WAVELET DOMAIN MULTI-VIEW ACTIVE LEARNING FOR HYPERSPECTRAL IMAGE ANALYSIS

This paper introduces a new wavelet based active learning approach for hyperspectral image (HSI) analysis. Specifically, it uses a redundant wavelet transform (RDWT) to construct a multi-view active learning framework for hyperspectral classification. We show that a wavelet decomposition provides a unique multi-view framework that results in improved active learning and classification, and apply the proposed method to a benchmark hyperspectral dataset. Experimental results demonstrate the efficacy of the proposed method compared to traditional learning methods, including random sampling, margin sampling, and multi-view active learning based on correlated subsets of contiguous bands.