A multi-crop identification model based on stepwise removal learning—support vector machine using remote sensing images

Abstract

The application of remote sensing to crop identification for crop monitoring has gained popularity in the past decade. Given many successful examples, however, the time required for data processing remains a challenge. In response, a multi-crop identification model was developed based on stepwise removal learning-support vector machine (SR-SVM) using remote sensing images. This classifier was constructed using a binary tree (BT) structure. In each layer it applies the SR learning algorithm with SVM to classification which reduces the calculation complexity and sample training time. The classification accuracies, however, are not affected. By adopting the new SR-SVM model, the efficiency of the multi-crop identification is improved. © 2007 Taylor & Francis Group, LLC.