## Robust Simultaneous Registration of Multiple Range Images (2002)

### Cached

### Download Links

Citations: | 39 - 8 self |

### BibTeX

@INPROCEEDINGS{Nishino02robustsimultaneous,

author = {Ko Nishino and Katsushi Ikeuchi},

title = {Robust Simultaneous Registration of Multiple Range Images},

booktitle = {},

year = {2002},

pages = {454--461}

}

### Years of Citing Articles

### OpenURL

### Abstract

The registration problem of multiple range images is fundamental for many applications that rely on precise geometric models. We propose a robust registration method that can align multiple range images comprised of a large number of data points. The proposed method minimizes an error function that is constructed to be global against all range images, providing the ability to diffusively distribute errors instead of accumulating them. The minimization strategy is designed to be efficient and robust against outliers by using conjugate gradient search utilizing M-estimator. Also, for "better" point correspondence search, the laser reflectance strength is used as an additional attribute of each 3D data point. For robustness against data noise, the framework is designed not to use secondary information, i.e. surface normals, in its error metric. We describe the details of the proposed method, and present experimental results applying the proposed method to real data.

### Citations

2366 |
A method of registration of 3-D shapes
- Besl, McKay
- 1992
(Show Context)
Citation Context ...interaction, while most feature-based approaches do not have this requirement. Error metric: point-point distance or point-plane distance Originally, point-based approaches, such as the ICP algorithm =-=[4, 28]-=-, set the error metric basis on the Euclidean distance between two points corresponding each other [10, 20]. However, since this error metric does not take the surface information into account, the po... |

638 | An algorithm for finding best matches in logarithmic expected time
- Friedman, Bentley, et al.
- 1977
(Show Context)
Citation Context ...point in each range image can easily dominate a critical portion of the overall computational time. To obtain point correspondences efficiently, we employ k-d tree structure to store the range images =-=[11]. -=-K-d tree’s k-d abbreviates kdimensional and it is a generalization of a binary-search tree for efficient search in high dimension space. The k-d tree is created by recursively splitting a data set d... |

609 |
Object modelling by registration of multiple range images
- Chen, Medioni
- 1992
(Show Context)
Citation Context ...the geometric model will be used as a basis of texture-mapping or appearance analysis, and so on. Simultaneous registration solves this error accumulation problem by aligning all range images at once =-=[1, 2, 5, 6, 8, 16, 20, 22, 23]-=-. This can be accomplished by defining an error minimization problem by using an error metric common among all range images. This approach can diffusively distribute the registration error over all ov... |

519 | Iterative point matching for registration of free-form curves and surfaces
- Zhang
- 1994
(Show Context)
Citation Context ...interaction, while most feature-based approaches do not have this requirement. Error metric: point-point distance or point-plane distance Originally, point-based approaches, such as the ICP algorithm =-=[4, 28]-=-, set the error metric basis on the Euclidean distance between two points corresponding each other [10, 20]. However, since this error metric does not take the surface information into account, the po... |

489 |
Zippered Polygon Meshed from Range Images
- Turk, Levoy
- 1994
(Show Context)
Citation Context ...registering multiple range images can be represented by two different approaches. The straightforward strategy is to focus on only two range images at a time, and register each range image to another =-=[25]-=-. After one range image pair is registered, a new pair including either range image in the former pair, positioned in the resulting coordinate, is registered. This is repeated till all range images ar... |

244 |
The representation, recognition and locating of 3-D objects
- Faugeras, Hebert
- 1986
(Show Context)
Citation Context ...istance or point-plane distance Originally, point-based approaches, such as the ICP algorithm [4, 28], set the error metric basis on the Euclidean distance between two points corresponding each other =-=[10, 20]. Ho-=-wever, since this error metric does not take the surface information into account, the point-based approaches based on point-point distance suffer from the inability to “slide” overlapping range i... |

188 | Multiview registration for large data sets - PULLI - 1999 |

135 |
Towards a General Multi-view Registration Techniques
- Bergevin, Soucy, et al.
- 1996
(Show Context)
Citation Context ...the geometric model will be used as a basis of texture-mapping or appearance analysis, and so on. Simultaneous registration solves this error accumulation problem by aligning all range images at once =-=[1, 2, 5, 6, 8, 16, 20, 22, 23]-=-. This can be accomplished by defining an error minimization problem by using an error metric common among all range images. This approach can diffusively distribute the registration error over all ov... |

125 | Spin-Images: A Representation for 3-D Surface Matching
- Johnson
- 1997
(Show Context)
Citation Context ...ting of the range data. Feature-based methods extract some signatures around 3D points, invariant to Euclidean transformation, in each target range image and make correspondences among those features =-=[6, 17, 18]-=-. Based on the assumption that all correspondences are matched correctly, the transformation for registration can be computed in a closed form manner. On the other hand, if the signatures computed fro... |

114 | Registering multiview range data to create 3d computer objects
- Blais, Levine
- 1995
(Show Context)
Citation Context ...the geometric model will be used as a basis of texture-mapping or appearance analysis, and so on. Simultaneous registration solves this error accumulation problem by aligning all range images at once =-=[1, 2, 5, 6, 8, 16, 20, 22, 23]-=-. This can be accomplished by defining an error minimization problem by using an error metric common among all range images. This approach can diffusively distribute the registration error over all ov... |

103 | Simultaneous registration of multiple range views for use in reverse engineering of CAD models
- Eggert, Fitzgibbon, et al.
- 1998
(Show Context)
Citation Context |

98 |
The 3D model acquisition pipeline
- BERNARDINI, RUSHMEIER
(Show Context)
Citation Context ...itionally, projects to construct precise geometric models based on observation of real world objects for the purpose of digital preservation of cultural heritage objects have drawn attention recently =-=[3, 9, 21]-=-. Because of their objective, these projects require very precise registration of multiple range images. In this paper, we propose a framework to register multiple range images robustly. Taking the po... |

88 | Registration and Integration of Textured 3D Data
- Johnson, Kang
- 1997
(Show Context)
Citation Context ...ue. We gradually reduce the number of the candidates m along the iteration so as to make it inversely proportional to the number of iterations. This utilization of the laser reflectance is similar to =-=[19]-=-, which uses color attributes to narrow down the closest point candidates. Figure 3 depicts how the point-point distance metric utilizing RSV as additional attribute works in the example case depicted... |

60 | Surface registration by matching oriented points
- Johnson, Hebert
- 1997
(Show Context)
Citation Context ...ting of the range data. Feature-based methods extract some signatures around 3D points, invariant to Euclidean transformation, in each target range image and make correspondences among those features =-=[6, 17, 18]-=-. Based on the assumption that all correspondences are matched correctly, the transformation for registration can be computed in a closed form manner. On the other hand, if the signatures computed fro... |

59 | Fast global registration of 3D sampled surfaces using a multi-z-buffer technique, Image and Vision Computing 17
- Benjemaa, Schmitt
- 1999
(Show Context)
Citation Context |

51 |
et al. The Digital Michelangelo Project: 3D Scanning of Large Statues
- Levoy
- 2000
(Show Context)
Citation Context ...itionally, projects to construct precise geometric models based on observation of real world objects for the purpose of digital preservation of cultural heritage objects have drawn attention recently =-=[3, 9, 21]-=-. Because of their objective, these projects require very precise registration of multiple range images. In this paper, we propose a framework to register multiple range images robustly. Taking the po... |

41 | Automatic modeling and localization for object recognition
- Wheeler
- 1996
(Show Context)
Citation Context ...tration algorithm which is i) based on the simultaneous strategy, ii) using points as matching units, iii) with the point-point distance metric. The framework is inspired by the work of Wheeler et al =-=[26, 27]-=-, that applied similar techniques for object recognition and localization. We want to construct the geometric model to be as accurate as possible. Also as future work, we would like to accomplish appe... |

40 |
Geometrical cloning of 3d objects via simultaneous registration of multiple range images
- Neugebauer
- 1997
(Show Context)
Citation Context |

35 | Three-dimensional registration using range and intensity information - Godin, Rioux, et al. - 1994 |

35 |
Registration and Integration of Multiple Range Images for 3-D Model Construction
- Masuda, Sakaue, et al.
- 1996
(Show Context)
Citation Context |

35 |
Range-imaging system utilizing nematic liquid crystal mask
- Sato, Inokuchi
- 1987
(Show Context)
Citation Context ...propose a framework to register multiple range images robustly. Taking the point cloud images obtained through use of a range sensor, e.g., laser range scanner [14, 13, 15], light-stripe range finder =-=[24]-=-, etc., as the input, we simultaneously register all range images to sit in one common coordinate system. We highly prioritize our efforts to make the resulting registered geometric 1 model accurate c... |

19 | The Great Buddha Project: Modelling Cultural Heritage through Observation
- Miyazaki, Ooishi, et al.
- 2000
(Show Context)
Citation Context ...itionally, projects to construct precise geometric models based on observation of real world objects for the purpose of digital preservation of cultural heritage objects have drawn attention recently =-=[3, 9, 21]-=-. Because of their objective, these projects require very precise registration of multiple range images. In this paper, we propose a framework to register multiple range images robustly. Taking the po... |

12 | From images to models: Automatic 3D object model construction from multiple views - Dorai, Wang, et al. - 1996 |

6 | Surface reconstruction from misregistered data
- Jin, Duchamp, et al.
- 1995
(Show Context)
Citation Context |

4 |
Sensor Modeling, Probablistic Hypothesis Generation, and Robust Localization for ObjectRecognition
- Wheeler, Ikeuchi
- 1995
(Show Context)
Citation Context ...tration algorithm which is i) based on the simultaneous strategy, ii) using points as matching units, iii) with the point-point distance metric. The framework is inspired by the work of Wheeler et al =-=[26, 27]-=-, that applied similar techniques for object recognition and localization. We want to construct the geometric model to be as accurate as possible. Also as future work, we would like to accomplish appe... |