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.

No abstract

Deep learning based methods have seen a massive rise in popularity for hyperspectral image classification over the past few years. However, the success of deep learning is attributed greatly to numerous labeled samples. It is still very challenging to use only a few labeled samples to train deep learning models to reach a high classification accuracy. An active deep-learning framework trained by an end-to-end manner is, therefore, proposed by this paper in order to minimize the hyperspectral image classification costs. First, a deep densely connected convolutional network is considered for hyperspectral image classification. Different from the traditional active learning methods, an additional network is added to the designed deep densely connected convolutional network to predict the loss of input samples. Then, the additional network could be used to suggest unlabeled samples that the deep densely connected convolutional network is more likely to produce a wrong label. Note that the additional network uses the intermediate features of the deep densely connected convolutional network as input. Therefore, the proposed method is an end-to-end framework. Subsequently, a few of the selected samples are labelled manually and added to the training samples. The deep densely connected convolutional network is therefore trained using the new training set. Finally, the steps above are repeated to train the whole framework iteratively. Extensive experiments illustrates that the method proposed could reach a high accuracy in classification after selecting just a few samples.

Deep learning based methods have made great progress in hyperspectral image classification. However, training a deep learning model often requires a large number of labeled samples, which are not always available in practical applications. In this paper, a simple but innovative classification paradigm to exploit morphological attribute profile cube is proposed to improve the small sample classification performance of hyperspectral image. First, morphological attribute profiles are constructed by applying different morphological filters to hyperspectral image. Morphological attribute profile cubes are then extracted as the feature of a sample. Second, the obtained morphological attribute profile cubes are scanned with multiple scale sliding windows to make full use of the rich spatial-spectral information. Finally, the features after multi-grained scanning are input into a deep forest classifier to complete the classification task. In this way, the proposed method could use a deep network structure to improve the classification accuracy. To demonstrate the effectiveness of the proposed method, the classification experiments are carried on three widely used hyperspectral data sets. The experimental results demonstrate that the proposed method can outperform the conventional semi-supervised methods and the state-of-the-art deep learning based methods. The demo code on the Salinas dataset is released on the page : https://github.com/liubing220524/MAPC-DRF-HSI. INDEX TERMS Hyperspectral image classification, morphological attribute profile cube, multi-grained scanning, deep random forest.

Recently, the deep learning models have achieved great success in hyperspectral images classification. However, most of the deep learning models fail to obtain satisfactory results under the condition of small samples due to the contradiction between the large parameter space of the deep learning models and the insufficient labeled samples in hyperspectral images. To address the problem, a deep model based on the induction network is designed in this paper to improve the classification performance of hyperspectral images under the condition of small samples. Specifically, the typical meta-training strategy is adopted, enabling the model to acquire stronger generalization ability, so as to accurately distinguish the new classes with only a few labeled samples (e.g., 5 samples per class). Moreover, in order to deal with the disturbance caused by the various characteristics of the samples in the same class in HSI, the classwise induction module is introduced utilizing the dynamic routing algorithm, which can induce the sample-wise representations to the class-wise level representations. The obtained class-wise level representations possess better separability, allowing the designed model to generate more accurate and robust classification results. Extensive experiments are carried out on three public hyperspectral images to verify the effectiveness of the proposed method. The results demonstrate that our method outperforms existing deep learning methods under the condition of small samples.