Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks208 from public data
- Knowledge Distillation: A Survey4,737
A comprehensive survey of knowledge distillation from the perspectives of knowledge categories, training schemes, teacher–student architecture, distillation algorithms, performance comparison and applications is provided.
- 2,515
- 1,191
- Camera self-calibration: Theory and experiments1,153
It is shown, using experiments with noisy data, that it is possible to calibrate a camera just by pointing it at the environment, selecting points of interest and then tracking them in the image as the camera moves.
- A theory of self-calibration of a moving camera819
The feasibility of camera calibration based on the epipolar transformation is demonstrated and two curves of degree six can be obtained in the dual plane such that one of the real intersections of the two yields the correct camera calibration.
- 677
- 667
- 608
- 590
- 585
- 444
- 431
- 378
- Theory of Reconstruction from Image Motion365
Theory of Reconstruction from Image Motion presents the mathematics underlying the reconstruction of camera motion from the movements of points in the camera image, employing mathematical methods drawn from linear algebra, projective geometry, algebraic geometry, the theory of transversality and the theories of least squares approximation.
- 321
- 306
- 293
- 292
- 244
- DCFNet: Discriminant Correlation Filters Network for Visual Tracking230
This paper presents an end-to-end lightweight network architecture, namely Discriminant Correlation Filter Network (DCFNet), where a differentiable DCF (discriminant correlation filter) layer is incorporated into a Siamese network architecture in order to learn the convolutional features and the correlation filter simultaneously.
- Asymmetric 3D Convolutional Neural Networks for action recognition209
The asymmetric 3D-CNN model outperforms all the traditional 3D-CNN models in both effectiveness and efficiency, and its performance is comparable with that of recent state-of-the-art action recognition methods on both benchmarks.
- 185
- 176
- Bridging Composite and Real: Towards End-to-End Deep Image Matting174
This paper proposes a novel Glance and Focus Matting network (GFM), which employs a shared encoder and two separate decoders to learn both tasks in a collaborative manner for end-to-end natural image matting and finds that a carefully designed composition route RSSN that aims to reduce the discrepancies can lead to a better model with remarkable generalization ability.
- 163
- Robust Detection of Degenerate Configurations while Estimating the Fundamental Matrix160
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.
- A Method for Interactive 3D Reconstruction of Piecewise PlanarObjects from Single Images154
This work presents an approach for 3D reconstruction of objects from a single image based on user-provided coplanarity, perpendicularity and parallelism constraints, used to calibrate the image and perform3D reconstruction.
- 153
- 153
- Feedback Graph Convolutional Network for Skeleton-Based Action Recognition152
This is the first work that introduces a feedback mechanism into GCNs for action recognition, and the proposed FGCN achieves the state-of-the-art performance on all three datasets.
- Incremental Tensor Subspace Learning and Its Applications to Foreground Segmentation and Tracking152
Experimental evaluations against state-of-the-art algorithms demonstrate the promise and effectiveness of the proposed incremental tensor subspace learning algorithm, and its applications to foreground segmentation and object tracking.
- Bayesian Tensor Approach for 3-D Face Modeling147
A decoupled probabilistic algorithm, named Bayesian tensor analysis (BTA), which can automatically and suitably determine dimensionality for different modalities of tensor data and empirical studies on expression retargeting justify the advantages of BTA.
- 142
- 138
- 137
- 132
- Algorithm-Dependent Generalization Bounds for Multi-Task Learning131
This paper presents novel algorithm-dependent generalization bounds for MTL by exploiting the notion of algorithmic stability and shows that the similarity of feature structures in MTL will lead to specific regularizations for predicting, which enables the learning algorithms to generalize fast and correctly from a few examples.
- 128
- 120
- The ADVISOR Visual Surveillance System106
The article explores the main difficulties encountered during the design and implementation of ADVISOR and describes the ways in which they were solved and points the way for future development and ways of deployment of the techniques used in the system.
- 103
- Spatiotemporal Attacks for Embodied Agents99
This work generates spatiotemporal perturbations to form 3D adversarial examples, which exploit the interaction history in both the temporal and spatial dimensions and adversarially perturb the physical properties of the contextual objects that appeared in the most important scene views.
- 98
- 97
- Cross-Architecture Knowledge Distillation89
A novel cross-architecture knowledge distillation method is proposed, which supports bi-directional distillation including from Transformer to CNN and from CNN to Transformer and multi-view robust training scheme is presented, to improve the robustness of the framework for distillation.
- 89
- Interaction-Aware Spatio-Temporal Pyramid Attention Networks for Action Classification88
An effective interaction-aware self-attention model which can extract information about the interactions between feature vectors to learn attention maps is proposed which can be embedded in any general CNN to form a video-level end-to-end attention network for action recognition.
- 88
- 82
- A Robust Tracking System for Low Frame Rate Video81
Experimental results demonstrate that the proposed tracking system can effectively tackle the difficulties caused by LFR, and an integral image based parameter calculation is constructed, which greatly reduces the computational load.
- 80
- 80
- DUT: Learning Video Stabilization by Simply Watching Unstable Videos77
This paper attempts to tackle the video stabilization problem in a deep unsupervised learning manner, which borrows the divide-and-conquer idea from traditional stabilizers while leveraging the representation power of DNNs to handle the challenges in real-world scenarios.
- 68
- Human Action Recognition under Log-Euclidean Riemannian Metric68
A new local spatio-temporal feature is proposed to represent the cuboids detected in video sequences that utilizes the covariance matrix to capture the self-correlation information of the low-level features within each cuboid.
- 68
- Hyperspectral Image Spectral-Spatial Feature Extraction via Tensor Principal Component Analysis62
The traditional principal component analysis is extended to its tensorial version tensor PCA, which is applied to the spectral-spatial features of hyperspectral image data, and the classification accuracy obtained is significantly higher than the accuracies obtained by its rivals.
- 61
- 59
- 59
- D2C: Deep cumulatively and comparatively learning for human age estimation56
This work uses a convolutional neural network which can learn discriminative aging features from raw face images without any handcrafting and proposes a novel cumulative hidden layer which is supervised by a point-wise cumulative signal to combat the sample imbalance problem.
- 56
- 53
- Exposure Trajectory Recovery From Motion Blur50
Exposure trajectories are defined, which represent the motion information contained in a blurry image and explain the causes of motion blur, and a novel motion offset estimation framework is proposed to model pixel-wise displacements of the latent sharp image at multiple timepoints.
- Vehicle Trajectory Approximation and Classification48
A variational technique for finding low curvature smooth approximations to trajectories in the plane to find the sequence of internal states for which the observed behaviour of the vehicle has the highest probability.
- Activity recognition using a supervised non-parametric hierarchical HMM47
The problem of classifying human activities occurring in depth image sequences is addressed and a two level hierarchical Hidden Markov Model, with independent Markov chains for the joint positions and depth image pattern, is used to model the features.
- 46
- Semi-Supervised Tensor-Based Graph Embedding Learning and Its Application to Visual Discriminant Tracking45
This paper treats an image patch as a 2-order tensor which preserves the original image structure and proposes a transfer-learningbased semi-supervised strategy to iteratively adjust the embedding space into which discriminative information obtained from earlier times is transferred.
- 45
- 41
- Learning Enhanced 3D Models for Vehicle Tracking40
The method is shown to enhance model-based tracking, particularly in the presence of clutter and occlusion, and to provide a basis for identification and to be compared to previous (edge-based) iconic evaluation techniques.
- 36
- A case against epipolar geometry35
It is demonstrated configurations for which 3D projective invariants can be computed from perspective stereo pairs, but epipolar geometry (and full projective structure) cannot, and the connection with camera calibration is investigated.
- 34
- 34
- Wide-Angle Image Rectification: A Survey33
This paper comprehensively survey progress in wide-angle image rectification from transformation models to rectification methods, evaluating the performance of state-of-the-art methods on public datasets and showing that although both kinds of methods can achieve good results, these methods only work well for specific camera models and distortion types.
- CDPM: convolutional deformable part models for semantically aligned person re-identification32
Convolutional Deformable Part Models (CDPM) works by decoupling the complex part alignment procedure into two easier steps: first, a vertical alignment step detects each body part in the vertical direction, with the help of a multi-task learning model; second, a horizontal refinement step based on attention suppresses the background information around each detected body part.
- Visible and infrared image registration based on region features and edginess32
This paper presents a coarse-to-fine procedure for registering visible and infrared images based on stable region features and edginess and shows that the proposed method provides more robust and accurate registration than the existing methods.
- 31
- Robust hand tracking via novel multi-cue integration30
The experimental results show that the integration algorithm outperforms tracking algorithms that only use a single cue, and the proposed tracking algorithm is more robust in complex environments than other state-of-art algorithms.
- 30
- 30
- 29
- Filter for Car Tracking Based on Acceleration and Steering Angle29
Experiments show that the CUF is better than the EKF at predicting future positions of the car; and the filter outputs can be used to control the measurement process, leading to improved ability to recover from errors in predictive tracking.
- Probabilistic analysis of the application of the cross ratio to model based vision: Misclassification27
It is shown that for a fixed overall probability of misclassification there is a maximum possible model cross ratio σm, and there is the same for a range of invariants useful in model based vision.
- 26
- 26
- 25
- 25
- A filter for visual tracking based on a stochastic model for driver behaviour25
The stochastic model is the basis of a new filter for tracking and predicting the motion of the car, using measurements obtained by fitting a rigid 3D model to a monocular sequence of video images.
- Classification based on the cross ratio24
The performance of the cross ratio in model based vision is quantified for a decision rule which is computationally expensive, but which makes good use of the available information.
- 23
- Leukocyte Recognition Using EM-Algorithm23
This document describes a method for classifying images of blood cells using the maximum a posteriori decision rule using three different classes of cells: Band Neutrophils, Eosinophils and Lymphocytes.
- 22
- Action classification using a discriminative multilevel HDP-HMM22
Experiments with two different datasets show that action class models learnt using the HDP-HMM technique produce good classification results.
- Properties of essential matrices22
It is shown that there are at most three camera displacements compatible with a dense set of correspondences, and the symmetric and the antisymmetric parts of an essential matrix are related.
- 21
- 21
- 21
- The Fisher-Rao Metric for Projective Transformations of the Line21
Experiments with the algorithm suggest that it can detect a projective transformation of the line even when the correspondences between the components of the measurements in the domain and the range of the projective Transformation are unknown.
- 20
- 20
- Visual music and musical vision20
A prototype cross-media retrieval system is built, using principles established in the neuroscientific study of synaesthesia, to bridge human hearing and vision from the viewpoint of database search for images or music.
- 19
- 19
- 19
- 18
- 18
- 18
- Facial expression transfer method based on frequency analysis17
A novel expression transfer method based on an analysis of the frequency of multi-expression facial images and the shape deformations between a neutral expression and non-neutral expressions to allow a more precise interpretation of facial expressions.
- Fisher-Rao Metric17
The information in a mass of data is summarised by fitting to the data a probability density function chosen from a parameterised family of pdfs, which provides a measure of the accuracy with which parameter values can be estimated from the data.
- Fisher-Rao metric17
The information in a mass of data is summarised by fitting to the data a probability density function chosen from a parameterised family of pdfs, which provides a measure of the accuracy with which parameter values can be estimated from the data.
- The Fisher-Rao metric17
The information in a mass of data is summarised by fitting to the data a probability density function chosen from a parameterised family of pdfs, which provides a measure of the accuracy with which parameter values can be estimated from the data.
- Gabor wavelets and AdaBoost in feature selection for face verification17
The experiments suggest that the feature selection algorithm for face verification selects the features corresponding to the unique characteristics rather than common characteristics, and a large example size statistically benefits AdaBoost feature selection.
- The critical line congruence for reconstruction from three images17
The relative positions of the lines can be reconstructed uniquely up to a collineation of space and it is shown that every Bordiga surface parameterises a critical set of lines.
- 16
- Probabilistic Tensor Analysis with Akaike and Bayesian Information Criteria16
The probabilistic tensor analysis (PTA) is developed, which selects suitable model for tensorformat data modeling based on Akaike information criterion and Bayesian information criterion, and empirical studies demonstrate that PTA associated with AIC and BIC selects correct number of models.
- End-to-end Animal Image Matting.15
This work proposes a novel Glance and Focus Matting network (GFM), which employs a shared encoder and two separate decoders to learn both tasks in a collaborative manner for end-to-end animal image matting and demonstrates that GFM outperforms state-of-the-art methods and effectively reduces the generalization error.
- MDL, Colllineations and the Fundamental Matrix15
The main result is that there is little reason for preferring the fundamental matrix model over the collineation model, even when the former is the ‘true’ model.
- 15
- 14
- A probabilistic definition of salient regions for image matching14
Experiments carried out using images in the Middlebury stereo database show that if the salience of a given image region is high, then there are relatively few background regions that have a better match to the given region than the true matching region.
- 13
- 13
- Apparent area of a rigid moving body13
The rate of change of apparent area of a rigid moving body is expressed in the form of an integral and an approximate expression is obtained for the time to contact with the body.
- Dual L1-Normalized Context Aware Tensor Power Iteration and Its Applications to Multi-object Tracking and Multi-graph Matching12
In this paper, multi-dimensional assignment is formulated as a rank-1 tensor approximation problem and a dual L1-normalized context/hyper-context aware tensor power iteration optimization method is proposed, applied to multi-object tracking and multi-graph matching.
- Automatic Visual Surveillance of Vehicles and People12
It is essential to devise autonomous surveillance systems, which can interpret the images and alert a human operator only when suspicious events occur, because operators rapidly become bored and lose concentration.
- Fusing $${\mathcal {R}}$$ R Features and Local Features with Context-Aware Kernels for Action Recognition11
A new global feature to capture the detailed geometrical distribution of interest points is proposed by using the 3D R transform which is defined as an extended 3D discrete Radon transform, followed by the application of a two-directional two-dimensional principal component analysis.
- An IR and visible image sequence automatic registration method based on optical flow11
This work uses motion features associated with single pixels in the two image sequences to carry out automatic registration of IR–visible camera registration using a normalized optical flow time sequence for each image pixel.
- 11
- 11
- 11
- Handcrafted vs. learned representations for human action recognition10
Both feature learning and feature engineering have their advantages in visual representation which, however, remains less explored in the video domain for action recognition.
- A Fisher-Rao Metric for Paracatadioptric Images of Lines10
The task of line detection is reduced to the task of finding isolated peaks in a Sobel image and an experimental comparison is made between these two algorithms and third algorithm taken from the literature and based on the Hough transform.
- 9
- An Improved Hierarchical Dirichlet Process-Hidden Markov Model and Its Application to Trajectory Modeling and Retrieval9
A hierarchical Bayesian model, an improved hierarchical Dirichlet process-hidden Markov model (iHDP-HMM), for visual document analysis that is capable of clustering visual documents and capturing the temporal correlations between the visual words within a visual document while identifying the number of document clusters and thenumber of visual topics adaptively.
- Application of the Fisher-Rao Metric to Ellipse Detection9
The experimental results suggest that ellipses can be reliably detected in a given low resolution image and that the number of false detections can be reduced using the multiresolution algorithm.
- 9
- 8
- Image recognition via two-dimensional random projection and nearest constrained subspace8
A new classifier called NCSC (Nearest Constrained Subspace Classifier) is designed and applied to image recognition with the 2D features and it outperforms its rivals including NN, NFL, NS and the orthonormal l2l2-norm classifier.
- 8
- Ambiguity in reconstruction from image correspondences8
This work uses a new proof that there are, in general, exactly ten different reconstructions compatible with five image correspondences, subject to a cubic polynomial constraint.
- Bearings-Only Tracking in the Plane7
A new finite dimensional recursive filter which is applied to bearings only tracking computes an approximation to the density for the target position conditional on the measurements, which is accurate provided only the most recent measurements are used.
- Stochastic properties of the cross ratio7
In computer vision the cross ratio is the basis of many recognition algorithms which do not require estimates of the position and orientation of the camera and is also employed in vehicle navigation, in which it is necessary to identify artificial landmarks or beacons placed in the environment.
- Application of the Fisher-Rao Metric to Structure Detection6
The main result is a derivation of an asymptotic approximation to the Fisher-Rao metric, under the assumption that the measurement noise is small, to within the level of accuracy allowed by the measurement errors.
- 5
- Generalized Visual Information Analysis Via Tensorial Algebra5
The generalized t-matrix algorithms, namely TSVD, THOSVD, TPCA, T2DPCA and TGCA, are applied to low-rank approximation, reconstruction and supervised classification of images and experiments show that the t-Matrix algorithms compare favorably with standard matrix algorithms.
- 5
- Iteration Functions re-visited5
Two classes of Iteration Functions (IFs) are derived from Joseph Traub using a different approach to the authors' and are demonstrably shown to be more informationally efficient than the first.
- Fisher information and model selection for projective transformations of the line5
The Rao measure is used to obtain a closed-form approximation to the probability of misclassifying a projective transformation of the line as an affine transformation.
- 5
- Intrinsic dimension estimation via nearest constrained subspace classifier4
The proposed classifier with an accurately estimated dimension parameter generally outperforms its competitors in terms of classification accuracy and a fast version of the classifier using a neighborhood representation to reduce its computational complexity is proposed.
- 4
- VEHICLE TRACKING WITH APPLICATIONS TO COLLISION ALERT4
An overview of a vision system that uses a camera mounted on a moving vehicle for locating, recognizing and tracking multiple vehicles and 3D model-based techniques are used for obtaining information regarding temporal and spatial vehicle movements in order to detect potential hazards in time.
- A Second Look at the Least-Squares Algorithm for Recovering Information from Optical Flow4
Experimental results show that the approximation to epsilon ( nu ) is astonishingly accurate and can be used to predict the performance of the algorithm in situations of practical interest.
- 3
- Self-supervised Exposure Trajectory Recovery for Dynamic Blur Estimation.3
Under mild constraints, the learned motion offsets can recover dense, (non-)linear exposure trajectories, which significantly reduce temporal disorder and ill-posed problems and further contribute to motion-aware image deblurring and warping-based video extraction from a single blurry image.
- 3
- 3
- Erratum to: A Robust Tracking System for Low Frame Rate Video3
This paper presents a meta-modelling system that automates the very labor-intensive and therefore time-heavy and expensive and therefore expensive process of manually calculating the value of a response in a discrete-time system.
- Stereo matching-based definition of saliency via sample-based Kullback–Leibler divergence estimation3
Experiments with the stereo images in the Middlebury database show that the proposed estimation of the Kullback–Leibler divergence is faster than the estimation based on parametric models for the pdfs, with no loss of accuracy.
- 3
- Shared feature extraction for semi-supervised image classification3
A manifold regularized MTL (MRMTL) algorithm is proposed to discover the latent shared subspace by treating the high-dimensional image space as a sub-manifold embedded in an ambient space and suggests that MRMTL can properly extract the common features for image representation and thus improve the generalization performance of the image classification models.
- Building 3D Models of Vehicles for Computer Vision3
A technique to build and manipulate three dimensional models of vehicles, to be used by a computer vision system, which will help find a suitable class of models for the vehicles and the trajectories, which are restricted to shape of cars and their trajectories.
- Reply to Pizlo, Rosenfeld, and Weiss3
It is possible that invariants are not appropriate for FCDPs, and PRW define a new set of attempts which captures two aspects of vision, transformations from the object plane to the image, the namely the information available to the observer and the fixed center directional perspective transformations decisions made by the observer on the basis of that infor(FCDPs).
- Applications of Algebraic Geometry to Computer Vision3
Algebraic geometry may in future provide methods for assessing the stability of algorithms against small perturbations in the data.
- Optical flow and the Taylor expansion3
An algorithm is proposed for computing the direction of translational motion of a moving rigid body from the resulting optical flow field that requires the coefficients of a Taylor expansion of the flow field to second order.
- A note on nearest neighbour error rates3
It is shown that if q>k and A"q","m = A"k","l then E"q,''m @? E"k,''l then If q> k and A"-q"," m = A"-k then E"-m @?" E" k","l".
- Ocean: Object-aware Anchor-free Tracking with Matching-relation Learning2
An object-aware anchor-free network for object tracking that upgrades the Siamese network by using automated search of the matching network based on binary channel operations to improve the stability of matching-relation learning for object tracking.
- 2
- Temporal Correlation Meets Embedding: Towards a 2nd Generation of JDE-based Real-Time Multi-Object Tracking2
A new learning approach using cross-correlation to capture temporal information of objects is proposed, and the feature extraction network is no longer trained solely on appearance features from each frame but learns richer motion features by utilizing feature heatmaps from consecutive frames, which addresses the challenge of inter-class feature similarity.
- Optical flow estimation using the Fisher–Rao metric2
It is shown that the Fisher–Rao algorithm performs well in comparison with two state of the art algorithms for estimating optical flow from AER measurements.
- The Fisher-Rao Metric in Computer Vision2
It is shown that the Fisher-Rao metric can be used to compare the amount of information in point features with the amount in edge element features, and is shown to be invariant under reparametrerisation.
- CDPM: Convolutional Deformable Part Models for Person Re-identification.2
Convolutional Deformable Part Models (CDPM) works by decoupling the complex part alignment procedure into two easier steps, and is able to accurately align flexible body parts without the need of any outside information.
- A Fisher–Rao Metric for Curves Using the Information in Edges2
Two curves which are close together in an image are indistinguishable given a measurement, in that there is no compelling reason to associate the measurement with one curve rather than the other, using the parametric version of the Fisher–Rao metric.
- THE FISHER–RAO METRIC FOR LINES IN A CONVEX IMAGE2
In this work, the Fisher–Rao metric on the parameter space for the set of lines in a two-dimensional convex image is approximated under the assumption that the errors in the measurements are small.
- 2
- 1
- 1
- Adversarial Attacks for Embodied Agents.1
This work generates spatiotemporal perturbations to form 3D adversarial examples, which exploit the interaction history in both the temporal and spatial dimensions and adversarially perturb the physical properties of the contextual objects that appeared in the most important scene views.
- 1
- GRMA: Generalized Range Move Algorithms for the Efficient Optimization of MRFs1
This paper proposes two generalized range move algorithms (GRMA) for the efficient optimization of MRFs, and extends the GRMAs to more general energy functions by restricting the chosen labels in each move so that the energy function is submodular on the chosen subset.
- Action classification using a discriminative non-parametric Hidden Markov Model1
A Bayesian framework based on Hierarchical Dirichlet Process is used to automatically infer the cardinality of hidden states and formulate a discriminative function based on distance between Gaussian distributions to improve classification performance.
- Approximation to the Fisher–Rao metric for the focus of expansion1
The Fisher-Rao metric for the focus of expansion is approximated, under the assumption that the focus is estimated from correspondences between two images taken by a translating camera, if the errors in the image correspondences are small.
- 1
- On the Use of Colour Filtering in an Integrated Real-Time People Tracking System1
It is shown that the quality of colour information is degraded by electrical interference and image compression to such an extent that it is no longer useful for local edge detection.
- Path prediction and classification based on non-linear filtering1
A path-based filter is described which is applied to the model-based tracking of cars and estimates the velocity, acceleration and steering angle of the car.
- 1
- Scalable Image Restoration and Enhancement towards Diverse Downstream Applications–
This paper proposes a scalable framework for image restoration and enhancement built upon the reinforcement learning (RL) paradigm, and confirms the model’s superior few-shot and zero-shot capabilities compared to existing methods, as well as its flexibility in addressing multiple competing objectives.
- Arbitrary Ratio Feature Compression via Next Token Prediction–
A novel and flexible Arbitrary Ratio Feature Compression (ARFC) framework, which supports any compression ratio with a single model, eliminating the need for multiple specialized models, and consistently outperforms existing approaches at various compression ratios.
- –
- –
- –
- –
- Connecting the Dots: Image Classification via Sparse Representation from a Constrained Subspace Perspective–
A classifier based on this model has a high classification accuracy similar to that of the conceptual NM (Nearest Manifold) classifier and the possibility of a proposed classifier called NCSC-CSR which has higher classification accuracy and robustness.
- –
- Message from the Guest Editors-In-Chief–
This special issue contains five papers which is the extended version of the conference paper presented at MIRU2009, the 12th Meeting on Image Recognition and Understanding held in Matsue, Japan, in July 2009.
- –
- –
- –
- A multivariate distribution for sub-images–
There is good agreement between the calculated and the experimental probabilities of obtaining a unique match that is also a correct match and if the probability of accepting a correct match is relatively large, then there is good agreement between the calculated and the experimental probabilities of obtaining a unique match.
- –
- –
- –
- Reconstruction from Minimal Data–
It is shown in Sect.2.1 that reconstruction from image correspondences is algebraic in that it can be expressed as the problem of finding the common zeros of a set of polynomial equations.
- Reconstruction from Image Velocities–
Reconstruction from image velocities is a limiting case of reconstruction from image correspondences, as the distances between corresponding points become small.
- Reconstruction from Image Correspondences–
Reconstruction is the task of finding the relative positions of the two viewpoints compatible with the image correspondences, and the relative position is determined up to a single unknown scale factor and a single 180° twist about the line joining the optical centres of theTwo cameras.
- –
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.