## A Variational Approach to Problems in Calibration of Multiple Cameras (2004)

### Cached

### Download Links

- [www.ece.wisc.edu]
- [www.ece.gatech.edu]
- [www.ece.gatech.edu]
- [www.vision.cs.ucla.edu]
- [vision.ucla.edu]
- DBLP

### Other Repositories/Bibliography

Venue: | Proc. IEEE Conf. Computer Vision and Pattern Recognition |

Citations: | 5 - 3 self |

### BibTeX

@INPROCEEDINGS{Unal04avariational,

author = {Gozde Unal and Anthony Yezzi},

title = {A Variational Approach to Problems in Calibration of Multiple Cameras},

booktitle = {Proc. IEEE Conf. Computer Vision and Pattern Recognition},

year = {2004},

pages = {172--178}

}

### OpenURL

### Abstract

This paper addresses the problem of calibrating camera parameters using variational methods. One problem addressed in this paper is the severe lens distortion in wide angle/inexpensive camera lenses. The camera distortion effects lead to inaccurate 3D reconstructions and geometrical measurements if not accounted for. A second problem is the color calibration problem caused by variations in camera responses which results in different color measurements and affects the algorithms that depend on these measurements. We present multi-view stereo techniques based on variational ideas to address these calibration problems. To reduce computational complexity of such algorithms, we utilize a prior knowledge on the calibration object which is used in the process, and evolve the pose, orientation, and scale parameters of such a 3D model object. We derive the evolution equations for the distortion coefficients, the color calibration parameters of the cameras, and present experimental results which demonstrate their potential use. 1.

### Citations

3152 |
Multiple View Geometry in Computer Vision, 2nd ed
- Hartley, Zisserman
- 2004
(Show Context)
Citation Context ... of a known calibration pattern and intrinsic camera parameters as well as camera pose and orientation (extrinsic camera parameters) are estimated by a minimization of an overall cost functional [5], =-=[6]-=-, [7], [8], [9], [10], [11], [12], [13]. Many calibration techniques use both nonlinear minimization and closed-form solutions as in [14]. In this paper, we develop a coordinated refinement technique ... |

1030 |
A versatile camera calibration technique for high-accuracy 3d machine vision metrology using off-the-shelf tv cameras and lenses
- Tsai
- 1987
(Show Context)
Citation Context ...mages of a known calibration pattern and intrinsic camera parameters as well as camera pose and orientation (extrinsic camera parameters) are estimated by a minimization of an overall cost functional =-=[5]-=-, [6], [7], [8], [9], [10], [11], [12], [13]. Many calibration techniques use both nonlinear minimization and closed-form solutions as in [14]. In this paper, we develop a coordinated refinement techn... |

873 | Fronts propagating with curvature dependent speed: Algorithms based on Hamilton-Jacobi formulations
- Osher, Sethian
- 1988
(Show Context)
Citation Context ...ure is then obtained by finding the partial differential equation (PDE) that is the gradient descent flow of the energy E. A popular class of numerical techniques, known ð1Þ ð2Þ as Level Sets Methods =-=[30]-=-, is utilized to evolve the surface S via the evolution of a 3D function :IR 3 !IR. Nevertheless, an update of the level set function is required after each iteration of the associated PDE, and even w... |

829 | A Flexible New Technique for Camera Calibration
- Zhang
(Show Context)
Citation Context ...and intrinsic camera parameters are first obtained by a pinhole camera model neglecting lens distortion. Then, distortion calibration is performed while holding the other parameters fixed [17], [18], =-=[19]-=-. This is possible because the mapping from 3D world coordinates to the 2D image plane can be decomposed into a perspective projection and a mapping that models the deviations from the ideal pinhole c... |

638 | A Mathematical Introduction to Robotic Manipulation
- Murray, Li, et al.
- 1994
(Show Context)
Citation Context ...5; @ Yo Xo 0 |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} RsNo + ^Xo Rs ^Xo; RsNo ; where we utilize exponential coordinates (see =-=[33]-=- for details on this representation) for the global rotation parameters of the surface. We note that a matrix in an inner-product expression, when operated on a vector, will incorporate each of its ro... |

468 |
Multiple View Geometry
- Hartley, Zisserman
- 2000
(Show Context)
Citation Context .../Contributions 1.1.1 Lens Distortion The linear model in an ideal pinhole camera model leads to imaging of world lines as lines on the image plane, and simplifies many computations and considerations =-=[3]-=-. However, for most real cameras with wide-angle or inexpensive lenses this assumption does not hold, and nonlinearities introduced by a well-known phenemonen referred to as a lens distortion should b... |

366 | Camera selfcalibration: Theory and experiments
- Faugeras, Luong, et al.
- 1992
(Show Context)
Citation Context ...ibration pattern and intrinsic camera parameters as well as camera pose and orientation (extrinsic camera parameters) are estimated by a minimization of an overall cost functional [5], [6], [7], [8], =-=[9]-=-, [10], [11], [12], [13]. Many calibration techniques use both nonlinear minimization and closed-form solutions as in [14]. In this paper, we develop a coordinated refinement technique for the extrins... |

327 | A fast level set method for propagating interfaces
- Adalsteinsson, Sethian
- 1995
(Show Context)
Citation Context ...via the evolution of a 3D function :IR 3 !IR. Nevertheless, an update of the level set function is required after each iteration of the associated PDE, and even with more efficient narrowband schemes =-=[31]-=-, there is a considerable amount of computation involved. For our intended applications, in which there is a calibration object whose shape can be roughly known a priori, rather than deforming the sur... |

283 |
Gool, ”Self-Calibration and Metric Reconstruction in spite of Varying and
- Pollefeys, Koch, et al.
- 1997
(Show Context)
Citation Context ...and intrinsic camera parameters as well as camera pose and orientation (extrinsic camera parameters) are estimated by a minimization of an overall cost functional [5], [6], [7], [8], [9], [10], [11], =-=[12]-=-, [13]. Many calibration techniques use both nonlinear minimization and closed-form solutions as in [14]. In this paper, we develop a coordinated refinement technique for the extrinsic and intrinsic c... |

202 |
Camera calibration with distortion models and accuracy evaluation
- Weng, Cohen, et al.
- 1992
(Show Context)
Citation Context ...insic and intrinsic camera parameters are first obtained by a pinhole camera model neglecting lens distortion. Then, distortion calibration is performed while holding the other parameters fixed [17], =-=[18]-=-, [19]. This is possible because the mapping from 3D world coordinates to the 2D image plane can be decomposed into a perspective projection and a mapping that models the deviations from the ideal pin... |

194 |
A Four-step Camera Calibration Procedure with Implicit Image Correction
- Heikkilä, Silvén
- 1997
(Show Context)
Citation Context ... estimated by a minimization of an overall cost functional [5], [6], [7], [8], [9], [10], [11], [12], [13]. Many calibration techniques use both nonlinear minimization and closed-form solutions as in =-=[14]-=-. In this paper, we develop a coordinated refinement technique for the extrinsic and intrinsic camera parameters: lens distortion, focal lengths, skew, and estimation of camera color calibration param... |

194 | Variational principles, surface evolution, PDE’s, level set methods and the stereo problem
- Faugeras, Keriven
- 1998
(Show Context)
Citation Context ... a joint region-based image segmentation and simultaneous 3D stereo reconstruction technique. Note that early variational approaches to the 3D reconstruction problem were pioneered by Faugeras et.al. =-=[23]-=- who however also relied on local feature matching. In our distortion calibration algorithm, we use a white bar object, made from a foam core as shown in Fig.1 on the left. We capture its views before... |

110 | True multi-image alignment and its application to mosaicing and lens distortion correction
- Sawhney, Kumar
- 1999
(Show Context)
Citation Context ... are arranged into a tensor, which is computed with estimated distortion parameters at each step to minimize a reprojection error in an iterative manner. In another group of methods as in [25], [26], =-=[27]-=-, a direct solution strategy is employed to find camera calibration parameters by incorporating lens distortion as well. Our contribution is a new distortion calibration technique that does not rely o... |

89 | Non-metric calibration of wide-angle lenses and polycameras
- Swaminathan, Nayar
- 2000
(Show Context)
Citation Context ...uch each detected segment is distorted. The curved lines in the image that do not really correspond to 3D line segments will constitute outliers in this optimization procedure [17], [20], [21], [22], =-=[23]-=-. Other techniques such as [24] rely on point correspondences. Given a set of 3D points, the associated epipolar and trilinear constraints are arranged into a tensor, which is computed with estimated ... |

88 |
Introduction to shape optimization: shape sensitivity analysisSpringer
- Sokolowski, Zolesio
- 1992
(Show Context)
Citation Context ...I k i Þ @ðgsXoÞ ; rf @ k S dAo; ð3Þ where F k i ðfkðXÞ Ik 2 i ð giðXÞÞÞ ðbk Ik i Þ2ŠX iðXÞ ðXiÞ is the Mumford-Shah term from (2) (also in [2]). The derivation follows from shape optimization tools =-=[32]-=- that provide the shape derivatives in curve and surface evolution framework. N denotes the surface normal vector. Note that the visibility function X iðgsðXoÞÞ, included in the data term F k i ð Þ, i... |

86 | Simultaneous linear estimation of multiple view geometry and lens distortion
- Fitzgibbon
(Show Context)
Citation Context ... constraints are arranged into a tensor, which is computed with estimated distortion parameters at each step to minimize a reprojection error in an iterative manner. In another group of methods as in =-=[25]-=-, [26], [27], a direct solution strategy is employed to find camera calibration parameters by incorporating lens distortion as well. Our contribution is a new distortion calibration technique that doe... |

81 | A convenient multi-camera self-calibration for virtual environments. PRESENCE: Teleoperators and Virtual Environments
- Svoboda, Martinec, et al.
- 2005
(Show Context)
Citation Context ...trinsic camera parameters as well as camera pose and orientation (extrinsic camera parameters) are estimated by a minimization of an overall cost functional [5], [6], [7], [8], [9], [10], [11], [12], =-=[13]-=-. Many calibration techniques use both nonlinear minimization and closed-form solutions as in [14]. In this paper, we develop a coordinated refinement technique for the extrinsic and intrinsic camera ... |

68 | Lens distortion calibration using point correspondences
- Stein
- 1997
(Show Context)
Citation Context ...storted. The curved lines in the image that do not really correspond to 3D line segments will constitute outliers in this optimization procedure [17], [20], [21], [22], [23]. Other techniques such as =-=[24]-=- rely on point correspondences. Given a set of 3D points, the associated epipolar and trilinear constraints are arranged into a tensor, which is computed with estimated distortion parameters at each s... |

65 | Stereoscopic segmentation
- Yezzi, Soatto
- 2001
(Show Context)
Citation Context ...ly variational approaches to the 3D reconstruction problem were pioneered by Faugeras and Keriven [1] who also relied on local feature matching. A more recent variational approach by Yezzi and Soatto =-=[2]-=-, [3] proposed a joint region-based image segmentation and simultaneous 3D stereo reconstruction technique. This paper addresses camera calibration techniques built on this latter stereo reconstructio... |

61 | Straight lines have to be straight
- Devernay, Faugeras
- 2006
(Show Context)
Citation Context ...r extrinsic and intrinsic camera parameters are first obtained by a pinhole camera model neglecting lens distortion. Then, distortion calibration is performed while holding the other parameters fixed =-=[17]-=-, [18], [19]. This is possible because the mapping from 3D world coordinates to the 2D image plane can be decomposed into a perspective projection and a mapping that models the deviations from the ide... |

53 |
Decentering distortion of lenses
- Brown
- 1966
(Show Context)
Citation Context ...a measure of how much each detected segment is distorted. The curved lines in the image that do not really correspond to 3D line segments will constitute outliers in this optimization procedure [17], =-=[20]-=-, [21], [22], [23]. Other techniques such as [24] rely on point correspondences. Given a set of 3D points, the associated epipolar and trilinear constraints are arranged into a tensor, which is comput... |

51 |
Self-calibration of the intrinsic parameters of cameras for active vision systems” CVPR
- Du, Brady
- 1993
(Show Context)
Citation Context ...raints are arranged into a tensor, which is computed with estimated distortion parameters at each step to minimize a reprojection error in an iterative manner. In another group of methods as in [25], =-=[26]-=-, [27], a direct solution strategy is employed to find camera calibration parameters by incorporating lens distortion as well. Our contribution is a new distortion calibration technique that does not ... |

32 | The coliseum immersive teleconferencing system
- Baker, Tanguay, et al.
(Show Context)
Citation Context ...Fig. 1 on the left. We capture its views before a dark background with the multiview stereo rig system, a desktop multicamera system designed for remote multimedia collaboration, developed by HP Labs =-=[28]-=-. The images of the calibration object captured from three of the five cameras in the rig are given in Fig. 1. Many desktop multicamera systems use wideangle and inexpensive cameras that produce sever... |

31 | pose and calibration from 4 or 5 known 3d points - Triggs, “Camera - 1999 |

21 |
On Calibrating Computer Controlled Cameras for Perceiving 3D
- Sobel
- 1974
(Show Context)
Citation Context ...constructing reliable digital representations of 3D scenes, accurate camera calibrations are essential. There has been a great deal of research on the camera calibration problem as early as the 1970s =-=[4]-=-. In most of the previous techniques, some sets of features were extracted from images of a known calibration pattern and intrinsic camera parameters as well as camera pose and orientation (extrinsic ... |

18 | Structure from motion for scenes without features
- Yezzi, Soatto
- 2003
(Show Context)
Citation Context ...riational approaches to the 3D reconstruction problem were pioneered by Faugeras and Keriven [1] who also relied on local feature matching. A more recent variational approach by Yezzi and Soatto [2], =-=[3]-=- proposed a joint region-based image segmentation and simultaneous 3D stereo reconstruction technique. This paper addresses camera calibration techniques built on this latter stereo reconstruction fra... |

18 | P.Sturm. “Methods and Geometry for Plane-Based SelfCalibration
- Gurdjos
(Show Context)
Citation Context ...n calibration pattern and intrinsic camera parameters as well as camera pose and orientation (extrinsic camera parameters) are estimated by a minimization of an overall cost functional [5], [6], [7], =-=[8]-=-, [9], [10], [11], [12], [13]. Many calibration techniques use both nonlinear minimization and closed-form solutions as in [14]. In this paper, we develop a coordinated refinement technique for the ex... |

16 | Nonmetric lens distortion calibration: Closed-form solutions, robust estimation and model selection
- El-Melegy, Farag
- 2003
(Show Context)
Citation Context ...ure of how much each detected segment is distorted. The curved lines in the image that do not really correspond to 3D line segments will constitute outliers in this optimization procedure [17], [20], =-=[21]-=-, [22], [23]. Other techniques such as [24] rely on point correspondences. Given a set of 3D points, the associated epipolar and trilinear constraints are arranged into a tensor, which is computed wit... |

9 | Semiautomatic Methods for Recovering Radial Distortion Parameters from a Single Image
- Kang
- 1997
(Show Context)
Citation Context ... how much each detected segment is distorted. The curved lines in the image that do not really correspond to 3D line segments will constitute outliers in this optimization procedure [17], [20], [21], =-=[22]-=-, [23]. Other techniques such as [24] rely on point correspondences. Given a set of 3D points, the associated epipolar and trilinear constraints are arranged into a tensor, which is computed with esti... |

7 |
Fast and Accurate Self Calibration
- Oliensis
- 1999
(Show Context)
Citation Context ...ttern and intrinsic camera parameters as well as camera pose and orientation (extrinsic camera parameters) are estimated by a minimization of an overall cost functional [5], [6], [7], [8], [9], [10], =-=[11]-=-, [12], [13]. Many calibration techniques use both nonlinear minimization and closed-form solutions as in [14]. In this paper, we develop a coordinated refinement technique for the extrinsic and intri... |

6 | Nonmetric calibration of wide-angle lenses - Swaminathan, Nayar - 1998 |

5 |
A Linear Iterative Method for Auto-Calibration Using the DAC Equation
- Seo, Heyden, et al.
- 2001
(Show Context)
Citation Context ...ion pattern and intrinsic camera parameters as well as camera pose and orientation (extrinsic camera parameters) are estimated by a minimization of an overall cost functional [5], [6], [7], [8], [9], =-=[10]-=-, [11], [12], [13]. Many calibration techniques use both nonlinear minimization and closed-form solutions as in [14]. In this paper, we develop a coordinated refinement technique for the extrinsic and... |

3 |
On-Line Color Camera Calibration
- Marszalec
- 1994
(Show Context)
Citation Context ...ome these unwanted effects. A common approach taken toward this problem is to calibrate each camera independently through comparisons with known colors on a color calibration object/environment [28], =-=[29]-=-. The color calibration object we use, shown in Fig. 2, is a color cube with patches of known colors whose images are captured from each camera. Demosaicing coefficients are calculated independently f... |

3 |
Stereoscopic segmentation,” Int
- Yezzi, Soatto
- 2003
(Show Context)
Citation Context ...tween the reprojection of the 3D calibration object and the image measurements by evolving the distortion parameters of the cameras. This is based on a recent variational approach by Yezzi and Soatto =-=[21, 22]-=- which proposed a joint region-based image segmentation and simultaneous 3D stereo reconstruction technique. Note that early variational approaches to the 3D reconstruction problem were pioneered by F... |

2 |
A Variational Approach to Calibration of Multiple Cameras
- Unal, Yezzi, et al.
- 2004
(Show Context)
Citation Context ...nd their correct estimation is one of the first phases of a camera calibration system. 1. An initial version of this work that addresses lens distortion and color calibration can be found in [15] and =-=[16]-=-, and an initial work addressing extrinsic camera calibration appears in [3]. 0162-8828/07/$25.00 ß 2007 IEEE Published by the IEEE Computer SocietysUNAL ET AL.: A VARIATIONAL APPROACH TO PROBLEMS IN ... |

1 |
Pose and Calibration from 4 or 5 Known 3D
- Triggs, “Camera
- 1999
(Show Context)
Citation Context ... known calibration pattern and intrinsic camera parameters as well as camera pose and orientation (extrinsic camera parameters) are estimated by a minimization of an overall cost functional [5], [6], =-=[7]-=-, [8], [9], [10], [11], [12], [13]. Many calibration techniques use both nonlinear minimization and closed-form solutions as in [14]. In this paper, we develop a coordinated refinement technique for t... |