Outlier-robust dimension reduction and its impact on hyperspectral endmember extraction

Hyperspectral endmember extraction is a process to extract end-member signatures from the observed hyperspectral data of an area. The presence of outliers in the data has been proved to pose a serious problem in endmember extraction. In this paper, unlike conventional outlier detectors which may be sensitive to window settings, we propose a robust affine set fitting (RASF) algorithm for joint dimension reduction and outlier detection without any window setting. Given the number of endmembers in advance, the RASF algorithm is to find a data-representative affine set from the corrupted data, while making the effects of outliers minimum, in the least-squares error sense. The proposed RASF algorithm is then combined with Neyman-Pearson hypothesis testing, termed RASF-NP, to further estimate the number of outliers present in the data. Computer simulations demonstrate the efficacy of the proposed method, and its impact on existing endmember extraction algorithms.