Dimensionality Reduction of Hyperspectral Images: A Data-Driven Approach for Band Selection

Abstract

Hyperspectral image (HSI) contains a large number of correlated bands. The processing of such image is associated with computational challenges like Hughes phenomenon, for which the classification accuracy decreases with increasing number of bands. Hence, dimensionality reduction (DR) of HSI is performed to enhance the efficiency of data processing. However, most of the DR techniques fail to suggest the optimum number of significant bands to achieve a good classification accuracy. In reality, user also cannot perceive the required number of bands before the analysis actually starts. In order to overcome this limitation, a multi-feature-based data-driven algorithm for unsupervised band selection is proposed in this paper. In the proposed method, bands are grouped using multi-feature analysis followed by band prioritization using signal-to-noise ratio (SNR). The performance of the proposed method is compared with other state-of-the-art methods using two benchmark hyperspectral dataset in terms of overall classification accuracy and execution time. The experimental results show that the proposed method achieved more accurate results without any user intervention in the band selection step.