Automatic atmospheric correction for shortwave hyperspectral remote sensing data using a time-dependent deep neural network

Sun, Jian; Xu, Fangcao; Cervone, Guido; Gervais, Melissa; Wauthier, C.; Salvador, Mark

doi:10.1016/j.isprsjprs.2021.02.007

Cited by 18 publications

(11 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, Xu et al 8 employs a Convolutional Neural Network (CNN) and autoencoder model structure to perform atmospheric correction and target detection. Sun et al 9 uses similar methods to perform near-real-time atmospheric correction from time-dependent (video) training data. Similarly, in treating the atmospheric propagation of light as a noising step, Basener et al 10,11 use Gaussian Processes (GPs) and de-noising convolutional autoencoders to predict reflectance from observed radiance values with good success.…”

Section: Atmospheric Correction and Machine Learningmentioning

confidence: 99%

A neural differential equation formulation for modeling atmospheric effects in hyperspectral images

Koch

Forland

Doster

et al. 2023

Algorithms, Technologies, and Applications for Multispectral and Hyperspectral Imaging XXIX

View full text Add to dashboard Cite

Atmospheric correction is the process for removing atmospheric effects from spectral data; a necessary step for recovering salient spectral properties. The complex interactions between the atmosphere and light are dominated by absorbance and scattering physics. Existing methods for modeling atmospheric interactions typically rely on deep knowledge of relevant environmental conditions and high-fidelity numerical simulations of the governing physics in order to obtain accurate estimates of these effects. Additionally, existing approaches often require a subject matter expert for pre/post-processing of the data. Model-based approaches for removing atmospheric effects struggle in situations where such domain expertise is not available, and require significant human effort and computational power even when that expertise is available. In contrast, we propose a data-driven approach the uses Neural Differential Equations (NDEs) to accurately learn the interactions between electromagnetic radiation and the atmospheric without access to location specific environmental information. Once trained, the NDE can be applied bi-directionally; to apply or remove atmospheric effects. We demonstrate the effectiveness and utility of these techniques on an example multi-spectral scene.

show abstract

Section: Atmospheric Correction and Machine Learningmentioning

confidence: 99%

A neural differential equation formulation for modeling atmospheric effects in hyperspectral images

Koch

Forland

Doster

et al. 2023

Algorithms, Technologies, and Applications for Multispectral and Hyperspectral Imaging XXIX

View full text Add to dashboard Cite

show abstract

“…Therefore, we used the ENVI RPC model to perform geometric corrections for these deformations. Furthermore, for atmospheric correction, the ENVI FLAASH module was used to eliminate the absorption and scattering of electromagnetic waves from sunlight and ground objects by the atmosphere (Aquino et al 2018, Sun et al 2021. Seven kinds of visible vegetation indexes were extracted (Du & Noguchi 2016, Choi et al 2020, Boonpook et al 2021: the visible-band difference vegetation index (VDVI -eqn.…”

Section: Iforest -Biogeosciences and Forestrymentioning

confidence: 99%

A geographically weighted deep neural network model for research on the spatial distribution of the down dead wood volume in Liangshui National Nature Reserve (China)

Sun¹,

Ao²,

Jia³

et al. 2021

iForest

View full text Add to dashboard Cite

In natural forest ecosystems, there is often abundant down dead wood (DDW) due to wind disasters, which greatly changes the size and structure of forests. Accurately determining the DDW volume (DDWV) is crucial for sustaining forest management, predicting the dynamic changes in forest resources and assessing the risks of natural disasters or disturbances. However, existing models cannot accurately express the significant spatial nonstationarity or complexity in their spatial relationships. To this end, we established a geographically weighted deep neural network (GWDNN) model that constructs a spatially weighted neural network (SWNN) through geographic location data and builds a neural network through stand factors and remote sensing factors to improve the interpretability of the spatial model of DDWV. To verify the effectiveness of this method, using 2019 data from Liangshui National Nature Reserve, we compared model fit, predictive ability and residual spatial autocorrelation among the GWDNN model and four other spatial models: an ordinary least squares (OLS) model, a linear mixed model (LMM), a geographically weighted regression (GWR) model and a deep neural network (DNN) model. The experimental results show that the GWDNN model is far superior to the other four models according to various indicators; the coefficient of determination R 2 , root mean square error (RMSE), mean absolute error (MAE), Moran's I and Z-statistic values of the GWDNN model were 0.95, 1.05, 0.77, -0.01 and -0.06, respectively. In addition, compared with the other models, the GWDNN model can more accurately depict local spatial variations and details of the DDWV in Liangshui National Nature Reserve.

show abstract

“…11 Recent works with various machine learning approaches have indicated similar or improved performance to state-of-the-art in-scene models. [13][14][15][16][17] While forward modeling of atmospheric correction can be difficult, the reverse process can be modeled accurately and robustly via physics-based radiative transport (RT) modeling. [18][19][20] In RT models, a line-of-sight (LoS) is constructed from the target (i.e.-surface material) to the observer (i.e.-sensor platform).…”

Section: Introductionmentioning

confidence: 99%

Diffusion learning for atmospheric correction

Stelter,

Sundberg

2023

Imaging Spectrometry XXVI: Applications, Sensors, and Processing

View full text Add to dashboard Cite

Diffusion learning is a generative technique commonly applied to create new images or audio directly from sampled noise. The machine learning approach works by applying degrading signals, such as noise, continuously and learning the denoising process with a neural network. In place of noise, other operations can be performed, such as the addition of atmosphere effects using a physics-based radiative transport code. In this paper, we explore coupling the MODTRAN software to a diffusion learning framework. The goal is to apply atmosphere systematically for a variety of reflective surfaces and use diffusion learning to train models for atmospheric correction. To achieve this, we generate a scoped dataset containing randomized Lambertian surfaces with differing solar illumination and surface angles.

show abstract

Automatic atmospheric correction for shortwave hyperspectral remote sensing data using a time-dependent deep neural network

Cited by 18 publications

References 19 publications

A neural differential equation formulation for modeling atmospheric effects in hyperspectral images

A neural differential equation formulation for modeling atmospheric effects in hyperspectral images

A geographically weighted deep neural network model for research on the spatial distribution of the down dead wood volume in Liangshui National Nature Reserve (China)

Diffusion learning for atmospheric correction

Contact Info

Product

Resources

About