Urban image classification

Per-pixel classifiers, subpixel analysis, object-based image analysis, and geospatial methods

Soe Myint, Victor Mesev, Dale A. Quattrochi, Elizabeth Wentz

Research output: Chapter in Book/Report/Conference proceedingChapter

Abstract

Remote-sensing methods used to generate base maps to analyze the urban environment rely predominantly on digital sensor data from spaceborne platforms. This is due in part from new sources of high-spatial-resolution data covering the globe, a variety of multispectral and multitemporal sources, sophisticated statistical and geospatial methods, and compatibility with geographic information system or science (GIS) data sources and methods. The goal of this chapter is to review the four groups of classification methods for digital sensor data from spaceborne platforms; per-pixel, subpixel, object-based (spatial-based), and geospatial methods. Per-pixel methods are widely used methods that classify pixels into distinct categories based solely on the spectral and ancillary information within that pixel. They are used for simple calculations of environmental indices (e.g., normalized difference vegetation index [NDVI]) to sophisticated expert systems to assign urban land covers (Stefanov et al., 2001). Researchers recognize, however, that even with the smallest pixel size, the spectral information within a pixel is really a combination of multiple urban surfaces. Subpixel classification methods therefore aim to statistically quantify the mixture of surfaces to improve overall classification accuracy (Myint, 2006a). While within-pixel variations exist, there is also significant evidence that groups of nearby pixels have similar spectral information and therefore belong to the same classification category. Object-oriented methods have emerged that group pixels prior to classification based on spectral similarity and spatial proximity. Classification accuracy using object-based methods shows significant success and promise for numerous urban applications (Myint et al., 2011). Like the object-oriented methods that recognize the importance of spatial proximity, geospatial methods for urban mapping also utilize neighboring pixels in the classification process. The primary difference though is that geostatistical methods (e.g., spatial autocorrelation methods) are utilized during both the pre- and postclassification steps (Myint and Mesev, 2012).

Original languageEnglish (US)
Title of host publicationRemotely Sensed Data Characterization, Classification, and Accuracies
PublisherCRC Press
Pages219-230
Number of pages12
Volume1
ISBN (Electronic)9781482217872
ISBN (Print)9781482217865
DOIs
StatePublished - Jan 1 2015

Fingerprint

Image classification
image classification
image analysis
Image analysis
pixel
Classifiers
Pixels
Geographic information systems
method
analysis
Information science
Sensors
Autocorrelation
sensor
Expert systems
Remote sensing
expert system
NDVI
autocorrelation
land cover

ASJC Scopus subject areas

  • Engineering(all)
  • Environmental Science(all)
  • Earth and Planetary Sciences(all)

Cite this

Myint, S., Mesev, V., Quattrochi, D. A., & Wentz, E. (2015). Urban image classification: Per-pixel classifiers, subpixel analysis, object-based image analysis, and geospatial methods. In Remotely Sensed Data Characterization, Classification, and Accuracies (Vol. 1, pp. 219-230). CRC Press. https://doi.org/10.1201/b19294

Urban image classification : Per-pixel classifiers, subpixel analysis, object-based image analysis, and geospatial methods. / Myint, Soe; Mesev, Victor; Quattrochi, Dale A.; Wentz, Elizabeth.

Remotely Sensed Data Characterization, Classification, and Accuracies. Vol. 1 CRC Press, 2015. p. 219-230.

Research output: Chapter in Book/Report/Conference proceedingChapter

Myint, S, Mesev, V, Quattrochi, DA & Wentz, E 2015, Urban image classification: Per-pixel classifiers, subpixel analysis, object-based image analysis, and geospatial methods. in Remotely Sensed Data Characterization, Classification, and Accuracies. vol. 1, CRC Press, pp. 219-230. https://doi.org/10.1201/b19294
Myint S, Mesev V, Quattrochi DA, Wentz E. Urban image classification: Per-pixel classifiers, subpixel analysis, object-based image analysis, and geospatial methods. In Remotely Sensed Data Characterization, Classification, and Accuracies. Vol. 1. CRC Press. 2015. p. 219-230 https://doi.org/10.1201/b19294
Myint, Soe ; Mesev, Victor ; Quattrochi, Dale A. ; Wentz, Elizabeth. / Urban image classification : Per-pixel classifiers, subpixel analysis, object-based image analysis, and geospatial methods. Remotely Sensed Data Characterization, Classification, and Accuracies. Vol. 1 CRC Press, 2015. pp. 219-230
@inbook{0537681446524256b89744fc22b9556b,
title = "Urban image classification: Per-pixel classifiers, subpixel analysis, object-based image analysis, and geospatial methods",
abstract = "Remote-sensing methods used to generate base maps to analyze the urban environment rely predominantly on digital sensor data from spaceborne platforms. This is due in part from new sources of high-spatial-resolution data covering the globe, a variety of multispectral and multitemporal sources, sophisticated statistical and geospatial methods, and compatibility with geographic information system or science (GIS) data sources and methods. The goal of this chapter is to review the four groups of classification methods for digital sensor data from spaceborne platforms; per-pixel, subpixel, object-based (spatial-based), and geospatial methods. Per-pixel methods are widely used methods that classify pixels into distinct categories based solely on the spectral and ancillary information within that pixel. They are used for simple calculations of environmental indices (e.g., normalized difference vegetation index [NDVI]) to sophisticated expert systems to assign urban land covers (Stefanov et al., 2001). Researchers recognize, however, that even with the smallest pixel size, the spectral information within a pixel is really a combination of multiple urban surfaces. Subpixel classification methods therefore aim to statistically quantify the mixture of surfaces to improve overall classification accuracy (Myint, 2006a). While within-pixel variations exist, there is also significant evidence that groups of nearby pixels have similar spectral information and therefore belong to the same classification category. Object-oriented methods have emerged that group pixels prior to classification based on spectral similarity and spatial proximity. Classification accuracy using object-based methods shows significant success and promise for numerous urban applications (Myint et al., 2011). Like the object-oriented methods that recognize the importance of spatial proximity, geospatial methods for urban mapping also utilize neighboring pixels in the classification process. The primary difference though is that geostatistical methods (e.g., spatial autocorrelation methods) are utilized during both the pre- and postclassification steps (Myint and Mesev, 2012).",
author = "Soe Myint and Victor Mesev and Quattrochi, {Dale A.} and Elizabeth Wentz",
year = "2015",
month = "1",
day = "1",
doi = "10.1201/b19294",
language = "English (US)",
isbn = "9781482217865",
volume = "1",
pages = "219--230",
booktitle = "Remotely Sensed Data Characterization, Classification, and Accuracies",
publisher = "CRC Press",

}

TY - CHAP

T1 - Urban image classification

T2 - Per-pixel classifiers, subpixel analysis, object-based image analysis, and geospatial methods

AU - Myint, Soe

AU - Mesev, Victor

AU - Quattrochi, Dale A.

AU - Wentz, Elizabeth

PY - 2015/1/1

Y1 - 2015/1/1

N2 - Remote-sensing methods used to generate base maps to analyze the urban environment rely predominantly on digital sensor data from spaceborne platforms. This is due in part from new sources of high-spatial-resolution data covering the globe, a variety of multispectral and multitemporal sources, sophisticated statistical and geospatial methods, and compatibility with geographic information system or science (GIS) data sources and methods. The goal of this chapter is to review the four groups of classification methods for digital sensor data from spaceborne platforms; per-pixel, subpixel, object-based (spatial-based), and geospatial methods. Per-pixel methods are widely used methods that classify pixels into distinct categories based solely on the spectral and ancillary information within that pixel. They are used for simple calculations of environmental indices (e.g., normalized difference vegetation index [NDVI]) to sophisticated expert systems to assign urban land covers (Stefanov et al., 2001). Researchers recognize, however, that even with the smallest pixel size, the spectral information within a pixel is really a combination of multiple urban surfaces. Subpixel classification methods therefore aim to statistically quantify the mixture of surfaces to improve overall classification accuracy (Myint, 2006a). While within-pixel variations exist, there is also significant evidence that groups of nearby pixels have similar spectral information and therefore belong to the same classification category. Object-oriented methods have emerged that group pixels prior to classification based on spectral similarity and spatial proximity. Classification accuracy using object-based methods shows significant success and promise for numerous urban applications (Myint et al., 2011). Like the object-oriented methods that recognize the importance of spatial proximity, geospatial methods for urban mapping also utilize neighboring pixels in the classification process. The primary difference though is that geostatistical methods (e.g., spatial autocorrelation methods) are utilized during both the pre- and postclassification steps (Myint and Mesev, 2012).

AB - Remote-sensing methods used to generate base maps to analyze the urban environment rely predominantly on digital sensor data from spaceborne platforms. This is due in part from new sources of high-spatial-resolution data covering the globe, a variety of multispectral and multitemporal sources, sophisticated statistical and geospatial methods, and compatibility with geographic information system or science (GIS) data sources and methods. The goal of this chapter is to review the four groups of classification methods for digital sensor data from spaceborne platforms; per-pixel, subpixel, object-based (spatial-based), and geospatial methods. Per-pixel methods are widely used methods that classify pixels into distinct categories based solely on the spectral and ancillary information within that pixel. They are used for simple calculations of environmental indices (e.g., normalized difference vegetation index [NDVI]) to sophisticated expert systems to assign urban land covers (Stefanov et al., 2001). Researchers recognize, however, that even with the smallest pixel size, the spectral information within a pixel is really a combination of multiple urban surfaces. Subpixel classification methods therefore aim to statistically quantify the mixture of surfaces to improve overall classification accuracy (Myint, 2006a). While within-pixel variations exist, there is also significant evidence that groups of nearby pixels have similar spectral information and therefore belong to the same classification category. Object-oriented methods have emerged that group pixels prior to classification based on spectral similarity and spatial proximity. Classification accuracy using object-based methods shows significant success and promise for numerous urban applications (Myint et al., 2011). Like the object-oriented methods that recognize the importance of spatial proximity, geospatial methods for urban mapping also utilize neighboring pixels in the classification process. The primary difference though is that geostatistical methods (e.g., spatial autocorrelation methods) are utilized during both the pre- and postclassification steps (Myint and Mesev, 2012).

UR - http://www.scopus.com/inward/record.url?scp=85054716669&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85054716669&partnerID=8YFLogxK

U2 - 10.1201/b19294

DO - 10.1201/b19294

M3 - Chapter

SN - 9781482217865

VL - 1

SP - 219

EP - 230

BT - Remotely Sensed Data Characterization, Classification, and Accuracies

PB - CRC Press

ER -