Deep learning-based hyperspectral target detection (HTD) methods are limited by the lack of prior information. Self-supervised learning is a kind of unsupervised learning, which mainly mines its own self-supervised information from unlabeled data. By training the model with such constructed valid posterior information, a valuable representation model can be learned and can get rid of the dependence of deep models on prior information. To this end, this paper proposes a self-supervised spectral-level contrastive learning-based HTD (SCLHTD) method to train a model with spectral difference discrimination capability for HTD in a self-supervised manner. Firstly, the hyperspectral images to be detected are sampled in odd and even bands, and the obtained band subsets are then used to train the corresponding adversarial convolutional autoencoders. Feature extraction part of the trained encoder is then used as the data augmentation function, where the positive and negative pairs are constructed through data augmentation, and the backbone is used to extract the representative vectors of the augmented samples. Secondly, the representative vectors are mapped to the spectral contrast space using spectral contrastive head, where the similarity and dissimilarity of spectra are learned by maximizing the similarity of positive pairs while minimizing the similarity of negative pairs, so that the backbone can discriminate spectral differences. Finally, aiming at suppressing the background, edge-preserving filters are used in conjunction with space information to process the detection results acquired by utilizing spectrum information via cosine similarity to generate the final detection results. Experimental results illustrate that the proposed SCLHTD method can achieve superior performances for HTD.

Hyperspectral image (HSI) provided with a substantial number of correlated bands causes calculation consumption and an undesirable “dimension disaster” problem for the classification. Band selection (BS) is an effective measure to reduce the information redundancy with the physics spectrum preserved for HSI. Although the existing BS methods have achieved noticeable progress, the correlation between neighbor bands still needs to be mined deeply for an effective selection criterion. This article proposes a BS approach to collecting the discriminative band subset for HSI classification (HSIC), which adopts the self-supervised learning paradigm to implement the BS by the auxiliary spectrum rebuilding (SR) task. In specific, we utilized a convolutional neural network (CNN) and a graph convolutional network (GCN) for the spectral–spatial feature extraction. Next, GCN and CNN are developed for the refinement of the band correlation sequentially. Afterward, the selected bands in terms of the acquired correlation are fed into the presented self-supervised SR network for spectral reconstruction. Simultaneously, the proposed architecture completed the selection with the optimization of the band reconstruction by a defined loss function. In this way, we supply substitution for selection criterion and path searching through the end-to-end framework. The extensive experimental results and analysis demonstrated that the proposed hybrid architecture provided a competitive band subset for the classification, and the accuracies with different types of classifiers are more effective than the compared BS methods.

Hyperspectral target detection plays a pivotal role in various civil and military applications. Although recent advancements in deep learning have largely embraced supervised learning approaches, they often hindered by the limited availability of labeled data. Unsupervised learning, therefore, emerges as a promising alternative, yet its potential has not been fully realized in current methodologies. This article proposes an innovative unsupervised learning framework employing a momentum contrastive learning-based transformer network specifically tailored for hyperspectral target detection. The proposed approach innovatively combines transformer-based encoder and momentum encoder networks to enhance feature extraction capabilities, adeptly capturing both local spectral details and long-range spectral dependencies through the novel overlapping spectral patch embedding and a cross-token feedforward layer. This dual-encoder design significantly improves the model's ability to discern relevant spectral features amidst complex backgrounds. Through unsupervised momentum contrastive learning, a dynamically updated queue of negative sample features is utilized so that the model can demonstrate superior spectral discriminability. This is further bolstered by a unique background suppression mechanism leveraging nonlinear transformations of cosine similarity detection results, with two nonlinearly pull-up operations, significantly enhancing target detection sensitivity, where the nonlinearly operations are the exponential function with its normalization and the power function with its normalization, respectively. Comparative analysis against seven state-of-the-art hyperspectral target detection methods across four real hyperspectral images demonstrates the effectiveness of the proposed method for hyperspectral target detection, with an increase in detection accuracy and a competitive computational efficiency. An extensive ablation study further validates the critical components of the proposed framework, confirming its comprehensive capability and applicability in hyperspectral target detection scenarios.

Coal fires are a common and serious problem in most coal-bearing countries. Thus, it is very important to monitor changes in coal fires. Remote sensing provides a useful technique for investigating coal fields at a large scale and for detecting coal fires. In this study, the spreading direction of a coal fire in the Wuda Coal Field (WCF), northwest China, was analyzed using multi-temporal Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper (ETM+) thermal infrared (TIR) data. Using an automated method and based on the land surface temperatures (LST) that were retrieved from these thermal data, coal fires related to thermal anomalies were identified; the locations of these fires were validated using a coal fire map (CFM) that was developed via field surveys; and the cross-validation of the results was also carried out using the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) thermal infrared images. Based on the results from longtime series of satellite TIR data set, the spreading directions of the coal fires were determined and the coal fire development on the scale of the entire coal field was predicted. The study delineated the spreading direction using the results of the coal fire dynamics analysis, and a coal fire spreading direction map was generated. The results showed that the coal fires primarily spread north or northeast in the central part of the WCF and south or southwest in the southern part of the WCF. In the northern part of the WCF, some coal fires were spreading north, perhaps coinciding with the orientation of the coal belt. Certain coal fires scattered in the northern and southern parts of the WCF were extending in bilateral directions. A quantitative analysis of the coal fires was also performed; the results indicate that the area of the coal fires increased an average of approximately 0.101 km² per year.

Despite continuing to progress in hyperspectral image classification (HSIC) based on deep learning, the classification accuracy is limited to furtherly improve in the absence of labeled samples. To address this issue, the metric-based prototypical networks for few-shot learning have enjoyed widespread popularity. However, the conventional prototypical networks are vulnerable to the selected examples and fail to accomplish representative predictions for the prototypes in complicated situations. In this paper, we propose a multi-view calibrated prototype-learning framework for few-shot HSIC, which consists of three rectified strategies from different views to improve the robustness of prototypes in the embedding space. Specifically, the calibrated aggregation network is the first presented to calibrate the representations with local patches aggregation for the enhancement of the prototypes. Moreover, to improve the compactness of the intraclass expression, the calibrated metric learning with regularization terms is designed to strengthen the discrimination of the prototypes. Furthermore, we calibrate the feature distribution of supervised samples by transferring statistical knowledge to eliminate the local bias in the test phase. The extensive experimental results and analysis of three hyperspectral image datasets demonstrate the superiority of the proposed architecture compared with other advanced methods.

Land surface temperature (LST) plays a crucial role in the physical and chemical processes of the land-atmosphere system. Remote sensing technology has greatly advanced the measurement of thermal infrared land surface temperature (TIR LST), which is the most widely utilized surface temperature product. However, cloud cover and mist often cause significant data loss in TIR LST. To address this issue and reconstruct the MYD11A1 LST under cloudy conditions, this study proposes an all-weather LST generation method based on the eXtreme Gradient Boosting (XGBoost) model. This method incorporates spatial-seamless passive microwave land surface temperature (PMW LST) to capture the nonlinear relationship between TIR LST and other variables. Compared to the MYD11A1 LST, the generated all-weather LST provides continuous spatial texture information without a significant boundary reconstruction effect, improving the accuracy of spatiotemporal variations in LST in China. In situ validation demonstrated the high accuracy of the generated all-weather LST, with mean R ² , Bias and unbiased root mean square error (ubRMSE) of 0.96 (0.91), 1.08 K (3.61 K) and 2.92 K (4.54 K) under clear (cloudy) daytime conditions, and 0.92 (0.95), -0.93 K (-2.96 K) and 3.09 K (3.04 K) under clear (cloudy) nighttime conditions. These results indicate the feasibility and reasonableness of the all-weather LST generation method developed in this study and affirm its ability to generate highly accurate all-weather LST.