Integration of conditional random fields and attribute grammars for range data interpretation of man-made objects

Jörg Schmittwilken, Michael Ying Yang, Wolfgang Förstner, Lutz Plümer

Research output: Contribution to journalArticleAcademicpeer-review

1 Citation (Scopus)

Abstract

A new concept for the integration of low- and high-level reasoning for the interpretation of images of man-made objects is described. The focus is on the 3D reconstruction of facades, especially the transition area between buildings and the surrounding ground. The aim is the identification of semantically meaningful objects such as stairs, entrances, and windows. A low-level module based on random sample consensus (RANSAC) algorithm generates planar polygonal patches. Conditional random fields (CRFs) are used for their classification, based on local neighborhood and priors from the grammar. An attribute grammar is used to represent semantic knowledge including object partonomy and observable geometric constraints. The AND-OR tree-based parser uses the precision of the classified patches to control the reconstruction process and to optimize the sampling mechanism of RANSAC. Although CRFs are close to data, attribute grammars make the high-level structure of objects explicit and translate semantic knowledge in observable geometric constraints. Our approach combines top-down and bottom-up reasoning by integrating CRF and attribute grammars and thus exploits the complementary strengths of these methods.

Original languageEnglish
Pages (from-to)117-126
Number of pages10
JournalAnnals of GIS
Volume15
Issue number2
DOIs
Publication statusPublished - 1 Jan 2009
Externally publishedYes

Fingerprint

data interpretation
Semantics
Stairs
Facades
top-down approach
Sampling
sampling
attribute

Keywords

  • Attribute grammars
  • Conditional random fields
  • Facade interpretation
  • High- and low-level integration
  • Range data

Cite this

Schmittwilken, Jörg ; Yang, Michael Ying ; Förstner, Wolfgang ; Plümer, Lutz. / Integration of conditional random fields and attribute grammars for range data interpretation of man-made objects. In: Annals of GIS. 2009 ; Vol. 15, No. 2. pp. 117-126.
@article{0b8c358b666a4cbfb92ba9b8996561e7,
title = "Integration of conditional random fields and attribute grammars for range data interpretation of man-made objects",
abstract = "A new concept for the integration of low- and high-level reasoning for the interpretation of images of man-made objects is described. The focus is on the 3D reconstruction of facades, especially the transition area between buildings and the surrounding ground. The aim is the identification of semantically meaningful objects such as stairs, entrances, and windows. A low-level module based on random sample consensus (RANSAC) algorithm generates planar polygonal patches. Conditional random fields (CRFs) are used for their classification, based on local neighborhood and priors from the grammar. An attribute grammar is used to represent semantic knowledge including object partonomy and observable geometric constraints. The AND-OR tree-based parser uses the precision of the classified patches to control the reconstruction process and to optimize the sampling mechanism of RANSAC. Although CRFs are close to data, attribute grammars make the high-level structure of objects explicit and translate semantic knowledge in observable geometric constraints. Our approach combines top-down and bottom-up reasoning by integrating CRF and attribute grammars and thus exploits the complementary strengths of these methods.",
keywords = "Attribute grammars, Conditional random fields, Facade interpretation, High- and low-level integration, Range data",
author = "J{\"o}rg Schmittwilken and Yang, {Michael Ying} and Wolfgang F{\"o}rstner and Lutz Pl{\"u}mer",
year = "2009",
month = "1",
day = "1",
doi = "10.1080/19475680903464696",
language = "English",
volume = "15",
pages = "117--126",
journal = "Annals of GIS",
issn = "1947-5683",
publisher = "Taylor & Francis",
number = "2",

}

Integration of conditional random fields and attribute grammars for range data interpretation of man-made objects. / Schmittwilken, Jörg; Yang, Michael Ying; Förstner, Wolfgang; Plümer, Lutz.

In: Annals of GIS, Vol. 15, No. 2, 01.01.2009, p. 117-126.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - Integration of conditional random fields and attribute grammars for range data interpretation of man-made objects

AU - Schmittwilken, Jörg

AU - Yang, Michael Ying

AU - Förstner, Wolfgang

AU - Plümer, Lutz

PY - 2009/1/1

Y1 - 2009/1/1

N2 - A new concept for the integration of low- and high-level reasoning for the interpretation of images of man-made objects is described. The focus is on the 3D reconstruction of facades, especially the transition area between buildings and the surrounding ground. The aim is the identification of semantically meaningful objects such as stairs, entrances, and windows. A low-level module based on random sample consensus (RANSAC) algorithm generates planar polygonal patches. Conditional random fields (CRFs) are used for their classification, based on local neighborhood and priors from the grammar. An attribute grammar is used to represent semantic knowledge including object partonomy and observable geometric constraints. The AND-OR tree-based parser uses the precision of the classified patches to control the reconstruction process and to optimize the sampling mechanism of RANSAC. Although CRFs are close to data, attribute grammars make the high-level structure of objects explicit and translate semantic knowledge in observable geometric constraints. Our approach combines top-down and bottom-up reasoning by integrating CRF and attribute grammars and thus exploits the complementary strengths of these methods.

AB - A new concept for the integration of low- and high-level reasoning for the interpretation of images of man-made objects is described. The focus is on the 3D reconstruction of facades, especially the transition area between buildings and the surrounding ground. The aim is the identification of semantically meaningful objects such as stairs, entrances, and windows. A low-level module based on random sample consensus (RANSAC) algorithm generates planar polygonal patches. Conditional random fields (CRFs) are used for their classification, based on local neighborhood and priors from the grammar. An attribute grammar is used to represent semantic knowledge including object partonomy and observable geometric constraints. The AND-OR tree-based parser uses the precision of the classified patches to control the reconstruction process and to optimize the sampling mechanism of RANSAC. Although CRFs are close to data, attribute grammars make the high-level structure of objects explicit and translate semantic knowledge in observable geometric constraints. Our approach combines top-down and bottom-up reasoning by integrating CRF and attribute grammars and thus exploits the complementary strengths of these methods.

KW - Attribute grammars

KW - Conditional random fields

KW - Facade interpretation

KW - High- and low-level integration

KW - Range data

UR - http://www.scopus.com/inward/record.url?scp=84878675669&partnerID=8YFLogxK

UR - https://ezproxy2.utwente.nl/login?url=https://library.itc.utwente.nl/login/2009/isi/yang_int.pdf

U2 - 10.1080/19475680903464696

DO - 10.1080/19475680903464696

M3 - Article

VL - 15

SP - 117

EP - 126

JO - Annals of GIS

JF - Annals of GIS

SN - 1947-5683

IS - 2

ER -