Abstract:Hyperspectral imaging,which measures electromagnetic energy scattered in the instantaneous field view in hundreds of bands,contributes significantly to earth observation.As the electromagnetic spectrum covers visible,near infrared,and shortwave infrared spectral bands,it holds abundant spectral information.The subtle discriminative spectral characteristics also make material identification possible.However,the signals recorded by low-resolution hyperspectral sensors from certain pixels under a complex background are a mixture of substances.To improve the accuracy of classification and subpixel object detection,Hyperspectral Unmixing (HU) represents an important solution,especially in the remote sensing area.Mixed pixels comprise a few materials called endmembers.The fraction of each endmember in a pixel is called abundance.Given the hyperspectral data exhibiting high dimension,special spectral correlation,and huge quantity,the determination of the number of endmembers can be regarded as dimensionality reduction.The objective of HU is to estimate and extract the spectral signatures and abundance of endmembers.In this study,mixing models are systematically discussed.On the basis of mixing models,the core problem of HU is presented,which mainly comprises three aspects:estimation of number of endmembers,endmember extraction,and abundance fraction estimation.For each aspect,we summarize the basic theory and development of the HU processing methods.Through the basic theory model,this work performs an initial classification of each aspect.The physical or mathematical problems involved are also discussed along with classical and state-of-the-art methods used to address the problems.Furthermore,a general overview of each category is provided,and a comprehensive analysis of the advantages and disadvantages of classic algorithms is performed.Through the comparison of different methods,this study offers a perspective on the potential and emerging challenges in the process of HU.The research on HU has been continuous.Even though extensive processing results have been achieved and numerous analyses have been performed,obstacles remain and call for solutions.Five challenges are identified in this work:curse of dimensionality,establishment of a high-accuracy mixed inversion model,endmember variation,establishment of remote sensing operational product,and development of algorithms for real-time processing.According to each challenge,the study performs further analysis and presents several remarks.Then,suggestions with regard to future research directions are offered.This study provides a brief overview of HU.With the development of hyperspectral imaging,the performance of imaging data has undergone a qualitative leap.However,different application requires different data types.That is,the most suitable data,and not the data with the highest index,should be adopted to solve the problem.Considering the future development trend of HU,this work suggests that new methods be combined with the demands of users to promote the operational application of HU and support hyperspectral remote sensing engineering applications.

UnmixDiff: Unmixing-based Diffusion Model for Hyperspectral Image Synthesis

Realistic Mixing Miniature Scene Hyperspectral Unmixing: from Benchmark Datasets to Autonomous Unmixing

Unmixing Diffusion for Self-Supervised Hyperspectral Image Denoising

Unmixing Before Fusion: A Generalized Paradigm for Multi-Source-Based Hyperspectral Image Synthesis

Diffusion-Model-Based Hyperspectral Unmixing Using Spectral Prior Distribution

Diverse Hyperspectral Remote Sensing Image Synthesis with Diffusion Models

From the Abundance Perspective: Multi-modal Scene Fusion-Based Hyperspectral Image Synthesis

Hyperspectral Unmixing Network Accounting for Spectral Variability Based on a Modified Scaled and a Perturbed Linear Mixing Model

From the Abundance Perspective

Spectral and Spatial Complexity-Based Hyperspectral Unmixing

Correntropy-Based Spatial-Spectral Robust Sparsity-Regularized Hyperspectral Unmixing

Spectral Mixture Model Inspired Network Architectures for Hyperspectral Unmixing

Pixel-to-Abundance Translation: Conditional Generative Adversarial Networks Based on Patch Transformer for Hyperspectral Unmixing

SWDiff: Stage-wise Hyperspectral Diffusion Model for Hyperspectral Image Classification

A Multimodal Hyperspectral Unmixing Method under Spectral Variability

Sparse Unmixing of Hyperspectral Data with Bandwise Model

SpectralDiff: A Generative Framework for Hyperspectral Image Classification with Diffusion Models

UnmixingSR: Material-aware Network with Unsupervised Unmixing as Auxiliary Task for Hyperspectral Image Super-resolution

Mixing Support Detection-Based Alternating Direction Method of Multipliers for Sparse Hyperspectral Image Unmixing

Research Progress on Unmixing of Hyperspectral Remote Sensing Imagery