Classification of iowa wetlands using an airborne hyperspectral image: A comparison of the spectral angle mapper classifier and an object-oriented approach
Canadian Journal of Remote Sensing
Wetlands mapping using multispectral imagery from Landsat multispectral scanner (MSS) and thematic mapper (TM) and Système pour l’observation de la Terre (SPOT) does not in general provide high classification accuracies because of poor spectral and spatial resolutions. This study tests the feasibility of using high-resolution hyperspectral imagery to map wetlands in Iowa with two nontraditional classification techniques: the spectral angle mapper (SAM) method and a new nonparametric object-oriented (OO) classification. The software programs used were ENVI and eCognition. Accuracies of these classified images were assessed by using the information collected through a field survey with a global positioning system and high-resolution color infrared images. Wetlands were identified more accurately with the OO method (overall accuracy 92.3%) than with SAM (63.53%). This paper also discusses the limitations of these classification techniques for wetlands, as well as discussing future directions for study. © 2005 Canadian Aeronautics and Space Institute.
Original Publication Date
DOI of published version
Harken, James and Sugumaran, Ramanathan, "Classification of iowa wetlands using an airborne hyperspectral image: A comparison of the spectral angle mapper classifier and an object-oriented approach" (2005). Faculty Publications. 3031.