Share Email Print

Journal of Applied Remote Sensing • Open Access

Evaluation of the contribution of LiDAR data and postclassification procedures to object-based classification accuracy
Author(s): Diane M. Styers; L. Monika Moskal; Jeffrey J. Richardson; Meghan A. Halabisky

Paper Abstract

Object-based image analysis (OBIA) is becoming an increasingly common method for producing land use/land cover (LULC) classifications in urban areas. In order to produce the most accurate LULC map, LiDAR data and postclassification procedures are often employed, but their relative contributions to accuracy are unclear. We examined the contribution of LiDAR data and postclassification procedures to increase classification accuracies over using imagery alone and assessed sources of error along an ecologically complex urban-to-rural gradient in Olympia, Washington. Overall classification accuracy and user’s and producer’s accuracies for individual classes were evaluated. The addition of LiDAR data to the OBIA classification resulted in an 8.34% increase in overall accuracy, while manual postclassification to the imagery+LiDAR classification improved accuracy only an additional 1%. Sources of error in this classification were largely due to edge effects, from which multiple different types of errors result.

Paper Details

Date Published: 6 November 2014
PDF: 17 pages
J. Appl. Rem. Sens. 8(1) 083529 doi: 10.1117/1.JRS.8.083529
Published in: Journal of Applied Remote Sensing Volume 8, Issue 1
Show Author Affiliations
Diane M. Styers, Western Carolina Univ. (United States)
L. Monika Moskal, Univ. of Washington (United States)
Jeffrey J. Richardson, Univ. of Washington (United States)
Meghan A. Halabisky, Univ. of Washington (United States)

© SPIE. Terms of Use
Back to Top