The majority of existing hyperspectral anomaly detection (HAD) methods use
the low-rank representation (LRR) model to separate the background and anomaly
components, where the anomaly component is optimized by handcrafted sparse
priors (e.g., $\ell_{2,1}$-norm). However, this may not be ideal since they
overlook the spatial structure present in anomalies and make the detection
result largely dependent on manually set sparsity. To tackle these problems, we
redefine the optimization criterion for the anomaly component in the LRR model
with a self-supervised network called self-supervised anomaly prior (SAP). This
prior is obtained by the pretext task of self-supervised learning, which is
customized to learn the characteristics of hyperspectral anomalies.
Specifically, this pretext task is a classification task to distinguish the
original hyperspectral image (HSI) and the pseudo-anomaly HSI, where the
pseudo-anomaly is generated from the original HSI and designed as a prism with
arbitrary polygon bases and arbitrary spectral bands. In addition, a
dual-purified strategy is proposed to provide a more refined background
representation with an enriched background dictionary, facilitating the
separation of anomalies from complex backgrounds. Extensive experiments on
various hyperspectral datasets demonstrate that the proposed SAP offers a more
accurate and interpretable solution than other advanced HAD methods.

该研究提出了一种基于自监督网络的自监督异常先验（SAP），将低秩表示模型中的异常分量优化为了更好地适应高光谱异常，同时提出了一种双纯化策略来从复杂背景中分离异常，并通过实验验证了该方法在各种高光谱数据集上的准确性和可解释性。

自监督异常先验的高光谱异常检测

Hyperspectral Anomaly Detection with Self-Supervised Anomaly Prior

Remote sensing data has been widely used for various Earth Observation (EO)
missions such as land use and cover classification, weather forecasting,
agricultural management, and environmental monitoring. Most existing remote
sensing data-based models are based on supervised learning that requires large
and representative human-labelled data for model training, which is costly and
time-consuming. Recently, self-supervised learning (SSL) enables the models to
learn a representation from orders of magnitude more unlabelled data. This
representation has been proven to boost the performance of downstream tasks and
has potential for remote sensing applications. The success of SSL is heavily
dependent on a pre-designed pretext task, which introduces an inductive bias
into the model from a large amount of unlabelled data. Since remote sensing
imagery has rich spectral information beyond the standard RGB colour space, the
pretext tasks established in computer vision based on RGB images may not be
straightforward to be extended to the multi/hyperspectral domain. To address
this challenge, this work has designed a novel SSL framework that is capable of
learning representation from both spectra-spatial information of unlabelled
data. The framework contains two novel pretext tasks for object-based and
pixel-based remote sensing data analysis methods, respectively. Through two
typical downstream tasks evaluation (a multi-label land cover classification
task on Sentienl-2 multispectral datasets and a ground soil parameter retrieval
task on hyperspectral datasets), the results demonstrate that the
representation obtained through the proposed SSL achieved a significant
improvement in model performance.

本研究设计了一种新的自监督学习框架，可以从未标记数据的光谱 - 空间信息中学习表示，并通过两个典型的下游任务评估证明，所提出的 SSL 获得了模型性能的显著提高。

一种通用的自监督学习 (SSL) 框架，用于从未标记的遥感图像的光谱 - 空间特征中进行表征学习

A generic self-supervised learning (SSL) framework for representation  learning from spectra-spatial feature of unlabeled remote sensing imagery

High spectral resolution imagery of the Earth's surface enables users to
monitor changes over time in fine-grained scale, playing an increasingly
important role in agriculture, defense, and emergency response. However, most
current algorithms are still confined to describing local features and fail to
incorporate a global perspective, which limits their ability to capture
interactions between global features, thus usually resulting in incomplete
change regions. In this paper, we propose a Global Multi-head INteractive
self-attention change Detection network (GlobalMind) to explore the implicit
correlation between different surface objects and variant land cover
transformations, acquiring a comprehensive understanding of the data and
accurate change detection result. Firstly, a simple but effective Global Axial
Segmentation (GAS) strategy is designed to expand the self-attention
computation along the row space or column space of hyperspectral images,
allowing the global connection with high efficiency. Secondly, with GAS, the
global spatial multi-head interactive self-attention (Global-M) module is
crafted to mine the abundant spatial-spectral feature involving potential
correlations between the ground objects from the entire rich and complex
hyperspectral space. Moreover, to acquire the accurate and complete
cross-temporal changes, we devise a global temporal interactive multi-head
self-attention (GlobalD) module which incorporates the relevance and variation
of bi-temporal spatial-spectral features, deriving the integrate potential same
kind of changes in the local and global range with the combination of GAS. We
perform extensive experiments on five mostly used hyperspectral datasets, and
our method outperforms the state-of-the-art algorithms with high accuracy and
efficiency.

本文提出了一种名为 GlobalMind 的改变检测网络，它采用了全球多头交互自注意力机制，结合全局轴向分割和全局时变交互多头注意力模块，用于超光谱图像，有效提高了跨时空变化检测精度和效率。