Deep learning has achieved great success in hyperspectral image classification. However, when processing new hyperspectral images, the existing deep learning models must be retrained from scratch with sufficient samples, which is inefficient and undesirable in practical tasks. This paper aims to explore how to accurately classify new hyperspectral images with only a few labeled samples, i.e., the hyperspectral images few-shot classification. Specifically, we design a new deep classification model based on relational network and train it with the idea of meta-learning. Firstly, the feature learning module and the relation learning module of the model can make full use of the spatial–spectral information in hyperspectral images and carry out relation learning by comparing the similarity between samples. Secondly, the task-based learning strategy can enable the model to continuously enhance its ability to learn how to learn with a large number of tasks randomly generated from different data sets. Benefitting from the above two points, the proposed method has excellent generalization ability and can obtain satisfactory classification results with only a few labeled samples. In order to verify the performance of the proposed method, experiments were carried out on three public data sets. The results indicate that the proposed method can achieve better classification results than the traditional semisupervised support vector machine and semisupervised deep learning models.

This article proposes a novel hierarchical residual network with attention mechanism (HResNetAM) for hyperspectral image spectral-spatial classification to improve the performance of conventional deep learning networks. The straightforward convolutional neural network based models have limitations in exploiting the multi-scale spatial and spectral features, and this is the key factor in dealing with the high-dimensional nonlinear characteristics present in hyperspectral images. The proposed hierarchical residual network can extract multi-scale spatial and spectral features at a granular level, so the receptive fields range of this network will be increased, which can enhance the feature representation ability of the model. Besides, we utilize the attention mechanism to set adaptive weights for spatial and spectral features of different scales, and this can further improve the discriminative ability of extracted features. Furthermore, the double branch structure is also exploited to extract spectral and spatial features with corresponding convolution kernels in parallel, and the extracted spatial and spectral features of multiple scales are fused for hyperspectral image classification. Four benchmark hyperspectral datasets collected by different sensors and at different acquisition time are employed for classification experiments, and comparative results reveal that the proposed method has competitive advantages in terms of classification performance when compared with other state-of-the-art deep learning models.

Recently, data-driven methods represented by deep learning (DL) have been widely used in hyperspectral image (HSI) classification and achieved the promising success. However, using less labeled samples to obtain higher classification accuracy is still a challenging task. In this study, we propose a patch-free bilateral network (PBiNet) for HSI classification. In order to make better use of the features with different scales, PBiNet uses the spatial path and the semantic path to obtain different level features for classification. The spatial path with small stride is used to retain the spatial detail information. The semantic path with fast down sampling rate is used to retain high-level semantic information. Using fast downsampling rates is to expand the scope of receptive field, so that semantic branch can better focus on global information. Then we design a feature fusion module to fuse the features obtained by the two paths. Finally, we use the classification maps produced by different scale features to calculate the loss function to optimize the whole model. Due to the better use of different levels of features, the proposed method could achieve higher classification accuracy with limited labeled samples. More importantly, because the whole HSI is used as the input, the proposed method has higher computational efficiency. To verify the effectiveness of the proposed method, we carried out classification experiments on four popular HSI datasets. Quantitative and qualitative experimental results show that the accuracy of the proposed method exceeds the compared methods.