondences21
It is shown that the the camera may be calibrated at the same time that the scene geometry is determined, and the scene may be reconstructed exactly (up to a scaled Euclidean transformation).
- 21
- Plane-Based Calibration and Auto-calibration of a Fish-Eye Camera20
By taking images of a planar scene with a fish-eye camera, this method automatically estimates the centre-of-distortion (COD), distortion parameters, and other conventional camera intrinsics by a three-stage algorithm.
- Multilinear Relationships between Coordinates of Corresponding Image Points and Lines20
How the fundamental matrix, F, the trifocal tensor T jk i and the quadrilinear relationship existing between corresponding points in four uncalibrated projective images may be derived in a common framework involving matrix determinants is described.
- 20
- 19
- A Multiple Model Probability Hypothesis Density Tracker for Time-Lapse Cell Microscopy Sequences19
This work proposes a new closed-form recursion that incorporates a state-dependent transition probability matrix and introduces a general framework for particle tracking using the proposed filter to multi-target tracking in total internal reflection fluorescence microscopy sequences.
- 19
- 18
- Application of the IMM-JPDA Filter to Multiple Target Tracking in Total Internal Reflection Fluorescence Microscopy Images18
A multi-target tracking method using an Interacting Multiple Model Joint Probabilistic Data Association (IMM-JPDA) filter for tracking vesicles in total internal reflection fluorescence microscopy (TIRFM) sequences is proposed.
- 18
- 18
- 18
- 18
- 18
- A Generalized Projective Reconstruction Theorem and Depth Constraints for Projective Factorization17
A generalized version of the classic projective reconstruction theorem is presented, which enables us to present classes of depth constraints under which any reconstruction of cameras and points projecting into given image points is projectively equivalent to the true camera-point configuration.
- 17
- Recognizing Objects Using Color-Annotated Adjacency Graphs17
A new algorithm for identifying objects in cluttered images, based on approximate subgraph matching, that is robust under moderate variations in the camera viewpoints and somewhat insensitive to minor graph variations is introduced.
- Dualizing Scene Reconstruction Algorithms17
The nature of the duality mapping is explored and its application to reconstruction ambiguity is discussed and an explicit method for dualizing any projective reconstruction algorithm is given.
- 17
- Post-hoc Calibration of Neural Networks.16
It is proved that even though the base network does not lead to the global optimum of NLL, by adding additional layers and minimizing NLL by optimizing the parameters of g one can obtain a calibrated network and provides a theoretical justification of post-hoc calibration methods.
- Identity-Preserving Face Recovery from Portraits16
A new Identity-preserving Face Recovery from Portraits (IFRP) to recover latent photorealistic faces from unaligned stylized portraits by embedding spatial transformer networks into the SRN, which can compensate for misalignments of stylized faces automatically and output aligned realistic face images.
- Iteratively reweighted graph cut for multi-label MRFs with non-convex priors16
This work introduces an algorithm that iteratively approximates the original energy with an appropriately weighted surrogate energy that is easier to minimize and demonstrates the benefits of the method over state-of-the-art MRF energy minimization techniques on stereo and inpainting problems.
- 16
- 16
- Pairwise Similarity Knowledge Transfer for Weakly Supervised Object Localization15
This work argues that learning only an objectness function is a weak form of knowledge transfer and proposes to learn a classwise pairwise similarity function that directly compares two input proposals as well.
- 15
- 15
- 15
- Fixed-Lens camera setup and calibrated image registration for multifocus multiview 3D reconstruction14
This paper shows how a new multifocus multiview 3D reconstruction procedure which includes a new Fixed-Lens multifocus image capture and a calibrated image registration technique using analytic homography transformation significantly reduce the errors in the camera poses and produce more complete 3D reconstructed models as compared with those by the conventional moving lens image Capture and multifocus stacking.
- $${L_q}$$ L q -Closest-Point to Affine Subspaces Using the Generalized Weiszfeld Algorithm14
Experimental results for the triangulation problem confirm that the L_q$$Lq-closest-point method, for $$1 \le q < 2$$1≤q<2, is more robust to outliers than the $$L_2$$L2-clOSest- point method.
- Enforcing Monotonic Temporal Evolution in Dry Eye Images14
This work addresses the problem of identifying dry areas in the tear film as part of a diagnostic tool for dry-eye syndrome by segmenting dry regions using a multi-label graph-cut algorithm on the 3D spatio-temporal volume of frames from a video sequence.
- Conformal spherical representation of 3D genus-zero meshes14
This paper describes an approach of representing 3D shape by using a set of invariant spherical harmonic coefficients after conformal mapping, where the modification is based on Mobius factorization and aims at obtaining a canonical conformal mapping.
- 14
- 14
- 14
- The 2nd Workshop on UAVs in Multimedia: Capturing the World from a New Perspective13
This workshop will cover various topics related to UAV multimedia, including aerial image and video processing, machine learning for UAV data analysis, UAV swarm technology, and UAV-based multimedia applications.
- 13
- FlashMo: Geometric Interpolants and Frequency-Aware Sparsity for Scalable Efficient Motion Generation12
This work proposes FlashMo, a frequency-aware sparse motion diffusion model that prunes low-frequency tokens to enhance efficiency without custom kernel design, and introduces MotionSiT, a scalable diffusion transformer based on a joint-temporal factorized interpolant with Lie group geodesics over SO(3) manifolds, enabling principled generation of joint rotations.
- Probability Density Geodesics in Image Diffusion Latent Space12
This work shows that geodesics can be computed in diffusion latent space, where the norm induced by the spatially-varying inner product is inversely proportional to the probability density.
- Adversarial Purification with the Manifold Hypothesis12
This work forms a novel framework for adversarial robustness using the manifold hypothesis, which provides sufficient conditions for defending against adversarial examples and develops an adversarial purification method that combines manifold learning with variational inference.
- A learning-based markerless approach for full-body kinematics estimation in-natura from a single image12
The method opens new vistas to robust user-interaction free estimation of full body kinematics, a prerequisite to motion analysis, and achieves a score of 91.1 and 69.1 on mean Probability of Correct Keypoint measure and Average Precision of Keypoints measure for the frontal and sagittal datasets respectively.
- Homography Estimation with L ∞ Norm Minimization Method12
By using ∞ L image error instead of 2 L Euclidean distance, it is shown that the problem is still a quasi-convex problem and can be solved by bisection method but with linear programming (LP), and therefore the computation time is decreased.
- 12
- Exploring Bias in over 100 Text-to-Image Generative Models11
Analysis of bias trends in text-to-image generative models over time indicates that artistic and style-transferred models exhibit significant bias, whereas foundation models, benefiting from broader training distributions, are becoming progressively less biased.
- 11
- Learning Arbitrary Potentials in CRFs with Gradient Descent.11
A new inference and learning framework which can learn arbitrary pairwise CRF potentials is introduced based on the observation that CRF inference can be achieved via projected gradient descent and consequently, can be integrated in deep neural networks to allow for end-to-end training.
- Memory Efficient Max Flow for Multi-label Submodular MRFs11
This paper introduces a variant of the max-flow algorithm that requires much less storage and makes it possible to optimally solve multi-label submodular problems involving large numbers of variables and labels on a standard computer.
- Pipeline reconstruction from fisheye images11
This work presents a system which can reconstruct the inner surface of buried pipelines from multiple fisheye images captured inside the pipes, which will facilitate pipe condition analysis.
- 11
- 11
- 10
- 10
- 10
- 10
- 10
- 10
- Joint Unsupervised Learning of Optical Flow and Egomotion with Bi-Level optimization9
This work exploits the global relationship between optical flow and camera motion using epipolar geometry and uses implicit differentiation to enable backpropagation through the lower-level geometric optimization layer independent of its implementation, allowing end-toend training of the network.
- 9
- 9
- 9
- 9
- Translation Estimation from Omnidirectional Images9
A translation estimation method from omnidirectional images with missing data that shows more reliable results than the previous method and gives a robust result when there are measurement errors in the data.
- 9
- Invariants of Points Seen in Multiple Images9
This paper investigates projective invariants of geometric configurations in 3 dimensional projective space P, and most particularly the computation of invariants from two or more independent images, and the essential matrix defined by Longuet-Higgins is described.
- 9
- 8
- Few-shot Weakly-Supervised Object Detection via Directional Statistics8
This work proposes a probabilistic multiple-instance learning approach for few-shot Common Object Localization (COL) and few-shots Weakly Supervised Object Detection (WSOD) and finds that operating on features extracted from the last layer of a pretrained Faster-RCNN is more effective compared to previous episodic learning based few- shot COL methods.
- Kernels on Riemannian Manifolds8
Positive definite kernels on manifolds are defined that permit us to embed a given manifold with a corresponding metric in a reproducing kernel Hilbert space and make it possible to utilize algorithms developed for linear spaces on nonlinear manifold-valued data.
- 8
- Scientific publishing: a case study of Libyan scientists8
A bibliometric analysis of the papers published by scientists in research centres affiliated to Libya's National Centre for Scientific Research was conducted in order to investigate the extent and nature of scholarly communication of Libyan scientists.
- 8
- In Defence of the 8-point Algorithm8
It is shown that by preceding the g-point algorithm with a very simple normalization (translation and scaling) of the coordinates of the matched points, results are obtained comparable with the best iterative algorithms.
- Camera Geometry and Single View Geometry8
The geometry of a single perspective camera is concentrated on the geometry of a single perspective camera and the action of a camera on geometric objects other than finite points.
- 8
- 8
- 8
- 8
- Drawing polygons given angle sequences8
This paper presents an algorithm that, given a sequence of exterior angles summing to 2π, construct a simple polygon having that sequence of external angles.
- 7
- Dictionary Learning on Grassmann Manifolds7
This work addresses the problem of sparse coding and dictionary learning in Grassmann manifolds by embedding Grassman manifolds into the space of symmetric matrices and proposes positive definite kernels to handle nonlinearity in data.
- Sequential L infinity Norm Minimization for Triangulation7
This paper shows that the upper and lower bounds, which are two input parameters of the bi-section method, can be updated through the time sequence so that the gap between the two bounds is kept as small as possible.
- A Study of Hippocampal Shape Difference Between Genders by Efficient Hypothesis Test and Discriminative Deformation7
The Fisher permutation and bootstrap tests are used to differentiate hippocampal shape between genders and a "informative direction" is found to preserve the original discriminative information.
- Sequential L ∞ Norm Minimization for Triangulation7
This paper shows that the upper and lower bounds, which are two input parameters of the bi-section method, can be updated through the time sequence so that the gap between the two bounds is kept as small as possible.
- 7
- RANP: Resource Aware Neuron Pruning at Initialization for 3D CNNs6
A Resource Aware Neuron Pruning (RANP) algorithm that prunes 3D CNNs at initialization to high sparsity levels by obtaining an importance score for each neuron based on their sensitivity to the loss function and reweighting according to neuron resource consumption related to FLOPs or memory.
- 6
- How Useful is Projective Geometry?6
In this response, two different approaches to shape recognition are weighed: the use of restricted camera models as advocated in the paper of Pizlo et.
- 5
- 5
- 5
- Fast Multi-labelling for Stereo Matching5
The Divide and Conquer paradigm is used to separate the complexities induced by the label set and the variable set, and deal with each of them respectively, which improves the solution speed without depleting the memory resource.
- 5
- 5
- Statistical Significance as an Aid to System Performance Evaluation5
Using forensic fingerprint identification as a testbed, a statistical framework for analyzing system performance is presented and a set of independent Bernoulli trials is used to predict the success of the system with respect to a particular dataset.
- Fast and Differentiable Message Passing on Pairwise Markov Random Fields4
This work introduces two fast and differentiable message passing algorithms, namely, Iterative Semi-Global Matching Revised (ISGMR) and Parallel Tree-Reweighted Message Passing (TRWP) which are greatly sped up on a GPU by exploiting massive parallelism.
- 4
- 4
- Compressive Evaluation in Human Motion Tracking4
This work proposes a compressive anneal particle filter to exploit sparsity existing in image-based human motion tracking, and introduces a progressive multilevel wavelet decomposition staged at eachAnneal layer to accelerate the compressive evaluation in a coarse-to-fine fashion.
- 4
- Regularized Discriminative Direction for Shape Difference Analysis4
A regularized discrim inative direction is developed by requiring a shape to conform to its population distribution when it deforms along the discriminative direction, and shows superior performance in detecting and localizing the difference of hippocampal shapes for sex.
- New 3D Fourier Descriptors for Genus-Zero Mesh Objects4
A newly developed surface parametrization method–the discrete conformal mapping (DCM)—is used to propose a 3D Fourier Descriptor (3D-FD) for representing and recognizing arbitrarily-complex genus-zero mesh objects.
- The Trifocal Tensor4
The trifocal tensor only depends on the motion between views and the internal parameters of the cameras and is defined uniquely by the camera matrices of the views, however, it can be computed from image correspondences alone without requiring knowledge of the motion or calibration.
- Summary of the Panel Session4
The workshop ended with a 75 minute panel session, with Richard Hartley, P. Anandan, Jitendra Malik, Joe Mundy and Olivier Faugeras as panelists, where each panelist selected a topic related to the workshop theme and gave a short position statement on it.
- 4
- 4
- Motion Estimation for Nonoverlapping Multicamera Rigs: Linear Algebraic and {\rm L}_\infty Geometric3
- 3
- 3
- 3
- 3
- Canonic Signed Digit Arithmetic3
This chapter deals with the subject of constant multiplication through the use of shifts and adds, and extends the domain of interest to include techniques that apply also to pipelined parallel arithmetic.
- 3
- Manipulating and Mitigating Generative Model Biases Without Retraining2
This work proposes a dynamic and computationally efficient manipulation of T2I model biases by exploiting their rich language embedding spaces without model retraining, and shows that leveraging foundational vector algebra allows for a convenient control over language model embeddings to shift T2I model outputs and control the distribution of generated classes.
- 2
- 2
- Bringing Blurry Alive at High Frame-Rate with an Event Camera.2
A simple and effective approach to reconstruct a high-quality and high frame-rate shape video and achieves significant improvements in removing general blurs and reconstructing high temporal resolution video by optimizing the energy model.
- 2
- Estimating the projected frontal surface area of cyclists from images using a variational framework and statistical shape and appearance models2
The method successfully estimates pFSA of cyclists and opens new vistas for exploration of the relationship with aerodynamic drag, using a statistical cyclist shape and appearance models as priors to encourage the evolving curve to arrive at the desired solution.
- 2
- 2
- Projective Reconstruction2
From several images of a scene and the coordinates of corresponding points identified in the different images, it is possible to construct a 3-dimensional point-cloud model of the scene, and compute the camera locations.
- Geometric Alignment of Helical Tomographic Projection Data Using Pi-Lines2
The performance of the self-calibration of conebeam CT systems in the context of helical scanning trajectories is investigated and it is investigated as an enhancement to robustness of the currently used tomogram sharpness maximisation technique.
- Gradual Sampling and Mutual Information Maximisation for Markerless Motion Capture2
Noticing the limitation of silhouettes in resolving gesture ambiguity, this work incorporates appearance information in an illumination invariant way by maximising Mutual Information between an appearance model and the observation, which strengthens the effectiveness of the better-behaved observation likelihood.
- 2
- Feature Matching and Pose Estimation Using Newton Iteration2
This paper proposes a new method that is able to estimate the transformation and find the correct matches simultaneously, and formulate the problem as a nearest-matrix problem, and provides a different numerical technique.
- Rectification-Free Multibaseline Stereo for Non-ideal Configurations2
A new algorithm to process non-ideally configured multibaseline stereo system, which not only avoids the rectification procedure but also remains the efficiency of SSSD at the same time.
- Discrete Conformal Shape Representation and Reconstruction of 3D Mesh Objects2
A conformal representation for genus-zero mesh objects is proposed, by using the discrete conformal mapping technique and a new method to reconstruct the original shape from its conformal representations is proposed.
- 2
- 2
- Self-Calibration of Zooming Cameras from a Single Viewpoint2
A configuration that commonly occurs in many imaging applications is that in which a camera has its optic centre fixed in location but which may rotate and zoom, thus changing its internal parameters.
- Reply to Pizlo, Rosenfeld, and Weiss2
In this note, some of the issues suggested by Pizlo et al.
- 2
- Safety Without Semantic Disruptions: Editing-free Safe Image Generation via Context-preserving Dual Latent Reconstruction1
This work identifies trade-offs between safety and censorship, which presents a necessary perspective in the development of ethical AI models, and achieves state-of-the-art results on safe image generation benchmarks and offers intuitive control over the level of model safety.
- Semi-Supervised 3D Hand Shape and Pose Estimation with Label Propagation1
The Pose Alignment network is proposed to propagate 3D annotations from labelled frames to nearby unlabelled frames in sparsely annotated videos to improve the pose estimation accuracy and incorporate the alignment supervision on pairs of labelled-unlabelled frames.
- 1
- 1
- Deep Learning Superpixel Semantic Segmentation with Transparent Initialization and Sparse Encoder1
This paper jointly learn semantic segmentation with trainable superpixels by adding fully-connected layers with transparent initialization and an efficient logit uniformization with a sparse encoder to reduce the large computational complexity arising from indexing pixels bysuperpixels.
- 1
- 1
- 1
- Globally Optimal Solutions to Vision Using Convex and Quasi-convex Optimization1
This work will talk about progress in this area, through the use of methods such as Second Order Cone Programming, branch-and-bound fractional programming and semidefinite programming.
- Auto-Calibration of a Compound-Type Omnidirectional Camera1
An algorithm for automatically removing lens distortion by exploiting planar projective invariants is capable of estimating centre-of-distortion and distortion parameters efficiently without using any nonlinear iteration.
- 1
- 1
- 1
- 1
- 1
- 1
- Multiview Reconstruction of Static and Dynamic Scenes1
It turns out that for dynamic scenes, tensor- based techniques may be used to solve these problems, in a way analogous to the rigid- scene case.
- Plane-based methods for 3D reconstruction and video enhancement1
Methods whereby the existence of planar or almost planar sections of the scene may be exploited throughout the reconstruction process, and the homography induced by a plane to aid in point tracking, projective reconstruction, self calibration and model building are discussed.
- 1
- 1
- Quantitative measurement of manufactured diamond shape1
A new shape parameter, τ, is introduced to characterize diamond morphology and an approach to compute it from images is described, implemented in an image analysis system currently being used on a regular basis to classify diamonds at a manufacturing facility.
- 1
- Reporting Intersections of Polygons1
An algorithm is given for reporting pairs of intersecting polygons from among a given set of polygons, and the concept of pseudo-coordinates is introduced to handle the difficulty of reporting intersections between rectilinearly oriented rectangles.
- –
- –
- NF-SLAM: Effective, Normalizing Flow-supported Neural Field representations for object-level visual SLAM in automotive applications–
A novel, vision-only object-level SLAM framework for automotive applications representing 3D shapes by implicit signed distance functions by augmenting the standard neural representation by a normalizing flow network is proposed, which shows accurate and reliable performance comparable to alternative frameworks that make use of direct depth readings.
- Direct to CT screening pilot for patients with COPD in a deprived area: longitudinal analysis of the pulmonary nodule cohort outcomes–
Pulmonary nodules consisted significant part of the CT screening pilot findings in COPD patients with significant further conversion rate to lung cancer diagnosis after follow-up, suggesting CT screening of high-risk population in deprived areas has a role in detecting lung cancer and identifying pulmonary nodules.
- Abstract 020: Nox5 Regulates Vascular Smooth Muscle Cell De-differentiation In Human Hypertension–
It is demonstrated that Nox5 plays an important role in VSMC phenotypic switching associated with vascular dysfunction in hypertension and provides new insights into the proteomic changes related to vascular phenotype in hypertension.
- –
- –
- –
- –
- In Defense of Graph Inference Algorithms for Weakly Supervised Object Localization–
It is argued that learning only an objectness function is a weak form of knowledge transfer and proposed to learn a classwise pairwise similarity function that directly compares two input proposals as well, which is computationally efficient for large-scale applications.
- Fast and Differentiable Message Passing for Stereo Vision.–
This work introduces two fast and differentiable message passing algorithms, namely, Iterative Semi-Global Matching Revised (ISGMR) and Parallel Tree-Reweighted Message Passing (TRWP) which are greatly sped up on GPU by exploiting massive parallelism.
- –
- –
- –
- –
- –
- –
- Reliable Point Correspondences in Scenes Dominated by Highly Reflective and Largely Homogeneous Surfaces–
A novel method that uses descriptors evaluated along points in image edges to obtain a sufficiently high proportion of correct point correspondences is proposed that gives better results in recovering the epipolar geometry in scenes dominated by highly reflective and homogeneous surfaces.
- –
- –
- –
- –
- –
- –
- –
- Invariant and Calibration-Free Methods in Scene Reconstruction and Object Recognition.–
It is shown that by taking advantage of known, or hypothesized geometric structure of a scene being viewed, geometric invariants may be used for tasks of segmentation, grouping and object recognition.
- –
- Digit-Serial Cell Design–
The construction of a bit-sliced digit-serial multiplier is the most involved of all the common operators because the height of all types of cells must be the same for a given specified digit-size.
- 4985834 System and method employing pipelined parallel circuit architecture for displaying surface structures of the interior region of a solid body–
These parallel circuits, realizable on large scale integrated circuit chips, perform surface normal calculations, linear interpolations and signal comparisons in simultaneously operating circuit paths which are asynchronously enabled when input data appear.
- –
- Reassembling polygons from edges–
This paper gives a space-efficient algorithm for doing the reassembly of a set of edges of polygons that involves a forward and a backward sweep over the data.
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-14. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.