Detection of windows in IR building textures using masked correlation

Dorota Iwaszczuk, Ludwig Hoegner, Uwe Stilla

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

9 Scopus citations

Abstract

Infrared (IR) images depict thermal radiation of physical objects. Imaging the building hull with an IR camera allows thermal inspections. Mapping these images as textures on 3D building models, 3D geo-referencing of each pixel can be carried out. This is helpful for large area inspections. In IR images glass reflects the surrounding and shows false results for the temperature measurements. Consequently, the windows should be detected in IR images and excluded for the inspection. In this paper, an algorithm for window detection in textures extracted from terrestrial IR images is proposed. First, a local dynamic threshold is used to extract candidates for windows in the textures. Assuming a regular grid of windows masked correlation is used to find the position of windows. Finally, gaps in the window grid are replaced by hypothetical windows. Applying the method for a test dataset, 79% completeness and 80% correctness was achieved.

Original languageEnglish
Title of host publicationPhotogrammetric Image Analysis, ISPRS Conference, PIA 2011, Proceedings
Pages133-146
Number of pages14
DOIs
StatePublished - 2011
EventISPRS Conference on Photogrammetric Image Analysis, PIA 2011 - Munich, Germany
Duration: 5 Oct 20117 Oct 2011

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume6952 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceISPRS Conference on Photogrammetric Image Analysis, PIA 2011
Country/TerritoryGermany
CityMunich
Period5/10/117/10/11

Keywords

  • Image Sequences
  • Infrared
  • Structure Detection
  • Texture Mapping

Fingerprint

Dive into the research topics of 'Detection of windows in IR building textures using masked correlation'. Together they form a unique fingerprint.

Cite this