## Camera pose revisited: New linear algorithms (2002)

Venue: | In 14„eme Congr„es Francophone de Reconnaissance des Formes et Intelligence Artificielle. Paper in French |

Citations: | 13 - 0 self |

### BibTeX

@ARTICLE{Ameller02camerapose,

author = {Marc-andré Ameller and Bill Triggs and Long Quan},

title = {Camera pose revisited: New linear algorithms},

journal = {In 14„eme Congr„es Francophone de Reconnaissance des Formes et Intelligence Artificielle. Paper in French},

year = {2002},

pages = {2002}

}

### Years of Citing Articles

### OpenURL

### Abstract

Abstract. Camera pose estimation is the problem of determining the position and orientation of an internally calibrated camera from known 3D reference points and their images. We briefly survey several existing methods for pose estimation, then introduce four new linear algorithms. The first three give a unique linear solution from four points by SVD null space estimation. They are based on resultant matrices: the 24 × 24 method is the raw resultant matrix, and the 12 × 12 and 9 × 9 methods are compressed versions of this obtained by Gaussian elimination with pivoting on constant entries. The final method returns the four intrinsic solutions to the pose from 3 points problem. It is based on eigendecomposition of a 5 × 5 matrix. One advantage of all these methods is that they are simple to implement. In particular, the matrix entries are simple functions of the input data. Numerical experiments are given comparing the performance of the new algorithms with several existing algebraic and linear methods.

### Citations

2740 |
Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography
- Fischler, Bolles
- 1981
(Show Context)
Citation Context ...3 points, the problem generically has four possible solutions. Many closed form methods are known for finding these. The earliest was perhaps by Lagrange in 1795. See [17, 6–8, 15]. Fischler & Bolles =-=[6]-=- gave a method that has become popular in computer vision when they introduced their RANSAC paradigm for detecting outliers in the data. Haralick et al [8] review many old and new variants of the basi... |

481 |
Perceptual Organization and Visual Recognition
- Lowe
- 1985
(Show Context)
Citation Context ... and elimination. For handling redundant data, iterative methods have also been developed in [12, 18, 3]. Methods for camera pose from line segments instead of point features have also been developed =-=[9, 4, 2, 13, 11]-=-. 3 point methods intrinsically give multiple solutions. If a unique solution is required, additional information must be given. A fourth point generically suffices, but even with an infinite number o... |

314 | Fitting parameterized threedimensional models to images
- Lowe
- 1991
(Show Context)
Citation Context ...ject 21914 Cumuli.and carefully examine their numerical stabilities under different orders of substitution and elimination. For handling redundant data, iterative methods have also been developed in =-=[12, 18, 3]-=-. Methods for camera pose from line segments instead of point features have also been developed [9, 4, 2, 13, 11]. 3 point methods intrinsically give multiple solutions. If a unique solution is requir... |

246 |
The representation, recognition, and locating of 3D objects
- Faugeras, Hebert
- 1986
(Show Context)
Citation Context ...ir known world-frame coordinates is then estimated. The centres of gravity of the two point clouds are aligned by translation, then the aligning rotation(s) are found by the quaternion or SVD methods =-=[10, 5]-=-. 3 Linear algorithms for pose from 4 points 3.1 Linear methods for polynomial system solving Linear algebra is a useful tool for manipulating polynomials. A polynomial Pi(x) = ∑ α cα,i xα in variable... |

217 | Model-based object pose in 25 lines of code
- DeMenthon, Davis
- 1992
(Show Context)
Citation Context ...ject 21914 Cumuli.and carefully examine their numerical stabilities under different orders of substitution and elimination. For handling redundant data, iterative methods have also been developed in =-=[12, 18, 3]-=-. Methods for camera pose from line segments instead of point features have also been developed [9, 4, 2, 13, 11]. 3 point methods intrinsically give multiple solutions. If a unique solution is requir... |

210 |
Review and analysis of solutions of the three point perspective pose estimation problem
- Haralick, Lee, et al.
- 1994
(Show Context)
Citation Context ...n 1795. See [17, 6–8, 15]. Fischler & Bolles [6] gave a method that has become popular in computer vision when they introduced their RANSAC paradigm for detecting outliers in the data. Haralick et al =-=[8]-=- review many old and new variants of the basic 3-point method Submitted to ECCV’00. This work was supported in part by Esprit LTR project 21914 Cumuli.and carefully examine their numerical stabilitie... |

118 |
Determination f the attitude of 3D objects from a single perspective view
- Dhome, Richetin, et al.
- 1989
(Show Context)
Citation Context ... and elimination. For handling redundant data, iterative methods have also been developed in [12, 18, 3]. Methods for camera pose from line segments instead of point features have also been developed =-=[9, 4, 2, 13, 11]-=-. 3 point methods intrinsically give multiple solutions. If a unique solution is required, additional information must be given. A fourth point generically suffices, but even with an infinite number o... |

115 |
Closed form solutions of absolute orientation using unit quaternions
- HORN
- 1987
(Show Context)
Citation Context ...ir known world-frame coordinates is then estimated. The centres of gravity of the two point clouds are aligned by translation, then the aligning rotation(s) are found by the quaternion or SVD methods =-=[10, 5]-=-. 3 Linear algorithms for pose from 4 points 3.1 Linear methods for polynomial system solving Linear algebra is a useful tool for manipulating polynomials. A polynomial Pi(x) = ∑ α cα,i xα in variable... |

96 |
An Analytic Solution for the Perspective 4-Point Problem
- Horaud, Conio, et al.
- 1989
(Show Context)
Citation Context ...on and elimination. For handling redundant data, iterative methods have also been developed in [12,18,3]. Methods for camera pose from line segments instead of point features have also been developed =-=[9,4, 2,13,11]-=-. 3 point methods intrinsically give multiple solutions. If a unique solution is required, additional information must be given. A fourth point generically suffices, but even with an infinite number o... |

89 | Linear N–point camera pose determination
- Quan, Lan
- 1999
(Show Context)
Citation Context ...w its effect on the pose methods tested below. Motivated by a lack of methods that directly provide the unique pose solution in the redundant data case, a family of linear algorithms was presented in =-=[14]-=-. Unfortunately these methods are similar to algebraic ones in the sense that the matrix entries are complicated coefficients extracted from fourth degree polynomials. This makes implementation somewh... |

65 |
Determination of camera location from 2-D to 3-D line and point correspondences
- Liu, Huang, et al.
- 1990
(Show Context)
Citation Context ... and elimination. For handling redundant data, iterative methods have also been developed in [12, 18, 3]. Methods for camera pose from line segments instead of point features have also been developed =-=[9, 4, 2, 13, 11]-=-. 3 point methods intrinsically give multiple solutions. If a unique solution is required, additional information must be given. A fourth point generically suffices, but even with an infinite number o... |

51 | Multivariate polynomials, duality and structured Matrices, in "J
- MOURRAIN, PAN
(Show Context)
Citation Context ...which the root itself is easily extracted. The required ideal elements can be guessed by hand (as here), or generated more systematically using several classical and modern resultant-building methods =-=[1]-=-. Although they give generically sufficient monomial sets, these constructions are often far from minimal, and also may not suffice for certain special values of the input coefficients. 43.2 Linear a... |

32 |
Pose determination from line-to-plane correspondences: Existence condition and closed-form solutions
- Chen
- 1991
(Show Context)
Citation Context |

29 | Reliability analysis of parameter estimation in linear models with application to mensuration problems in computer vision - Förstner - 1987 |

7 |
Minimum Solutions for Orientation
- Wrobel
- 1992
(Show Context)
Citation Context ...nt generically suffices, but even with an infinite number of points there are certain degenerate cases for which no unique solution is possible. These critical configurations are known precisely. See =-=[16, 17]-=- for details, but briefly, all 3D points including the camera centre must lie on a special twisted cubic space curve (the horopter) that wraps around a circular cylinder (the dangerous cylinder). Nota... |

5 |
Space Resection: Failure Cases
- Thompson
- 1966
(Show Context)
Citation Context ...nt generically suffices, but even with an infinite number of points there are certain degenerate cases for which no unique solution is possible. These critical configurations are known precisely. See =-=[16, 17]-=- for details, but briefly, all 3D points including the camera centre must lie on a special twisted cubic space curve (the horopter) that wraps around a circular cylinder (the dangerous cylinder). Nota... |

5 |
A general phogrammetric solution for the determining object position and orientation
- Yuan
- 1989
(Show Context)
Citation Context ...ject 21914 Cumuli.and carefully examine their numerical stabilities under different orders of substitution and elimination. For handling redundant data, iterative methods have also been developed in =-=[12, 18, 3]-=-. Methods for camera pose from line segments instead of point features have also been developed [9, 4, 2, 13, 11]. 3 point methods intrinsically give multiple solutions. If a unique solution is requir... |