HSI-SSTRANS: Hyperspectral Image Classification with Spectral and Space Transformer

Abstract

Hyperspectral image (HSI) classification is an important technique in the field of remote sensing. In the HSI classification task, there is the phenomenon of different spectral information of the same substance and different substances of the same spectral information. In order to solve this problem, a method for processing spectral information and spatial information is proposed. At the same time, the method based on convolutional neural network (CNN) only considers local information and limits its representation ability. In order to obtain more features, the Transformer structure is used to extract global information in spectrum and space. Spectral and Space Transformer is built to join feature of spatial-spectral to obtain the HSI classification structure.We evaluate the classification performance of the proposed method on IndianPines and Houston by conducting experiments, showing the superiority over other transformer networks and achieving a improvement in comparison with other backbone networks.