A spatial information measure method for hyperspectral image visualization

Abstract

This paper presents a new band selection method for the visualization of hyperspectral images. The method selects three relevant spectral bands to build the red-green-blue composite. The proposed method uses not only mutual information, but also spatial information to select the bands. Results from the proposed method are presented and compared to other state-of-the-art methods in terms of informative content, gradient magnitude and variance. The results show that the proposed method compares favourably with alternative approaches.