Remote sensing image classification using extreme learning machine-guided collaborative coding

Remote sensing image classification is a very challenging problem and covariance descriptor can be introduced in the feature extraction and representation process for remote sensing image. However, due to the reason that covariance descriptor lies in non-Euclidean manifold, conventional extreme learning machine (ELM) cannot effectively deal with this problem. In this paper, we propose an improved ELM framework which incorporates the collaborative coding to tackle the covariance descriptor classification problem. First, a new ELM-guided dictionary learning and coding model is proposed. Then the iterative optimization algorithm is developed to solve the model. By evaluating the proposed approach on the UCMERCED high-resolution aerial image dataset, we show the effectiveness of the proposed strategy.