Assessing validation methods for building identification and extraction

Author

Wentz, Elizabeth A. ; Qunshan Zhao

Author_Institution

Sch. of Geogr. Sci. & Urban Planning, Arizona State Univ., Tempe, AZ, USA

fYear

2015

fDate

March 30 2015-April 1 2015

Firstpage

1

Lastpage

4

Abstract

As data sources and algorithm choices become increasingly more available for automatically extracting and reconstructing 3D buildings, methods are needed to assess the accuracy of the classification process. Our research goal is to compare validation data sources and methods to assess objectivity, sensitivity, and reliability of current validation approaches. Our results show that when relying on object-oriented methods for building classification and extraction, methods other than pixel assessment are needed.

Keywords

buildings (structures); feature extraction; geophysical image processing; image classification; image reconstruction; object-oriented methods; remote sensing; 3D building reconstruction; assessing validation method; building classification; building extraction; building identification; classification process accuracy assessment; data source; object-oriented method; objectivity assessment; reliability assessment; sensitivity assessment; Accuracy; Buildings; Data mining; Image reconstruction; Laser radar; Remote sensing; Three-dimensional displays;

fLanguage

English

Publisher

ieee

Conference_Titel

Urban Remote Sensing Event (JURSE), 2015 Joint

Conference_Location

Lausanne

Type

conf

DOI

10.1109/JURSE.2015.7120453

Filename

7120453