Active Learning for Hyperspectral Image Classification Using Kernel Sparse Representation Classifiers

Active learning (AL) is one of the popular approaches that can mitigate some of the drawbacks of supervised classification. Although sparse representation classifier (SRC) has already proven to be a robust classifier and successfully used in many applications, it is seldom used jointly with AL. In this letter, we propose a novel AL technique for SRCs. In the proposed model, the query function is designed by combining uncertainty and diversity criteria, both of which are defined by using the SRC in kernel space. The proposed technique outperforms other state-of-the-art methods in terms of classification performance.