Land Cover Mapping Using Combinationand Ensemble Classifiers

摘要

In recent years, large scale land cover maps constructed from remotely sensed data havernbecome important information sources for resource management. Classifiers arerncommonly used to predict land cover for unsampled map units; hence, they play a keyrnrole in map construction. Achieving adequate classifier accuracy is often problematic forrnhighly variable and difficult-to-sample landscapes. This article investigates a variety ofrnmethods for improving accuracy based on 1) combining a few different classifiers, and 2)rncreating ensembles of many classifiers. In addition, we derive an analytic expression forrnthe exact bagging к-nearest neighbor classifier.