Robust detection of degenerate configurations for the fundamental matrix

New methods are reported for the detection of multiple solutions (degeneracy) when estimating the fundamental matrix, with specific emphasis on robustness in the presence of data contamination (outliers). The fundamental matrix can be used as a first step in the recovery of structure from motion. If the set of correspondences is degenerate then this structure cannot be accurately recovered and many solutions will explain the data equally well. It is essential that we are alerted to such eventualities. However, current feature matchers are very prone to mismatching, giving a high rate of contamination within the data. Such contamination can make a degenerate data set appear non degenerate, thus the need for robust methods becomes apparent. The paper presents such methods with a particular emphasis on providing a method that will work on real imagery and with an automated (non perfect) feature detector and matcher. It is demonstrated that proper modelling of degeneracy in the presence of outliers enables the detection of outliers which would otherwise be missed. Results using real image sequences are presented. All processing, point matching, degeneracy detection and outlier detection is automatic.<<ETX>>

• Publication year

  1995

• Venue

  Proceedings of IEEE International Conference on Computer Vision

• Publication date

  1995-06-20

• Fields of study

  Mathematics, Computer Science

• Identifiers
  DOI 10.1109/ICCV.1995.466820
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• Degenerate correspondence sets are treated as cases where the fundamental matrix has multiple valid explanations.

  Confidence 0.87

  박진우 (dztg5apj7m) extraction뀨 (7c402c1b98) reviewAK (4715169a40) reviewB (s683577b42) review
• Point matching, degeneracy detection, and outlier detection are carried out automatically on real image sequences.

  Confidence 0.90

  박진우 (dztg5apj7m) extraction뀨 (7c402c1b98) reviewAK (4715169a40) reviewB (s683577b42) review
• Modeling degeneracy explicitly in contaminated correspondence data can reveal outliers that would otherwise be missed.

  Confidence 0.92

  박진우 (dztg5apj7m) extraction뀨 (7c402c1b98) reviewAK (4715169a40) reviewB (s683577b42) review
• Robust degeneracy detection for the fundamental matrix remains effective even when the correspondence set is contaminated by mismatches.

  Confidence 0.95

  박진우 (dztg5apj7m) extraction뀨 (7c402c1b98) reviewAK (4715169a40) reviewB (s683577b42) review

• automated feature matching

  method, pipeline

  An image-processing pipeline that detects and matches features without manual intervention.

  Aliases: automated point matching

  박진우 (dztg5apj7m) extraction뀨 (7c402c1b98) reviewAK (4715169a40) reviewB (s683577b42) review
• degenerate configuration

  phenomenon, configuration

  A correspondence arrangement in which the fundamental matrix is not uniquely determined and multiple solutions can fit the data.

  Aliases: degeneracy

  박진우 (dztg5apj7m) extraction뀨 (7c402c1b98) reviewAK (4715169a40) reviewB (s683577b42) review
• fundamental matrix

  method, mathematical object

  A two-view epipolar geometry object that encodes the geometric relationship between matched image points.

  박진우 (dztg5apj7m) extraction뀨 (7c402c1b98) reviewAK (4715169a40) reviewB (s683577b42) review
• outlier contamination

  error source, data condition

  Incorrect point correspondences mixed into the set of matches used to estimate the model.

  Aliases: outliers, data contamination

  박진우 (dztg5apj7m) extraction뀨 (7c402c1b98) reviewAK (4715169a40) reviewB (s683577b42) review
• real image sequences

  data, evaluation input

  Ordered sets of images captured from actual scenes and used as the evaluation input.

  Aliases: real imagery

  박진우 (dztg5apj7m) extraction뀨 (7c402c1b98) reviewAK (4715169a40) reviewB (s683577b42) review
• robust degeneracy detection

  method

  A detection approach that identifies degenerate fundamental-matrix cases while accounting for contaminated correspondences.

  Aliases: robust methods

  박진우 (dztg5apj7m) extraction뀨 (7c402c1b98) reviewAK (4715169a40) reviewB (s683577b42) review
• structure from motion

  task

  A computer vision task that reconstructs scene structure and camera motion from image correspondences.

  Aliases: SfM

  박진우 (dztg5apj7m) extraction뀨 (7c402c1b98) reviewAK (4715169a40) reviewB (s683577b42) review

• Motion Clustering using the Trilinear Constraint over Three Views

  1995cited by this paper
• Robust Vision

  1994cited by this paper
• Outlier detection and motion segmentation

  1993cited by this paper
• What can be seen in three dimensions with an uncalibrated stereo rig

  1992cited by this paper
• Estimation of Relative Camera Positions for Uncalibrated Cameras

  1992cited by this paper
• Geometric invariance in computer vision

  1992cited by this paper
• Structure-from-motion under orthographic projection

  1990cited by this paper
• Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography

  1981cited by this paper

• Dense Feature Matching Based on Homographic Decomposition

  2022cites this paper
• USAC 2021 with spatial coherence modelling robust to degenerate data

  2021cites this paper
• Improved Universal Framework for Random Sample Consensus Maksym Ivashechkin

  2019cites this paper
• A Descriptor-less Well-Distributed Feature Matching Method Using Geometrical Constraints and Template Matching

  2018cites this paper
• Monocular Visual Odometry for an Unmanned Sea-Surface Vehicle

  2017cites this paper
• Multiple structure recovery with T-linkage

  2017cites this paper
• Direct Pose Estimation and Refinement

  2016cites this paper
• 3 D Modeling : Comparison of some Innovative Techniques of Photogrammetry

  2016cites this paper
• Video Stabilization Using Point Feature Matching

  2015cites this paper
• End of Project Documentation

  2015cites this paper
• Design on Diffuse Optical Instrument Based on Computer Simulation Technology

  2014cites this paper
• Robust Sampling Consensus

  2013cites this paper
• Applying distribution of feature points to detect multiple plane

  2012cites this paper
• Variationelle 3D-Rekonstruktion aus Stereobildpaaren und Stereobildfolgen

  2012cites this paper
• Algebraic error analysis of collinear feature points for camera parameter estimation

  2011cites this paper
• Unscented feature tracking

  2011cites this paper
• A Planes Detection Algorithm Based on Feature Distribution

  2011cites this paper
• Estimating the fundamental matrix based on least absolute deviation

  2011cites this paper
• Metric measurement from street view sequences with simple operator assistance and phase correlation based frame selection

  2011cites this paper
• An M-estimator for high breakdown robust estimation in computer vision

  2011cites this paper
• Detection of moving objects from a moving platform in urban scenes

  2010cites this paper
• Specific Displacements for Self‐calibration

  2010cites this paper
• Recovering articulated non-rigid shapes, motions and kinematic chains from video

  2009cites this paper
• Exploiting uncertainty in random sample consensus

  2009cites this paper
• Dealing with degeneracy in essential matrix estimation

  2008influential citation
• Estimation and Detection with Applications to Navigation

  2008cites this paper
• Detecting spurious features using parity space

  2008cites this paper
• Robust Feature Tracking

  2007cites this paper
• Improving the Efficiency of a Least Median of Squares Schema for the Estimation of the Fundamental Matrix

  2006cites this paper
• Probabilistic Image-Based Rendering with Gaussian Mixture Model

  2006cites this paper
• Recovering Articulated Non-rigid Shapes, Motions and Kinematic Chains from Video

  2006cites this paper
• Study on Epipolar Geometry Restoration of Parallel Binocular Vision

  2006influential citation
• Global-to-Local Histogram Match Culling for Epipolar Geometry Estimation

  2006cites this paper
• Robust structure and motion from outlines of smooth curved surfaces

  2006cites this paper
• Multiscale segmentation by combining motion and intensity cues

  2005cites this paper
• A computational framework for stereo imaging

  2005cites this paper
• Efficient Algorithms for Determining 3-D Bi-Plane Imaging Geometry

  2005cites this paper
• Reconstrução Tridimensional de Ambientes Reais Usando Dados Laser e de Intensidade

  2004cites this paper
• Accelerated multiple-view reconstruction

  2004cites this paper
• Visual Odometry Using Commodity Optical Flow

  2004cites this paper
• Fundamental problems in computational video

  2004influential citation
• Structure and Motion from Images of Smooth Textureless Objects

  2004cites this paper
• Object Level Grouping for Video Shots

  2004cites this paper
• Image Mosaicing and Super-resolution

  2004cites this paper
• A linear iterative least-squares method for estimating the fundamental matrix

  2003cites this paper
• Markerless Motion Capture for Humans in Video

  2003cites this paper
• Geometric Representation and Grouping of Image Curves

  2003cites this paper
• Robust Estimation in High Noise and Highly Dimensional Data Sets with Applications to Machine Vision

  2002cites this paper
• Bayesian Model Estimation and Selection for Epipolar Geometry and Generic Manifold Fitting

  2002influential citation
• Maximum Likelihood 3D Reconstruction from One or More Images under Geometric Constraints

  2002cites this paper
• Vision Algorithms: Theory and Practice

  2002cites this paper
• NAPSAC: high noise, high dimensional model parameterisation - it's in the bag

  2002cites this paper
• Estimating the Fundamental Matrix by Transforming Image Points in Projective Space

  2001cites this paper
• Aerial video surveillance and exploitation

  2001cites this paper
• Indoor robot motion based on monocular images

  2001cites this paper
• Uncertainty analysis of 3D reconstruction from uncalibrated views

  2000cites this paper
• On the Estimation of the Fundamental Matrix: A Convex Approach to Constrained Least-Squares

  2000cites this paper
• MLESAC: A New Robust Estimator with Application to Estimating Image Geometry

  2000cites this paper
• A robust algorithm to estimate the fundamental matrix

  2000cites this paper
• Using Specific Displacements to Analyze Motion without Calibration

  1999influential citation
• Feature Based Methods for Structure and Motion Estimation

  1999cites this paper
• Model Selection for Two View Geometry: A Review

  1999cites this paper
• Development of Binocular Stereopsis for Vehicle Lateral Control , Longitudinal Control and Obstacle Detection UCB-ITS-PRR-99-37

  1999cites this paper
• Geometrical constraints of the synthetic method of estimating fundamental matrix and its analysing

  1999cites this paper
• Shape, Contour and Grouping in Computer Vision

  1999cites this paper
• Geometric Motion Segmentation and ModelSelectionBy

  1998cites this paper
• Image-based Visualization of Real 3d Scenes Iv Mathematical Notation

  1998cites this paper
• Video indexing based on mosaic representations

  1998cites this paper
• Holistic Matching

  1998cites this paper
• Robust Detection of Degenerate Configurations while Estimating the Fundamental Matrix

  1998cites this paper
• Determining the Epipolar Geometry and its Uncertainty: A Review

  1998cites this paper
• Concerning Bayesian Motion Segmentation, Model, Averaging, Matching and the Trifocal Tensor

  1998cites this paper
• Making good features track better

  1998cites this paper
• Robust feature tracking in underwater video sequences

  1998cites this paper
• Reconstruction of extended environments from image sequences

  1997cites this paper
• Recovering Perspective Pose with a Dual Step EM Algorithm

  1997cites this paper
• Development of Binocular Stereopsis for Vehicle Lateral Control, Longitudinal Control and Obstacle Detection

  1997cites this paper
• Robust parameterization and computation of the trifocal tensor

  1997cites this paper
• Vision for longitudinal vehicle control

  1997cites this paper
• Development of Binocular Stereopsis for Vehicle Lateral Control, Longitudinal Control and Obstable Detection

  1997cites this paper
• Reconstruction of extended environmentsfrom image sequences

  1997cites this paper
• An Image-Based Approach to Three-Dimensional Computer Graphics

  1997cites this paper
• Six-point synthetic method to estimate fundamental matrix

  1997cites this paper
• Development of Binocular Stereopsis for Vehicle Lateral Control, Longitudinal Control and Obstable Detection

  1997cites this paper
• The Development and Comparison of Robust Methods for Estimating the Fundamental Matrix

  1997cites this paper
• Performance characterization of fundamental matrix estimation under image degradation

  1997cites this paper
• Title Development Of Binocular Stereopsis For Vehicle Lateral Control , Longitudinal Control And Obstacle Detection Permalink

  1997cites this paper
• Image Analysis and Computer Vision: 1995

  1996cites this paper
• Parallax Geometry of Pairs of Points for 3D Scene Analysis

  1996cites this paper
• MUSE: robust surface fitting using unbiased scale estimates

  1996cites this paper
• Using Singular Displacements for Uncalibrated Monocular Visual Systems

  1996cites this paper
• Epipolar Geometry and Linear Subspace Methods: A New Approach to Weak Calibration

  1996cites this paper
• A unified approach to moving object detection in 2D and 3D scenes

  1996cites this paper
• On Uncalibrated Motion-Based Inspection for Conveyor-Belt Systems

  1996cites this paper
• 4.1 the Detailed Scene-based Representation

  year unknowncites this paper
• Robust estimation of the fundamental matrix and stereocorrespondencesNuno

  year unknowncites this paper