作者: Mohamed A. Bencherif , Yakoub Bazi , Abderrezak Guessoum , Naif Alajlan , Farid Melgani
DOI: 10.1109/LGRS.2014.2349538
关键词:
摘要: In this letter, we propose an efficient multiclass active learning (AL) method for remote sensing image classification. We fuse the capabilities of extreme machine (ELM) classifier and graph-based optimization methods to boost classification accuracy while minimizing user interaction. First, use ELM generate initial label estimation unlabeled pixels. Then, optimize a functional energy that integrates outputs as structure. As ELM, solution problem leads system linear equations. Due sparse Laplacian matrix built from lattice graph defined on pixels, is solved in time. experiments, report discuss results proposed AL two very high resolution images acquired by IKONOS-2 GoeEye-1, well well-known Pavia University hyperspectral image.