## Feature Extraction from Point Clouds (2001)

Venue: | In Proceedings of the 10 th International Meshing Roundtable |

Citations: | 58 - 0 self |

### BibTeX

@INPROCEEDINGS{Gumhold01featureextraction,

author = {Stefan Gumhold and Xinlong Wang and Rob Macleod},

title = {Feature Extraction from Point Clouds},

booktitle = {In Proceedings of the 10 th International Meshing Roundtable},

year = {2001},

pages = {293--305}

}

### Years of Citing Articles

### OpenURL

### Abstract

This paper describes a new method to extract feature lines directly from a surface point cloud. No surface reconstruction is needed in advance, only the inexpensive computation of a neighbor graph connecting nearby points.

### Citations

1019 | Surface simplification using quadric error metrics
- Garland, Heckbert
- 1997
(Show Context)
Citation Context ...ner location. We minimize the square distance to all of the k planes. The tool to solve this minimization problem is also known as quadric error metrics, originally introduced by Garland and Heckbert =-=[11]-=- in mesh simplification to find optimal locations for the mesh vertices after an edge collapse operation. With the knowledge of the vertex grouping around the junction point, we can finally recover th... |

818 | An optimal algorithm for approximate nearest neighbor searching in fixed dimensions
- Arya, Mount, et al.
- 1998
(Show Context)
Citation Context ...s to the k nearest neighbors, where we chose k between 10 and 16. For our purposes, it was enough to compute an approximation to the Riemannian graph with the approximate nearest neighbor library ANN =-=[4]-=- in near linear time. For the brain dataset, the Riemannian graph was computed in 7 to 10 seconds for k = 10 to 16. The Riemannian graph does not minimize the number of edges in the neighbor graph. To... |

692 | Surface reconstruction from unorganized points
- Hoppe, DeRose, et al.
- 1992
(Show Context)
Citation Context ... correlation matrix together with the corresponding eigenvaluessf 0 ; 1 ; 2 g, where 0 1 2 , define the correlation ellipsoid that adopts the general form of the neighbor points. Hoppe [13] used the smallest eigenvalue of the correlation matrix to determine the normal direction of the surface. Similar to Guy and Medioni [12] we use the eigenvalues to measure the probability of feature l... |

501 | Quickhull Algorithm for Convex Hulls
- Barber, Dobkin, et al.
- 1996
(Show Context)
Citation Context ...ing time. It is practical for data sets with a size up to fifty thousand points. First we computed the Delaunay tetrahedralization of the input point set with the public domain software package Qhull =-=[5]-=-. Then we followed the approach of Adamy et al. [2] and first filtered out all triangles of the Gabriel complex. These are the triangles, that do not contain any fourth point within their minimum circ... |

363 |
Dynamic 3D models with local and global deformations: deformable superquadrics
- Terzopoulos, Metaxas
- 1991
(Show Context)
Citation Context ...mputer vision, medical imaging, and computer assisted surgical planning. Our algorithm addresses some of the problems that arise in surface reconstruction of datasets that contain creases and corners =-=[19, 9, 18, 6, 3]-=-. For these approaches it is very helpful to extract the feature lines and junctions beforehand. Most of the reconstruction algorithms that can accommodate creases and corners do so either implicitly ... |

299 | R.Fisher., "Direct least-squares fitting of ellipses
- Pilu, Fitzgibbon
- 1999
(Show Context)
Citation Context ... derived from the vertex penalty functions exceeds a tolerance . The tolerance is a measure of the sensitivity of the crease detection. As all the penalty functions are normalized to the interval [0=-=; -=-1], it turns out that a value of = 1 is appropriate. If the dataset contains blunt creases, will require some fine tuning by the user. The remaining edges are entered into a queue. The MSG is built ... |

216 | The power crust
- Amenta, Choi, et al.
(Show Context)
Citation Context ...points on a near equilateral tetrahedron are perfectly manifold and there is no reason why the crease edge should be preferred over the crease cutting edge. The power crust algorithm of Amenta et al. =-=[16]-=- treats sharp edges and corners also in a post processing step. However, they reconstructed the features by extending the surfaces adjacent to the features and calculating the intersections. This idea... |

201 | Surface Modeling with Oriented Particle Systems
- Szeliski, Tonnesen
- 1992
(Show Context)
Citation Context ...mputer vision, medical imaging, and computer assisted surgical planning. Our algorithm addresses some of the problems that arise in surface reconstruction of datasets that contain creases and corners =-=[19, 9, 18, 6, 3]-=-. For these approaches it is very helpful to extract the feature lines and junctions beforehand. Most of the reconstruction algorithms that can accommodate creases and corners do so either implicitly ... |

80 |
Expressive rendering: A review of nonphotorealistic techniques
- LANSDOWN, SCHOFIELD
- 1995
(Show Context)
Citation Context ...ified surface reconstruction that removes triangles that interconnected different sides of a crease. We have not implemented this yet. 5.3 Non Photorealistic Rendering In non photorealistic rendering =-=[14]-=- surface display is often enhanced along feature lines and lines of high curvature. Our first example of surface enhancement was the bunny model in figure 1, in which we singled out the crease and bor... |

60 |
Detecting undersampling in surface reconstruction
- Dey, Giesen
- 2001
(Show Context)
Citation Context ...aces adjacent to the features and calculating the intersections. This idea is quite similar to our approach but their simple approach of surface extension does not tolerate noisy data. Dey and Giesen =-=[8]-=- proposed a method to detect undersampled regions, which allows the detection of border loops, creases and corners. The surface near the features is reconstructed in a similar way as Adamy et al. prop... |

57 | Inference of surfaces, 3D curves, and junctions from sparse, noisy 3D data
- Guy, Medioni
- 1997
(Show Context)
Citation Context ...crease lines and junctions. Sinha and Schunk [17] fit spline patches to depth images and adjusted the spline parameters such that the spline patches could bend sharply around creases. Guy and Medioni =-=[12]-=- described a robust algorithm to extract surfaces, feature lines and feature junctions from noisy point clouds. They discretized the space around the point cloud into a volume grid and accumulated for... |

39 |
A two-stage algorithm for discontinuity-preserving surface reconstruction
- Sinha, Schunk
- 1992
(Show Context)
Citation Context ...ed, contains a term that penalizes the generation of a crease line in order to avoid permanent creation of creases. This approach can also reconstruct the crease lines and junctions. Sinha and Schunk =-=[17]-=- fit spline patches to depth images and adjusted the spline parameters such that the spline patches could bend sharply around creases. Guy and Medioni [12] described a robust algorithm to extract surf... |

35 | Automatic reconstruction of unstructured 3D data : Combining a medial axis and implicit surfaces
- Bittar, Tsingos, et al.
- 1995
(Show Context)
Citation Context ...mputer vision, medical imaging, and computer assisted surgical planning. Our algorithm addresses some of the problems that arise in surface reconstruction of datasets that contain creases and corners =-=[19, 9, 18, 6, 3]-=-. For these approaches it is very helpful to extract the feature lines and junctions beforehand. Most of the reconstruction algorithms that can accommodate creases and corners do so either implicitly ... |

31 | New techniques for topologically correct surface reconstruction
- ADAMY, GIESEN, et al.
- 2000
(Show Context)
Citation Context ...des. The tetrahedron described by the problematic points need not be exactly equilateral as the locations of the input points could accidentally favor constellations that cut the crease. Adamy et al. =-=[2]-=- described corner and crease reconstruction as a post processing step of finding a triangulation describing a valid manifold surface. The previously described case is again a problem. Both constellati... |

25 |
Surface reconstruction from unstructured 3d data
- Algorri, Schmitt
- 1996
(Show Context)
Citation Context |

20 |
Graph-based surface reconstruction using structures in scattered point sets
- Mencl, Muller
- 1998
(Show Context)
Citation Context ...rehand. Most of the reconstruction algorithms that can accommodate creases and corners do so either implicitly or in a post processing step. The graph-based surface reconstruction of Mencl and Muller =-=[15]-=- handles sharp edges by maximizing for each point the sum of dihedral angles of the incident faces. The optimization is performed only locally and depends on the order in which the input points are pr... |

19 | Surface reconstruction in almost linear time under locally uniform sampling
- Dey, Funke, et al.
- 2001
(Show Context)
Citation Context ...detect undersampled regions, which allows the detection of border loops, creases and corners. The surface near the features is reconstructed in a similar way as Adamy et al. proposed. Funke and Ramos =-=[10]-=- described a surface reconstruction method based on a directional nearest neighbor search, that avoids the construction of a voronoi diagram and runs in near linear time. So far their algoFigure 2. a)... |

18 | Reconstructing Surfaces from Unstructured 3D Points
- FUA, SANDER
- 1991
(Show Context)
Citation Context |

15 |
Invariant surface reconstruction using weak continuity constraints
- Blake, Zisserman
- 1986
(Show Context)
Citation Context ...vision literature there exist quite old approaches for depth images that can handle features correctly and extract the feature lines. The weak constraint optimization technique by Blake and Zisserman =-=[7]-=- uses a non linear fitting approach. The global energy functional, which is minimized, contains a term that penalizes the generation of a crease line in order to avoid permanent creation of creases. T... |