Hierarchical Feature Fusion and Selection for Hyperspectral Image Classification

Most existing classification methods design complicated and large deep neural network (DNN) model to deal with the ubiquitous spectral variability and nonlinearity of hyperspectral images (HSIs). However, their application is blocked by limited training samples and considerable computational costs in real scenes. To solve these problems, we propose a simple spectral hierarchical feature fusion and selection network (HFFSNet). Specifically, we apply 1-D grouped convolution for dimensionality reduction and multilevel feature extraction, then the multilevel features are fused to assist the adaptive feature selection of different layer features via the soft attention mechanism, and finally the selected features are fused to further enhance the feature representation. Extensive experimental results on three hyperspectral datasets demonstrate the effectiveness of the proposed network.

Automated summary

Most existing classification methods for hyperspectral images (HSIs) rely on complicated and large deep neural network (DNN) models, which suffer from limited training samples and high computational costs in real scenarios. To address these issues, we propose a simple spectral hierarchical feature fusion and selection network (HFFSNet) that utilizes 1-D grouped convolution for dimensionality reduction and multilevel feature extraction. The multilevel features are fused using the soft attention mechanism to assist adaptive feature selection, and the selected features are further fused to enhance the overall feature representation. Extensive experimental results on three hyperspectral datasets demonstrate the effectiveness of our proposed network.