## Two Geometric Algorithms for Layout Analysis (2002)

Venue: | In Workshop on Document Analysis Systems |

Citations: | 48 - 12 self |

### BibTeX

@INPROCEEDINGS{Breuel02twogeometric,

author = {Thomas M. Breuel},

title = {Two Geometric Algorithms for Layout Analysis},

booktitle = {In Workshop on Document Analysis Systems},

year = {2002},

pages = {188--199},

publisher = {Springer-Verlag}

}

### OpenURL

### Abstract

This paper presents geometric algorithms for solving two key problems in layout analysis: finding a cover of the background whitespace of a document in terms of maximal empty rectangles, and finding constrained maximum likelihood matches of geometric text line models in the presence of geometric obstacles.

### Citations

73 | Segmentation of page images using the area voronoi diagram - Kise, Sato, et al. - 1998 |

64 | W.: Statistical approaches to feature-based object recognition
- Wells
- 1997
(Show Context)
Citation Context ...old. This match score corresponds to a maximum likelihood match in the presence of Gaussian error on location and in the presence of a uniform background of noise features, as shown in the literature =-=[7].-=- Let us assume that lines are parameterized by their distance r from the origin and the orientation θ of their normal. An additional parameter, d, gives the distance of the line of descenders from th... |

59 |
Fast recognition using adaptive subdivisions of transformation space
- Breuel
- 1992
(Show Context)
Citation Context ...y orientations even in incompletely segmented text. The approach to constrained text line finding underlying the algorithm in this paper has previously been described for geometric object recognition =-=[3]-=-, and applied to text line finding [5]. Let us represent each character on the page by the point at the bottom and center of its bounding box (the alignment point). In the absence of error, for most R... |

55 | Background structure in document images
- Baird
- 1994
(Show Context)
Citation Context ... the whitespace or background structure of documents in terms of rectangular covers. Background structure analysis as an approach to document layout analysis has been described by a number of authors =-=[13, 2, 12, 8, 1, 9]-=-. The work by Baird et al. [2] analyzes background structure in terms of rectangular covers, a computationally convenient and compact representation of the background. However, past algorithms for com... |

54 | Geometric Layout Analysis Techniques for Document Image Understanding: a Review”, ITC-IRST
- Cattoni, Coianiz, et al.
- 1998
(Show Context)
Citation Context ...hem are morphology or “smearing” based approaches, projection profiles (recursive X-Y cuts), texture-based analysis, analysis of the background structure, and others (for a review and references, =-=see [6]-=-). While layout analysis is a simpler problem than general image segmentation, it still raises challenging issues in geometric algorithms and image statistics. This paper presents algorithms for addre... |

43 |
Image segmentation by shape-directed covers
- Baird, Jones, et al.
- 1990
(Show Context)
Citation Context ... the whitespace or background structure of documents in terms of rectangular covers. Background structure analysis as an approach to document layout analysis has been described by a number of authors =-=[13, 2, 12, 8, 1, 9]-=-. The work by Baird et al. [2] analyzes background structure in terms of rectangular covers, a computationally convenient and compact representation of the background. However, past algorithms for com... |

42 |
Language-free layout analysis
- Ittner, Baird
- 1993
(Show Context)
Citation Context ... the whitespace or background structure of documents in terms of rectangular covers. Background structure analysis as an approach to document layout analysis has been described by a number of authors =-=[13, 2, 12, 8, 1, 9]-=-. The work by Baird et al. [2] analyzes background structure in terms of rectangular covers, a computationally convenient and compact representation of the background. However, past algorithms for com... |

29 |
A new algorithm for the largest empty rectangle problem
- Orlowski
- 1990
(Show Context)
Citation Context ... r∈U Q(r) where U = {r ⊆ rb|∀c ∈ C : r ∩ c = ∅} (2) 2.2 Algorithm As noted above, there are several algorithms for maximal empty rectangle problems, including those from computational geom=-=etry (e.g., [11]-=-) and document analysis (e.g., [2]). Unfortunately, such algorithms tend to be fairly complex to implement and have not found widespread use. The algorithm presented in this paper for the maximum empt... |

20 | Robust least square baseline finding using a branch and bound algorithm
- Breuel
- 2002
(Show Context)
Citation Context ...gmented text. The approach to constrained text line finding underlying the algorithm in this paper has previously been described for geometric object recognition [3], and applied to text line finding =-=[5]-=-. Let us represent each character on the page by the point at the bottom and center of its bounding box (the alignment point). In the absence of error, for most Roman fonts, each such point rests eith... |

13 |
An optimisation methodology for document structure extraction on latin character documents
- Liang, Phillips, et al.
(Show Context)
Citation Context ...tely, finding a complete and correct page segmentation without knowledge of the line structure is difficult. Globally integrated solutions to page layout analysis, like those proposed by Liang et al. =-=[10]-=- avoid this issue, but appear to be complex to implement and so far have not found wide application. Constrained line finding provides a simpler alternative. A constrained line finder only needs a lis... |

12 | Finding lines under bounded error
- Breuel
- 1996
(Show Context)
Citation Context ...An algorithm for finding globally optimal solutions to the unconstrained text line finding problem has been presented in [5], based on previous work on branch-and-bound methods for geometric matching =-=[4]-=-. We will briefly review the unconstrained method here. The basic idea is to consider rectangular subsets (boxes; cartesian products of line parameter intervals) of the three-dimensional space of text... |

2 |
Conception d'un syst eme d'analyse de documents
- Trincklin
- 1984
(Show Context)
Citation Context |

1 |
Page segementation by white streams
- Pavlidis, Zhou
- 1991
(Show Context)
Citation Context |