Research on hyperspectral remote sensing image classification method based on tensor decomposition

Currently, the majority of research on hyperspectral remote sensing image classification using tensor decomposition focuses on the spectral dimension during image processing. However, it is crucial to explore how to integrate spatial and spectral features of hyperspectral remote sensing images, leveraging their inherent three-dimensional characteristics for representation. This paper introduces a method for hyperspectral remote sensing image classification based on TT decomposition. By utilizing TT decomposition to decompose the hyperspectral remote sensing image tensor data, this approach directly extracts three-dimensional features from the image itself, effectively integrating spatial and spectral information. Experimental results on two publicly available datasets demonstrate that this model outperforms other methods in hyperspectral remote sensing image classification.