Comparison of Support Vector Machine, Artificial Neural Networks and Spectral Angle Mapper Classifiers on Fused Hyperspectral Data for Improved LULC Classification

An assessment of image classifiers for generating machine-learning training samples for mapping the invasive Campuloclinium macrocephalum (Less.) DC (pompom weed) using DESIS hyperspectral imagery

“…The SAM method was selected to represent the spectral matching algorithms in this study because SAM has been extensively assessed for mapping a variety of IAPs from spaceborne hyperspectral imagery (Ustin et al, 2002, Narumalani et al, 2006, Lawrence et al, 2006, Sahithi et al, 2019Kazmi et al, 2021). The SAM method exploits target object spectral properties to identify similar objects in hyperspectral data cubes.…”

“…The SVM is a machine-learning classification method derived from statistical learning theory and separates classes by maximizing the margin between them using a hyperplane decision surface (Mountrakis et al, 2011). The SVM was selected to represent nonparametric machine learning algorithms in this study because SVM has a low sensitivity to high dimensionality and been previously assessed for species-specific mapping of IAPs from spaceborne hyperspectral data (Melgani and Bruzzone, 2004;Mirik et al, 2013;Sahithi et al, 2019;Sabat-Tomala et al, 2020). The classification problem in SVM involves defining this optimal decision surface (Zhu and Blumberg, 2002)…”