Recent trends in classification of remote sensing data: active and semisupervised machine learning paradigms

摘要: This paper addresses the recent trends in machine learning methods for automatic classification of remote sensing (RS) images. In particular, we focus on two new paradigms: semisupervised and active learning. These paradigms allow one to address problems critical conditions where available labeled training samples are limited. operational very usual RS problems, due high cost time associated with collection samples. Semisupervised techniques enrich initial set information improve accuracy by exploiting unlabeled or requiring additional labeling phases from user, respectively. The aforementioned strategies theoretically experimentally analyzed considering SVM-based order highlight advantages disadvantages both strategies.