Semi-Supervised Graph Prototypical Networks for Hyperspectral Image Classification

Graph convolutional network (GCN) is one of the most favorable semi-supervised approaches, which demonstrates encouraging performance for hyperspectral image classification (HSIC), especially under the condition of small sample sizes. In this paper, we propose a novel semi-supervised graph prototypical network (SSGPN) for high-precise HSIC. Different from prevenient GCN, we devise a prototypical layer comprising a distance-based cross-entropy (DCE) loss function and a novel temporal entropy-based regularizer (TER) in the frameworks of SSGPN. This effective layer can facilitate to generate more discriminative embedding features along with the representative prototypes to each class, so as to achieve accurate identification of various land-cover categories. Additionally, to promote computational efficiency, we present a graph normalization (G-Norm) to accelerate the convergence speed and boost the training procedure. Experimental results demonstrate that our proposed SSGPN can obtain promising performance compared with the state-of-the-art methods.

1. Introduction

In recent years, deep learning-based supervised theories and methodologies have demonstrated excellent performance in hyperspectral image classification (HSI). The typical works include recurrent neural networks (RNN), generative adversarial networks (GAN), convolutional neural networks (CNN), to name a few. Especially for the most prevalent CNNs, Li et al. proposed an effective pixel pair feature-based CNN (PPF-CNN) [1] by combining the existed handful samples, which realized data augmentation for optimizing the classification result. Afterward, a diverse region CNN (DR-CNN) [2] was presented to excavate the abundant spectral-spatial information, which acquired improved performance. However, model overfitting and performance degradation arise with limited labeled samples as the network developing deeper.