#### DMCA

## Range scan registration using reduced deformable models (2009)

Venue: | EG |

Citations: | 30 - 1 self |

### Citations

11972 | Maximum Likelihood from Incomplete Data via the EM Algorithm
- Dempster, Laird, et al.
- 1977
(Show Context)
Citation Context ...,W We solve this problem in an iterative fashion, where we update the model parameters T and W until convergence. At each iteration, we perform an alternating optimization similar to the EM algorithm =-=[DLR77]-=- where we 1. (T-step) fix the current weights W to be constant and solve for the rigid transformations T , then 2. (W-step) using the rigid transformations T from the Tstep, solve for the weight funct... |

3926 |
Random Sample Consensus: A Paradigm for Model Fitting with Applications to Image Analysis and Automated Cartography,
- Fischler, Bolles
- 1981
(Show Context)
Citation Context ...atching with them alone gives rise to many incorrect correspondences due to repeated shape features. To deal robustly with such outliers, we treat each labeled region independently and perform RANSAC =-=[FB81]-=- to estimate a rigid transformation and remove spurious correspondences. This step proved to be useful in our experiments: on average anywhere from 20% to 40% of correspondence candidates were removed... |

3114 | A method for registration of 3-D shapes,” - Besl - 1992 |

1362 | Least squares quantization in pcm.
- Lloyd
- 1982
(Show Context)
Citation Context ... combine regions and throw away extra labels. Initial Segmentation. To initialize the weight function, we generate a random segmentation of the range scan, similar to the k-means clustering algorithm =-=[Llo82]-=-. First, we randomly pick initial bone locations for each bone j on the source shape. In this step, we ensure that these locations are sufficiently spaced apart by using best-candidate samc○ 2008 The ... |

356 | A new point matching algorithm for non-rigid registration. - Chui, Rangarajan - 2003 |

290 | The space of human body shapes: reconstruction and parameterization from range scans. - Allen, Curless, et al. - 2003 |

280 | Pose Space Deformations: A Unified Approach to Shape Interpolation and Skeleton-Driven Deformation. - LEWIS, CORDNER, et al. - 2000 |

168 | Spin-Images: A Representation for 3-D Surface Matching.
- Johnson
- 1997
(Show Context)
Citation Context ...ated using a closest point strategy with a modified distance measure. This measure computes a weighted average of two components: the distance between the points, and the distance between the normals =-=[Joh97]-=-. Given these correspondences, the total correspondence error is measured using the point-to-plane distance metric. To make this term more robust to outliers and missing geometry, we do not include a ... |

165 | Automatic rigging and animation of 3D characters.
- Baran, Popovic
- 2007
(Show Context)
Citation Context ...onably accurate starting guesses in almost all of our experiments. 5. Experimental Results We tested our algorithm with five datasets: one synthetic dataset of a walking figure generated by Pinocchio =-=[BP07]-=-, two real datasets of a car and a robot [PG08], a dataset of a moving hand [WLG07], and a dataset of human body scans [ACP02]. Each dataset consists of a sequence of depth scans of a moving object. A... |

152 | Articulated Body Deformation From Range Scan Data.
- ALLEN, CURLESS, et al.
- 2002
(Show Context)
Citation Context ...datasets: one synthetic dataset of a walking figure generated by Pinocchio [BP07], two real datasets of a car and a robot [PG08], a dataset of a moving hand [WLG07], and a dataset of human body scans =-=[ACP02]-=-. Each dataset consists of a sequence of depth scans of a moving object. All tests were performed on a single core of an Intel Core 2 Duo 2.66GHz with 2GB of RAM. Robot Walk Figure 9: More difficult r... |

134 | Skinning mesh animations.
- JAMES, TWIGG
- 2005
(Show Context)
Citation Context ...imization. In addition, we adapt the “abstract” notion of skeletal structure, where each “bone” of the skeleton is described implicitly by the weight function associated with each bone transformation =-=[JT05]-=-. In this case, a deformation is described by a function D(x) : R 3 → R 3 representing the deformation of space as a weighted sum of rigid transformations D(x) = ∑w j(x)Tj(x) = ∑w j(x) j j ( ) R j(x) ... |

129 | Building Efficient, Accurate Character Skins From Examples. - MOHR, GLEICHER - 2003 |

120 | Shape by Example. - SLOAN, R, et al. - 2001 |

116 | EigenSkin: Real time large deformation character skinning in hardware. - KRY, JAMES, et al. - 2002 |

113 | The Correlated Correspondence Algorithm for Unsupervised Registration of Nonrigid Surfaces. - Anguelov, Srinivasan, et al. - 2004 |

85 |
Spectrally optimal sampling for distribution ray tracing.
- MITCHELL
- 1991
(Show Context)
Citation Context ...nd Target Registration Influence Weights Car 0% Figure 8: Registration Error. The color-coded visualization shows the distance to the closest point as a percentage of the bounding box diagonal. pling =-=[Mit91]-=-. Then, we initialize binary weight values vc at each grid point. To do this, we find the bone j whose location is closest to the grid point, and then we set the j th weight component to be 1 and the ... |

85 | Example-based 3d scan completion. In - Pauly, Mitra, et al. - 2005 |

85 | Multi-Weight Enveloping: Least-Squares Approximation Techniques for Skin Animation. - WANG, PHILLIPS - 2002 |

81 |
Fast 3d scanning with automatic motion compensation.
- Weise, Leibe, et al.
- 2007
(Show Context)
Citation Context ... Results We tested our algorithm with five datasets: one synthetic dataset of a walking figure generated by Pinocchio [BP07], two real datasets of a car and a robot [PG08], a dataset of a moving hand =-=[WLG07]-=-, and a dataset of human body scans [ACP02]. Each dataset consists of a sequence of depth scans of a moving object. All tests were performed on a single core of an Intel Core 2 Duo 2.66GHz with 2GB of... |

80 |
Global correspondence optimization for non-rigid registration of depth scans
- Li, Sumner, et al.
- 2008
(Show Context)
Citation Context ...nment by Chang and Zwicker can be thought of as binary weights, but this leads to unnatural deformations. This problem is alleviated by optimizing continuous skinning weights in our method. Li et al. =-=[LSP08]-=- also use a reduced set of transformations to model the deformation. However, this work prescribes weights on the surface in the form of a deformation graph [SSP07], and the weights are not allowed to... |

78 | Global non-rigid alignment of 3-D scans. - Brown, Rusinkiewicz - 2007 |

72 | Optimal step nonrigid ICP algorithms for surface registration. In CVPR - Amberg, Romdhani, et al. - 2007 |

72 | Embedded deformation for shape manipulation
- Sumner, Schmid, et al.
- 2007
(Show Context)
Citation Context ... surface. However, neither approach solves for the skinning weights. Instead, Huang et al. solve for deformed sample positions and interpolate the rest of the shape using prescribed influence weights =-=[SSP07]-=-. The transformation assignment by Chang and Zwicker can be thought of as binary weights, but this leads to unnatural deformations. This problem is alleviated by optimizing continuous skinning weights... |

69 |
Non-rigid registration under isometric deformations. SGP
- Huang, Adams, et al.
- 2008
(Show Context)
Citation Context ...not model the individual motion of each point, but instead we explicitly solve for skinning weights that determine groups of points moving together. This is similar to the recent work by Huang et al. =-=[HAWG08]-=-, which optimizes the deformation by clustering the motion of the surface into rigid parts, and the work by Chang and Zwicker [CZ08], which determines a coherent assignment of transformations to the s... |

57 | Real-time enveloping with rotational regression - Wang, Pulli, et al. - 2007 |

56 | DESBRUN M.: Variational shape approximation - COHEN-STEINER, ALLIEZ |

50 | Recovering articulated object models from 3d range data. - Anguelov, Koller, et al. - 2004 |

47 | ZABIH R.: Fast approximate energy minimization via graph cuts - BOYKOV, VEKSLER |

47 | and Gérard Medioni. Object modelling by registration of multiple range images - Chen - 1992 |

46 | An extension of the ICP algorithm for modeling nonrigid objects with mobile robots. - HAHNEL, THRUN, et al. - 2003 |

46 | Dynamic geometry registration - MITRA, FLORY, et al. - 2007 |

46 | Reconstruction of deforming geometry from time-varying point clouds - WAND, JENKE, et al. - 2007 |

45 | Adaptive space deformations based on rigid cells
- Botsch, Pauly, et al.
- 2007
(Show Context)
Citation Context ...) + βEjoint(T ,W), (5) where α and β are coefficients to weight each term. Since we require the transformations T to be rigid, this becomes a non-linear optimization problem. Similar to Botsch et al. =-=[BPWG07]-=-, we solve this problem iteratively using the Gauss-Newton method, linearizing both the residuals and the rotation matrices. 4.2. Solving the W-step In the W-step, we keep the transformations T fixed ... |

38 |
Articulated object reconstruction and markerless motion capture from depth video
- Pekelny, Gotsman
- 2008
(Show Context)
Citation Context ...igid parts) can be expressed automatically using our system, whereas the graph-based approach requires that the graph nodes to be placed carefully in advance. Finally, the work by Pekelny and Gotsman =-=[PG08]-=- solves the problem using a manually specified segmentation to indicate each rigid part of the scanned object, but our approach does not require such user intervention. Skinning from Examples. Since w... |

34 | Space-time surface reconstruction using incompressible flow - Sharf, Alcantara, et al. - 2008 |

34 | Example-based skeleton extraction
- Schaefer, Yuksel
- 2007
(Show Context)
Citation Context ...d datasets. The red shape is the source, the blue shape is the target, and the green shape is the source deformed to match the target. squares (NNLS) problem using the approach by Schaefer and Yuksel =-=[SY07]-=-. This technique solves the least squares problem without the non-negativity constraint and repeatedly removes variables with the smallest negative values, effectively forcing their value to zero, unt... |

33 | Automatic conversion of mesh animations into skeletonbased animations - AGUIAR, THEOBALT, et al. - 2008 |

33 | What energy functions can be minimizedvia graph cuts - Kolmogorov, Zabih |

31 | C.: Skinning with dual quaternions
- Kavan, Collins
- 2007
(Show Context)
Citation Context ...ation model for its simplicity. In case LBS exhibits artifacts, such as the well known “candy wrapping” effect, our method could be adapted to optimize other RDMs, for example dualquaternion blending =-=[KCZO07]-=-. However, this would likely make the optimization procedure more computationally expensive. Our method requires sufficient separation between different parts of the shape in the source scan. If two p... |

22 | Reconstructing animated meshes from time-varying point clouds - Süßmuth, Winter, et al. |

8 |
M.: Automatic Registration for Articulated Shapes. Comput. Graph. Forum 27, 5 (2008). c© 2011 The Author(s) c© 2011 The Eurographics Association and
- CHANG, ZWICKER
(Show Context)
Citation Context ...ving together. This is similar to the recent work by Huang et al. [HAWG08], which optimizes the deformation by clustering the motion of the surface into rigid parts, and the work by Chang and Zwicker =-=[CZ08]-=-, which determines a coherent assignment of transformations to the surface. However, neither approach solves for the skinning weights. Instead, Huang et al. solve for deformed sample positions and int... |