Hyperspectral image classification method based on data expansion and consistency regularization with small samples

In the hyperspectral image (HSI) classification, convolutional neural network (CNN)-based approaches often struggle with the scarcity of labeled samples. The letter proposes an HSI classification method based on data expansion and consistency regularization with small samples. Specifically, we leverage the pixel-pair feature (PPF) to expand the data set, which facilitates the adequate tuning of CNN parameters and alleviates the issue of overfitting. Additionally, a designed CNN structure is employed to extract discriminative features from the limited number of labeled PPFs and numerous unlabeled PPFs. And the CNN is trained via minimizing the weighted sum of supervised and unsupervised losses, where the supervised loss is calculated through the cross entropy function while the unsupervised loss is evaluated with the consistency regularization item. Moreover, reliable references required in the consistency regularization item are provided after making exponential moving average (EMA) on the outputs of CNN at different training epochs. Ultimately, we conduct experiments on three real HSI data sets, and the results show that the proposed approach gains superior classification accuracy compared to several existing CNN-based approaches.