IDEA #2QPSHY Human Visual System-Based Hyperspectral Band Selection and Recoloring

Band selection (BS) is one of the most exciting and challenging hyperspectral imagery (HSI) in recent years. Typically, the highly correlated spectral bands bring about information redundancy and heavy computation troubles in HSI data analysis. Hyperspectral images' high data dimensionality also increases the load on data computation, storage, and transmission. One of the goal BS is to reduce redundancy of hyperspectral imagery by selecting some informative and distinctive bands from the original hyperspectral image cube. The definition of the objective function and the search strategy is essential and crucial for each band selection. This paper focuses on a human visual system-based hyperspectral band selection in an unsupervised manner. The proposed method is the first time, to the best of our knowledge. to use the HVS based BS measure to quantify the HSI band's goodness. Also, to evaluate the visualization outcomes, both subjective visual assessment and objective metrics are used. The effectiveness of the proposed approach is demonstrated with a few real hyperspectral datasets. The experimental results show that the proposed algorithm is robust and significantly outperforms the state-of-the-art competitors' accuracy and efficiency. The presented method can be used to classify pixels of the hyperspectral image by finding the relevant bands without reducing the classification accuracy rate.