Spectral unmixing is a technique for remotely sensed image interpretation that expresses each (possibly mixed) pixel as a combination of pure spectral signatures (endmembers) and their fractional abundances. In this paper, we develop a new technique for unsupervised unmixing which is based on a deep autoencoder network (DAEN). Our newly developed DAEN consists of two parts. The first part of the network adopts stacked autoencoders (SAEs) to learn spectral signatures, so as to generate a good initialization for the unmixing process. In the second part of the network, a variational autoencoder (VAE) is employed to perform blind source separation, aimed at obtaining the endmember signatures and abundance fractions simultaneously. By taking advantage from the SAEs, the robustness of the proposed approach is remarkable as it can unmix data sets with outliers and low signal-to-noise ratio. Moreover, the multihidden layers of the VAE ensure the required constraints (nonnegativity and sum-to-one) when estimating the abundances. The effectiveness of the proposed method is evaluated using both synthetic and real hyperspectral data. When compared with other unmixing methods, the proposed approach demonstrates very competitive performance.

Spatial–spectral classification (SSC) has become a trend for hyperspectral image (HSI) classification. However, most SSC methods mainly consider local information, so that some correlations may not be effectively discovered when they appear in regions that are not contiguous. Although many SSC methods can acquire spatial-contextual characteristics via spatial filtering, they lack the ability to consider correlations in non-Euclidean spaces. To address the aforementioned issues, we develop a new semisupervised HSI classification approach based on normalized spectral clustering with kernel-based learning (NSCKL), which can aggregate local-to-global correlations to achieve a distinguishable embedding to improve HSI classification performance. In this work, we propose a normalized spectral clustering (NSC) scheme that can learn new features under a manifold assumption. Specifically, we first design a kernel-based iterative filter (KIF) to establish vertices of the undirected graph, aiming to assign initial connections to the nodes associated with pixels. The NSC first gathers local correlations in the Euclidean space and then captures global correlations in the manifold. Even though homogeneous pixels are distributed in noncontiguous regions, our NSC can still aggregate correlations to generate new (clustered) features. Finally, the clustered features and a kernel-based extreme learning machine (KELM) are employed to achieve the semisupervised classification. The effectiveness of our NSCKL is evaluated by using several HSIs. When compared with other state-of-the-art (SOTA) classification approaches, our newly proposed NSCKL demonstrates very competitive performance. The codes will be available at https://github.com/yuanchaosu/TCYB-nsckl.

As an unsupervised learning tool, autoencoder has been widely applied in many fields. In this letter, we propose a new robust unmixing algorithm that is based on stacked nonnegative sparse autoencoders (NNSAEs) for hyperspectral data with outliers and low signal-to-noise ratio. The proposed stacked autoencoders network contains two main steps. In the first step, a series of NNSAE is used to detect the outliers in the data. In the second step, a final autoencoder is performed for unmixing to achieve the endmember signatures and abundance fractions. By taking advantage from nonnegative sparse autoencoding, the proposed approach can well tackle problems with outliers and low noise-signal ratio. The effectiveness of the proposed method is evaluated on both synthetic and real hyperspectral data. In comparison with other unmixing methods, the proposed approach demonstrates competitive performance.

Transformer holds significance in deep learning (DL) research. Node embedding (NE) and positional encoding (PE) are usually two indispensable components in a Transformer. The former can excavate hidden correlations from the data, while the latter can store locational relationships between nodes. Recently, the Transformer has been applied for hyperspectral image (HSI) classification because the model can capture long-range dependencies to aggregate global features for representation learning. In an HSI, adjacent pixels tend to be homogeneous, while the NE does not identify the positional information of pixels. Therefore, PE is crucial for Transformers to understand locational relationships between pixels. However, in this area, most Transformer-based methods randomly generate PEs without considering their physical meaning, which leads to weak representations. This article proposes a new graph generative structure-aware Transformer (GraphGST) to solve the above-mentioned PE problem when implementing HSI classification. In our GraphGST, a new absolute PE (APE) is established to acquire pixels’ absolute positional sequences (APSs) and is integrated into the Transformer architecture. Moreover, a generative mechanism with self-supervised learning is developed to achieve cross-view contrastive learning (CL), aiming to enhance the representation learning of the Transformer. The proposed GraphGST model can capture local-to-global correlations, and the extracted APSs can complement the spectral features of pixels to assist in NE. Several experiments with real HSIs are conducted to evaluate the effectiveness of our GraphGST. The proposed method demonstrates very competitive performance compared with other state-of-the-art (SOTA) approaches. Our source codes will be provided in the following link https://github.com/yuanchaosu/TGRS-graphGST .

Hyperspectral unmixing is an important problem for remotely sensed data interpretation. It amounts at estimating the spectral signatures of the pure spectral constituents in the scene (endmembers) and their corresponding subpixel fractional abundances. Although the unmixing problem is inherently nonlinear (due to multiple scattering), the nonlinear unmixing of hyperspectral data has been a very challenging problem. This is because nonlinear models require detailed knowledge about the physical interactions between the sunlight scattered by multiple materials. In turn, bilinear mixture models (BMMs) can reach good accuracy with a relatively simple model for scattering. In this article, we develop a new BMM and a corresponding unsupervised unmixing approach which consists of two main steps. In the first step, a deep autoencoder is used to linearly estimate the endmember signatures and their associated abundance fractions. The second step refines the initial (linear) estimates using a bilinear model, in which another deep autoencoder (with a low-rank assumption) is adapted to model second-order scattering interactions. It should be noted that in our developed BMM model, the two deep autoencoders are trained in a mutually interdependent manner under the multitask learning framework, and the relative reconstruction error is used as the stopping criterion. The effectiveness of the proposed method is evaluated using both synthetic and real hyperspectral data sets. Our experimental results indicate that the proposed approach can reasonably estimate the nature of nonlinear interactions in real scenarios. Compared with other state-of-the-art unmixing algorithms, the proposed approach demonstrates very competitive performance.

Transformer is a powerful tool for capturing long-range dependencies and has shown impressive performance in hyperspectral image (HSI) classification. However, such power comes with a heavy memory footprint and huge computation burden. In this paper, we propose two types of lightweight self-attention modules (a channel lightweight multi-head self-attention module and a position lightweight multi-head self-attention module) to reduce both memory and computation while associating each pixel or channel with global information. Moreover, we discover that transformers are ineffective in explicitly extracting local and multi-scale features due to the fixed input size and tend to overfit when dealing with a small number of training samples. Therefore, a lightweight transformer (LiT) network, built with the proposed lightweight self-attention modules, is presented. LiT adopts convolutional blocks to explicitly extract local information in early layers and employs transformers to capture long-range dependencies in deep layers. Furthermore, we design a controlled multi-class stratified sampling strategy to generate appropriately sized input data, ensure balanced sampling, and reduce the overlap of feature extraction regions between training and test samples. With appropriate training data, convolutional tokenization, and lightweight transformers, LiT mitigates overfitting and enjoys both high computational efficiency and good performance. Experimental results on several HSI datasets verify the effectiveness of our design.