Object vs. pixel: A systematic evaluation in urban environments

Soe Myint, Christopher S. Galletti, Shai Kaplan, Won Kyung Kim

Research output: Contribution to journalArticle

15 Citations (Scopus)

Abstract

The objective of this paper is to compare object-based and per-pixel classifiers in a systematic manner using high resolution urban imagery. The prevailing opinion is that object-based methods perform better than single-pixel classifiers, but there has been no formal investigation of this claim using multiple images and identical training samples in a detailed land-cover classification. Furthermore, there has been no standardized study of how different object-based segmentation and scale parameters improve high resolution urban classifications. We used two subsets of QuickBird over Phoenix and Scottsdale, Arizona, to test these issues. Our results show that small-scale segmentation (10) produces higher accuracy. A combination of equally balanced shape and spectral homogeneity (0.5) with compactness parameter of 0.5 is the most effective for image segmentation. The highest overall accuracy was achieved using a per-pixel Minimum distance classifier, but it was only marginally more accurate than the object-based classification.

Original languageEnglish (US)
Pages (from-to)657-678
Number of pages22
JournalGeocarto International
Volume28
Issue number7
DOIs
StatePublished - Nov 2013

Fingerprint

segmentation
pixel
evaluation
QuickBird
homogeneity
land cover
imagery
parameter
opinion
method
test

Keywords

  • object
  • QuickBird
  • segmentation
  • supervised
  • unsupervised

ASJC Scopus subject areas

  • Water Science and Technology
  • Geography, Planning and Development

Cite this

Object vs. pixel : A systematic evaluation in urban environments. / Myint, Soe; Galletti, Christopher S.; Kaplan, Shai; Kim, Won Kyung.

In: Geocarto International, Vol. 28, No. 7, 11.2013, p. 657-678.

Research output: Contribution to journalArticle

Myint, Soe ; Galletti, Christopher S. ; Kaplan, Shai ; Kim, Won Kyung. / Object vs. pixel : A systematic evaluation in urban environments. In: Geocarto International. 2013 ; Vol. 28, No. 7. pp. 657-678.
@article{877b089088124bcd9d11433f5b9b5c1d,
title = "Object vs. pixel: A systematic evaluation in urban environments",
abstract = "The objective of this paper is to compare object-based and per-pixel classifiers in a systematic manner using high resolution urban imagery. The prevailing opinion is that object-based methods perform better than single-pixel classifiers, but there has been no formal investigation of this claim using multiple images and identical training samples in a detailed land-cover classification. Furthermore, there has been no standardized study of how different object-based segmentation and scale parameters improve high resolution urban classifications. We used two subsets of QuickBird over Phoenix and Scottsdale, Arizona, to test these issues. Our results show that small-scale segmentation (10) produces higher accuracy. A combination of equally balanced shape and spectral homogeneity (0.5) with compactness parameter of 0.5 is the most effective for image segmentation. The highest overall accuracy was achieved using a per-pixel Minimum distance classifier, but it was only marginally more accurate than the object-based classification.",
keywords = "object, QuickBird, segmentation, supervised, unsupervised",
author = "Soe Myint and Galletti, {Christopher S.} and Shai Kaplan and Kim, {Won Kyung}",
year = "2013",
month = "11",
doi = "10.1080/10106049.2013.776642",
language = "English (US)",
volume = "28",
pages = "657--678",
journal = "Geocarto International",
issn = "1010-6049",
publisher = "Taylor and Francis Ltd.",
number = "7",

}

TY - JOUR

T1 - Object vs. pixel

T2 - A systematic evaluation in urban environments

AU - Myint, Soe

AU - Galletti, Christopher S.

AU - Kaplan, Shai

AU - Kim, Won Kyung

PY - 2013/11

Y1 - 2013/11

N2 - The objective of this paper is to compare object-based and per-pixel classifiers in a systematic manner using high resolution urban imagery. The prevailing opinion is that object-based methods perform better than single-pixel classifiers, but there has been no formal investigation of this claim using multiple images and identical training samples in a detailed land-cover classification. Furthermore, there has been no standardized study of how different object-based segmentation and scale parameters improve high resolution urban classifications. We used two subsets of QuickBird over Phoenix and Scottsdale, Arizona, to test these issues. Our results show that small-scale segmentation (10) produces higher accuracy. A combination of equally balanced shape and spectral homogeneity (0.5) with compactness parameter of 0.5 is the most effective for image segmentation. The highest overall accuracy was achieved using a per-pixel Minimum distance classifier, but it was only marginally more accurate than the object-based classification.

AB - The objective of this paper is to compare object-based and per-pixel classifiers in a systematic manner using high resolution urban imagery. The prevailing opinion is that object-based methods perform better than single-pixel classifiers, but there has been no formal investigation of this claim using multiple images and identical training samples in a detailed land-cover classification. Furthermore, there has been no standardized study of how different object-based segmentation and scale parameters improve high resolution urban classifications. We used two subsets of QuickBird over Phoenix and Scottsdale, Arizona, to test these issues. Our results show that small-scale segmentation (10) produces higher accuracy. A combination of equally balanced shape and spectral homogeneity (0.5) with compactness parameter of 0.5 is the most effective for image segmentation. The highest overall accuracy was achieved using a per-pixel Minimum distance classifier, but it was only marginally more accurate than the object-based classification.

KW - object

KW - QuickBird

KW - segmentation

KW - supervised

KW - unsupervised

UR - http://www.scopus.com/inward/record.url?scp=84887205991&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84887205991&partnerID=8YFLogxK

U2 - 10.1080/10106049.2013.776642

DO - 10.1080/10106049.2013.776642

M3 - Article

AN - SCOPUS:84887205991

VL - 28

SP - 657

EP - 678

JO - Geocarto International

JF - Geocarto International

SN - 1010-6049

IS - 7

ER -