FY-3G is the first polar-orbiting satellite equipped with a precipitation measurement radar(PMR)operating at Ku-andKa-band frequencies in China.In this study,we compare the reflectivity data from the FY-3G PMR Ku prod...FY-3G is the first polar-orbiting satellite equipped with a precipitation measurement radar(PMR)operating at Ku-andKa-band frequencies in China.In this study,we compare the reflectivity data from the FY-3G PMR Ku product and groundbasedradars(GRs)during 2024.Also,the FY-3G PMR is used as a third-party reference to evaluate the reflectivityconsistency among different GRs.The FY-3G PMR and GRs share similarities in their general distribution,characteristics,and intensity of reflectivity in strong precipitation cloud systems,though the former presents less detailed system structure.Systematic deviations between the FY-3G PMR and GRs and between GRs are comparable,albeit the reflectivity of the FY-3G PMR is generally slightly stronger than that of GRs(especially X-band GRs),with a mean bias ranging from 0.7 to 1.7dB.S-band GRs exhibit the smallest systematic deviation(STD=3.09 dB)from the FY-3G PMR,whereas the X-band GRsshow the largest(STD=3.61 dB),indirectly indicating the highest internal consistency among S-band GRs and the lowestamong X-band GRs.Besides,both S-and C-band GRs display similar deviations when paired with the FY-3G PMR as wellas when paired with their adjacent S/C-band GRs,suggesting good consistency between these two bands.In contrast,XbandGRs exhibit relatively poor consistency with S-band GRs and the FY-3G PMR,showing a deviation ranging from 3.0to 4.6 dB.展开更多
Cemented paste backfill(CPB)is a technology that achieves safe mining by filling the goaf with waste rocks,tailings,and other materials.It is an inevitable choice to deal with the development of deep and highly diffic...Cemented paste backfill(CPB)is a technology that achieves safe mining by filling the goaf with waste rocks,tailings,and other materials.It is an inevitable choice to deal with the development of deep and highly difficult mines and meet the requirements of environmental protection and safety regulations.It promotes the development of a circular economy in mines through the development of lowgrade resources and the resource utilization of waste,and extends the service life of mines.The mass concentration of solid content(abbreviated as“concentration”)is a critical parameter for CPB.However,discrepancies often arise between the on-site measurements and the pre-designed values due to factors such as groundwater inflow and segregation within the goaf,which cannot be evaluated after the solidification of CPB.This paper innovatively provides an in-situ non-destructive approach to identify the real concentration of CPB after curing for certain days using hyperspectral imaging(HSI)technology.Initially,the spectral variation patterns under different concentration conditions were investigated through hyperspectral scanning experiments on CPB samples.The results demonstrate that as the CPB concentration increases from 61wt%to 73wt%,the overall spectral reflectance gradually increases,with two distinct absorption peaks observed at 1407 and 1917 nm.Notably,the reflectance at 1407 nm exhibited a strong linear relationship with the concentration.Subsequently,the K-nearest neighbors(KNN)and support vector machine(SVM)algorithms were employed to classify and identify different concentrations.The study revealed that,with the KNN algorithm,the highest accuracy was achieved when K(number of nearest neighbors)was 1,although this resulted in overfitting.When K=3,the model displayed the optimal balance between accuracy and stability,with an accuracy of 95.03%.In the SVM algorithm,the highest accuracy of 98.24%was attained with parameters C(regularization parameter)=200 and Gamma(kernel coefficient)=10.A comparative analysis of precision,accuracy,and recall further highlighted that the SVM provided superior stability and precision for identifying CPB concentration.Thus,HSI technology offers an effective solution for the in-situ,non-destructive monitoring of CPB concentration,presenting a promising approach for optimizing and controlling CPB characteristic parameters.展开更多
This review summarizes studies of hydrothermal alteration minerals at the Qiucun gold deposit in southeastern China and focuses on characterization and mapping of the deposit using hyperspectral remote sensing.The dep...This review summarizes studies of hydrothermal alteration minerals at the Qiucun gold deposit in southeastern China and focuses on characterization and mapping of the deposit using hyperspectral remote sensing.The deposit exhibits multistage fluid-rock interaction,as evidenced by systematic alteration assemblages,including silicification,sericitization by white micas,the development of argillaceous clays,variable chloritization,and locally significant carbonate alteration.We describe the genetic importance of such mineral groups and emphasize their diagnostic Visible and Near-Infrared to Short-Wave Infrared(VNIR-SWIR)spectral signatures,especially Al-OH,Mg-OH/Fe-OH,and CO3 absorption bands,which make it possible to distinguish between minerals,not to mention the fact that,in some instances,compositional trends may be predicted.This review’s methodological advances are discussed beginning with data collection at satellite,airborne,and ground levels,proceeding to processing procedures,such as atmospheric and topographic correction,and culminating in spectral analysis,including continuum removal,spectral matching,and unmixing/classification techniques.An integrated study of hyperspectral findings reveals that alteration minerals develop spatially coherent zones that are strongly controlled by fault/fracture structures and host-rock reactivity,producing proximal silicification/sericitization cores and larger silicified/larcenies of argillaceous rocks owing to diverse apex coverings of carbonate.This should be combined with petrography and geochemistry to address overprinting,mixed pixels,and surface weathering,and to couple mineral maps with ore-forming processes.The review finds that hyperspectral remote sensing offers a solid modeling platform for the deposit-scale alteration at Qiucun and other hydrothermal gold systems,and outlines the directions for future research to integrate quantitatively and more threedimensional alteration characterization.展开更多
Also known as imaging spectroscopy,hyperspectral remote sensing is becoming a key technology for ecosystem and natural resource management sustainability.Hyperspectral observations can be used to measure tens to hundr...Also known as imaging spectroscopy,hyperspectral remote sensing is becoming a key technology for ecosystem and natural resource management sustainability.Hyperspectral observations can be used to measure tens to hundreds of narrow bands of reflected radiation to resolve diagnostic absorption bands and spectral shape variations associated with vegetation pigments,water status of the canopy,biochemical composition,mineralogies,and organic matter of the soil,and water quality constituents of aquatic water.These abilities allow one to make a transition between the descriptive mapping and the functional monitoring,the anticipation of stress and disturbance early,and the more accurate attribution of environmental change.This summary encompasses improvements on the entire sensor-to-product pipeline,including field and UAV(Unmanned Aerial Vehicle)system platform developments,airborne campaign and spaceborne mission developments,calibration and analysis-ready preprocessing improvements,empirical learning methodology improvements,radiative transfer-based inversion method,spectral unmixing,deep learning,and hybrid physics-machine learning.We underline the increased importance of the combination of data with LiDAR(Light Detection and Ranging),SAR(Synthetic Aperture Radar),and thermal features aimed at decreasing the level of ambiguity and enhancing operational resilience.Applications based on decision are evaluated in terms of biodiversity and habitat evaluation,vegetation functionality and restoration,stress and disturbance,sustainable agricultural production,inland water quality and coastal water quality,land degradation and soil status,and environmental impact assessment.Inhibiting factors to operational adoption have always been perceived to be domain shift by region,season,and sensor,ground truth and validation,mixed pixels and scale mismatch,preprocessing sensitivities,and desirable uncertainty quantification and product output that is interpretable.We conclude with the scalability,sustainability,service priorities,such as harmonization standards,representative benchmarking,uncertainty-aware delivery,and co-design of stakeholders.展开更多
Hyperspectral remote sensing has emerged as a transformative technology for sustainable natural resource management by providing unprecedented insight into the biochemical,biophysical,and compositional properties of E...Hyperspectral remote sensing has emerged as a transformative technology for sustainable natural resource management by providing unprecedented insight into the biochemical,biophysical,and compositional properties of Earth’s surface.The high spectral resolution of hyperspectral sensors allows a very specific discrimination of materials,monitoring of environmental stress at a very early stage,and provides quantitative retrieval of ecological and geochemical parameters in a wide range of landscapes.The booming technology in sensor design,machine learning,spectral unmixing,and multi-sensor data fusion has further improved the analysis potential and application of imaging spectroscopy to a large extent.This paper involves a discussion of the oversight of such technological advances and the manner in which they are utilized in the principal fields that include forestry,agriculture,water,mineral exploration,and coastal ecosystems.Case studies allow us to identify the potential practical consequences of both spaceborne and unmanned aerial vehicles(UAV)-based hyperspectral systems and AI-based workflows that can be used to aid in more efficient and accurate environmental review.Even though the issues associated with data volume,atmospheric impacts,lack of uniformity in the calibration process,and socioeconomic limits continue to exist,the new technology in sensor miniaturization,cloud computing,and artificial intelligence indicates a fast-changing environment.All these developments make hyperspectral remote sensing a key instrument in solving global sustainability problems and evidence-based management of natural resources in an evolving world.展开更多
Hyperspectral images(HSIs)are susceptible to various noise interferences during the imaging process,leading to degraded image quality and affecting the accuracy of information extraction.Efficient denoising methods ar...Hyperspectral images(HSIs)are susceptible to various noise interferences during the imaging process,leading to degraded image quality and affecting the accuracy of information extraction.Efficient denoising methods are crucial for ensuring the accuracy of subsequent remote sensing analysis and applications.In view of the characteristics of hyperspectral image data,such as high dimensionality,strong spectral correlation,and high computational complexity,a threedimensional visual state space U-Net(VSSU3D)was proposed in this paper.By introducing a visual state space module into the traditional U-Net,and combining the spatial-spectral characteristics of hyperspectral images with the core idea of the Mamba model,targeted optimizations wereachieved to effectively model global information dependencies while reducing computational complexity.Additionally,a simplified channel attention module was embedded between the encoder and decoder to enhance cross-scale feature fusion capabilities.Experimental results on multiple publicly available hyperspectral image datasets demonstrated that VSSU3D achieved denoising performance comparable to or superior to existing advanced methods,which verified its effectiveness.展开更多
Agricultural remote sensing has been developed and applied in monitoring soil,crop growth,weed infestation,insects,diseases and water status in farm fields to provide data and information to guide agricultural managem...Agricultural remote sensing has been developed and applied in monitoring soil,crop growth,weed infestation,insects,diseases and water status in farm fields to provide data and information to guide agricultural management practices.Precision agriculture has been implemented through prescription mapping of crop fields at different scales with the data remotely sensed from space-borne,airborne and ground-based platforms.Ground-based remote sensing techniques offer portability,flexibility and controllability in applications for precision agriculture.In weed management,crop injury from off-target herbicide spray drift and herbicide resistance in weeds are two important issues.For precision weed management,ground-based hyperspectral remote sensing techniques were developed for detection of crop injury from dicamba and differentiation between glyphosate resistant and sensitive weeds.This research presents the techniques for ground-based hyperspectral remote sensing for these two applications.Results illustrate the advantages of ground-based hyperspectral remote sensing for precision weed management.展开更多
The Global Precipitation Measurement(GPM)dual-frequency precipitation radar(DPR)products(Version 07A)are employed for a rigorous comparative analysis with ground-based operational weather radar(GR)networks.The reflect...The Global Precipitation Measurement(GPM)dual-frequency precipitation radar(DPR)products(Version 07A)are employed for a rigorous comparative analysis with ground-based operational weather radar(GR)networks.The reflectivity observed by GPM Ku PR is compared quantitatively against GR networks from CINRAD of China and NEXRAD of the United States,and the volume matching method is used for spatial matching.Additionally,a novel frequency correction method for all phases as well as precipitation types is used to correct the GPM Ku PR radar frequency to the GR frequency.A total of 20 GRs(including 10 from CINRAD and 10 from NEXRAD)are included in this comparative analysis.The results indicate that,compared with CINRAD matched data,NEXRAD exhibits larger biases in reflectivity when compared with the frequency-corrected Ku PR.The root-mean-square difference for CINRAD is calculated at 2.38 d B,whereas for NEXRAD it is 3.23 d B.The mean bias of CINRAD matched data is-0.16 d B,while the mean bias of NEXRAD is-2.10 d B.The mean standard deviation of bias for CINRAD is 2.15 d B,while for NEXRAD it is 2.29 d B.This study effectively assesses weather radar data in both the United States and China,which is crucial for improving the overall consistency of global precipitation estimates.展开更多
The cleanliness of seed cotton plays a critical role in the pre-treatment of cotton textiles,and the removal of impurity during the harvesting process directly determines the quality and market value of cotton textile...The cleanliness of seed cotton plays a critical role in the pre-treatment of cotton textiles,and the removal of impurity during the harvesting process directly determines the quality and market value of cotton textiles.By fusing band combination optimization with deep learning,this study aims to achieve more efficient and accurate detection of film impurities in seed cotton on the production line.By applying hyperspectral imaging and a one-dimensional deep learning algorithm,we detect and classify impurities in seed cotton after harvest.The main categories detected include pure cotton,conveyor belt,film covering seed cotton,and film adhered to the conveyor belt.The proposed method achieves an impurity detection rate of 99.698%.To further ensure the feasibility and practical application potential of this strategy,we compare our results against existing mainstream methods.In addition,the model shows excellent recognition performance on pseudo-color images of real samples.With a processing time of 11.764μs per pixel from experimental data,it shows a much improved speed requirement while maintaining the accuracy of real production lines.This strategy provides an accurate and efficient method for removing impurities during cotton processing.展开更多
Hyper Spectral Image Super-Resolution(HSI-SR) has gained significant attention in recent years due to its potential applications.However,the challenge of obtaining high-resolution hyperspectral images is compounded by...Hyper Spectral Image Super-Resolution(HSI-SR) has gained significant attention in recent years due to its potential applications.However,the challenge of obtaining high-resolution hyperspectral images is compounded by limitations in sensor resolution and the high dimensionality of spectral data.Traditional approaches,including interpolation-based methods and sparse representation techniques,often struggle to capture the intricate spectral-spatial dependencies in hyperspectral images.To address these limitations,this study proposes a Hadamard Self-Attention Network(HSAN) for fusing a High-resolution Multispectral Image(Hr-MSI) and a Low-resolution Hyper Spectral Image(Lr-HSI),achieving HSI-SR for obtaining a High-resolution Hyper Spectral Image(Hr-HSI).The core of HSAN is a new Hadamard self-attention mechanism that can be more efficient than traditional dot-product attention because it avoids matrix multiplications and softmax operations.Considering that deep learning-based data fusion typically entails a significant computational and storage burden,this new approach can be integrated with convolutional layers to form an unsupervised lightweight network,which significantly reduces dependence on computational resources.Experimental results across four datasets validate the effectiveness and advantages of HSAN,compared with state-of-the-art approaches.The source code will be available at https://github.com/zxnhkdm/HSAN.展开更多
In this study,eight different varieties of maize seeds were used as the research objects.Conduct 81 types of combined preprocessing on the original spectra.Through comparison,Savitzky-Golay(SG)-multivariate scattering...In this study,eight different varieties of maize seeds were used as the research objects.Conduct 81 types of combined preprocessing on the original spectra.Through comparison,Savitzky-Golay(SG)-multivariate scattering correction(MSC)-maximum-minimum normalization(MN)was identified as the optimal preprocessing technique.The competitive adaptive reweighted sampling(CARS),successive projections algorithm(SPA),and their combined methods were employed to extract feature wavelengths.Classification models based on back propagation(BP),support vector machine(SVM),random forest(RF),and partial least squares(PLS)were established using full-band data and feature wavelengths.Among all models,the(CARS-SPA)-BP model achieved the highest accuracy rate of 98.44%.This study offers novel insights and methodologies for the rapid and accurate identification of corn seeds as well as other crop seeds.展开更多
Imaging hyperspectral technology has distinctive advantages of non-destructive and non-contact measurement,and the integration of spectral and spatial data.These characteristics present new methodologies for intellige...Imaging hyperspectral technology has distinctive advantages of non-destructive and non-contact measurement,and the integration of spectral and spatial data.These characteristics present new methodologies for intelligent geological sensing in tunnels and other underground engineering projects.However,the in situ acquisition and rapid classification of hyperspectral images in underground still faces great challenges,including the difficulty in obtaining uniform hyperspectral images and the complexity of deploying sophisticated models on mobile platforms.This study proposes an intelligent lithology identification method based on partition feature extraction of hyperspectral images.Firstly,pixel-level hyperspectral information from representative lithological regions is extracted and fused to obtain rock hyperspectral image partition features.Subsequently,an SG-SNV-PCA-DNN(SSPD)model specifically designed for optimizing rock hyperspectral data,performing spectral dimensionality reduction,and identifying lithology is integrated.In an experimental study involving 3420 hyperspectral images,the SSPD identification model achieved the highest accuracy in the testing set,reaching 98.77%.Moreover,the speed of the SSPD model was found to be 18.5%faster than that of the unprocessed model,with an accuracy improvement of 5.22%.In contrast,the ResNet-101 model,used for point-by-point identification based on non-partitioned features,achieved a maximum accuracy of 97.86%in the testing set.In addition,the partition feature extraction methods significantly reduce computational complexity.An objective evaluation of various models demonstrated that the SSPD model exhibited superior performance,achieving a precision(P)of 99.46%,a recall(R)of 99.44%,and F1 score(F1)of 99.45%.Additionally,a pioneering in situ detection work was carried out in a tunnel using underground hyperspectral imaging technology.展开更多
In recent years,camouflage technology has evolved from single-spectral-band applications to multifunctional and multispectral implementations.Hyperspectral imaging has emerged as a powerful technique for target identi...In recent years,camouflage technology has evolved from single-spectral-band applications to multifunctional and multispectral implementations.Hyperspectral imaging has emerged as a powerful technique for target identification due to its capacity to capture both spectral and spatial information.The advancement of imaging spectroscopy technology has significantly enhanced reconnaissance capabilities,offering substantial advantages in camouflaged target classification and detection.However,the increasing spectral similarity between camouflaged targets and their backgrounds has significantly compromised detection performance in specific scenarios.Conventional feature extraction methods are often limited to single,shallow spectral or spatial features,failing to extract deep features and consequently yielding suboptimal classification accuracy.To address these limitations,this study proposes an innovative 3D-2D convolutional neural networks architecture incorporating depthwise separable convolution(DSC)and attention mechanisms(AM).The framework first applies dimensionality reduction to hyperspectral images and extracts preliminary spectral-spatial features.It then employs an alternating combination of 3D and 2D convolutions for deep feature extraction.For target classification,the LogSoftmax function is implemented.The integration of depthwise separable convolution not only enhances classification accuracy but also substantially reduces model parameters.Furthermore,the attention mechanisms significantly improve the network's ability to represent multidimensional features.Extensive experiments were conducted on a custom land-based hyperspectral image dataset.The results demonstrate remarkable classification accuracy:98.74%for grassland camouflage,99.13%for dead leaf camouflage and 98.94%for wild grass camouflage.Comparative analysis shows that the proposed framework is outstanding in terms of classification accuracy and robustness for camouflage target classification.展开更多
In hyperspectral image classification(HSIC),accurately extracting spatial and spectral information from hyperspectral images(HSI)is crucial for achieving precise classification.However,due to low spatial resolution an...In hyperspectral image classification(HSIC),accurately extracting spatial and spectral information from hyperspectral images(HSI)is crucial for achieving precise classification.However,due to low spatial resolution and complex category boundary,mixed pixels containing features from multiple classes are inevitable in HSIs.Additionally,the spectral similarity among different classes challenge for extracting distinctive spectral features essential for HSIC.To address the impact of mixed pixels and spectral similarity for HSIC,we propose a central-pixel guiding sub-pixel and sub-channel convolution network(CP-SPSC)to extract more precise spatial and spectral features.Firstly,we designed spatial attention(CP-SPA)and spectral attention(CP-SPE)informed by the central pixel to effectively reduce spectral interference of irrelevant categories in the same patch.Furthermore,we use CP-SPA to guide 2D sub-pixel convolution(SPConv2d)to capture spatial features finer than the pixel level.Meanwhile,CP-SPE is also utilized to guide 1D sub-channel con-volution(SCConv1d)in selecting more precise spectral channels.For fusing spatial and spectral information at the feature-level,the spectral feature extension transformation module(SFET)adopts mirror-padding and snake permutation to transform 1D spectral information of the center pixel into 2D spectral features.Experiments on three popular datasets demonstrate that ours out-performs several state-of-the-art methods in accuracy.展开更多
Taoren and Xingren are commonly used herbs in East Asian medicine with different medication functions but huge economic differences,and there are cases of adulterated sales in market transactions.An effective adultera...Taoren and Xingren are commonly used herbs in East Asian medicine with different medication functions but huge economic differences,and there are cases of adulterated sales in market transactions.An effective adulteration recognition based on hyperspectral technology and machine learning was designed as a non-destructive testing method in this paper.A hyperspectral dataset comprising 500 Taoren and 500 Xingren samples was established;six feature selection methods were considered in the modeling of radial basis function-support vector machine(RBF-SVM),whose interaction between the two optimization methods was further researched.Two mixed metaheuristics modeling methods,Mixed-PSO and Mixed-SA,were designed,which fused both band selection and hyperparameter optimization from two-stage into one with detailed process analysis.The metrics of this mixed model were improved by comparing with traditional two-stage method.The accuracy of Mixed-PSO was 89.2%in five-floods crossvalidation that increased 4.818%than vanilla RBF-SVM;the accuracy of Mixed-SA was 88.7%which could reach the same as the traditional two-stage method,but it only relied on 48 crux bands in full 100 bands in RBF-SVM model fitting.展开更多
Accurate cloud classification plays a crucial role in aviation safety,climate monitoring,and localized weather forecasting.Current research has been focusing on machine learning techniques,particularly deep learning b...Accurate cloud classification plays a crucial role in aviation safety,climate monitoring,and localized weather forecasting.Current research has been focusing on machine learning techniques,particularly deep learning based model,for the types identification.However,traditional approaches such as convolutional neural networks(CNNs)encounter difficulties in capturing global contextual information.In addition,they are computationally expensive,which restricts their usability in resource-limited environments.To tackle these issues,we present the Cloud Vision Transformer(CloudViT),a lightweight model that integrates CNNs with Transformers.The integration enables an effective balance between local and global feature extraction.To be specific,CloudViT comprises two innovative modules:Feature Extraction(E_Module)and Downsampling(D_Module).These modules are able to significantly reduce the number of model parameters and computational complexity while maintaining translation invariance and enhancing contextual comprehension.Overall,the CloudViT includes 0.93×10^(6)parameters,which decreases more than ten times compared to the SOTA(State-of-the-Art)model CloudNet.Comprehensive evaluations conducted on the HBMCD and SWIMCAT datasets showcase the outstanding performance of CloudViT.It achieves classification accuracies of 98.45%and 100%,respectively.Moreover,the efficiency and scalability of CloudViT make it an ideal candidate for deployment inmobile cloud observation systems,enabling real-time cloud image classification.The proposed hybrid architecture of CloudViT offers a promising approach for advancing ground-based cloud image classification.It holds significant potential for both optimizing performance and facilitating practical deployment scenarios.展开更多
Among hyperspectral imaging technologies, interferometric spectral imaging is widely used in remote sening due to advantages of large luminous flux and high resolution. However, with complicated mechanism, interferome...Among hyperspectral imaging technologies, interferometric spectral imaging is widely used in remote sening due to advantages of large luminous flux and high resolution. However, with complicated mechanism, interferometric imaging faces the impact of multi-stage degradation. Most exsiting interferometric spectrum reconstruction methods are based on tradition model-based framework with multiple steps, showing poor efficiency and restricted performance. Thus, we propose an interferometric spectrum reconstruction method based on degradation synthesis and deep learning.Firstly, based on imaging mechanism, we proposed an mathematical model of interferometric imaging to analyse the degradation components as noises and trends during imaging. The model consists of three stages, namely instrument degradation, sensing degradation, and signal-independent degradation process. Then, we designed calibration-based method to estimate parameters in the model, of which the results are used for synthesizing realistic dataset for learning-based algorithms.In addition, we proposed a dual-stage interferogram spectrum reconstruction framework, which supports pre-training and integration of denoising DNNs. Experiments exhibits the reliability of our degradation model and synthesized data, and the effectiveness of the proposed reconstruction method.展开更多
Images obtained from hyperspectral sensors provide information about the target area that extends beyond the visible portions of the electromagnetic spectrum.However,due to sensor limitations and imperfections during ...Images obtained from hyperspectral sensors provide information about the target area that extends beyond the visible portions of the electromagnetic spectrum.However,due to sensor limitations and imperfections during the image acquisition and transmission phases,noise is introduced into the acquired image,which can have a negative impact on downstream analyses such as classification,target tracking,and spectral unmixing.Noise in hyperspectral images(HSI)is modelled as a combination from several sources,including Gaussian/impulse noise,stripes,and deadlines.An HSI restoration method for such a mixed noise model is proposed.First,a joint optimisation framework is proposed for recovering hyperspectral data corrupted by mixed Gaussian-impulse noise by estimating both the clean data as well as the sparse/impulse noise levels.Second,a hyper-Laplacian prior is used along both the spatial and spectral dimensions to express sparsity in clean image gradients.Third,to model the sparse nature of impulse noise,anℓ_(1)−norm over the impulse noise gradient is used.Because the proposed methodology employs two distinct priors,the authors refer to it as the hyperspectral dual prior(HySpDualP)denoiser.To the best of authors'knowledge,this joint optimisation framework is the first attempt in this direction.To handle the non-smooth and nonconvex nature of the generalℓ_(p)−norm-based regularisation term,a generalised shrinkage/thresholding(GST)solver is employed.Finally,an efficient split-Bregman approach is used to solve the resulting optimisation problem.Experimental results on synthetic data and real HSI datacube obtained from hyperspectral sensors demonstrate that the authors’proposed model outperforms state-of-the-art methods,both visually and in terms of various image quality assessment metrics.展开更多
Quantitative inversion is a major topic in remote sensing science.The development of visible light-based hyperspectral reconstruction techniques has opened novel prospects for low-cost,high-precision remote sensing in...Quantitative inversion is a major topic in remote sensing science.The development of visible light-based hyperspectral reconstruction techniques has opened novel prospects for low-cost,high-precision remote sensing inversion in agriculture.The aim of this study was to assess the effectiveness of hyperspectral reconstruction technology in agricultural remote sensing applications.Hyperspectral images were reconstructed using the MST++hyperspectral reconstruction model and compared with the original visible light images in terms of their correlations with physiological parameters,the accuracy of single-feature modeling,and the accuracy of combined feature modeling.The results showed that compared to the visible light image,the reconstructed data exhibited a stronger correlation with the measured physiological parameters,and the accuracy was improved for both the single feature and combined feature inversion modes.However,compared to multispectral sensors,hyperspectral reconstruction provided limited improvement of the inversion model accuracy.The results suggest that for physiological parameters that are not easy to observe directly,deep mining of features in visible light data through hyperspectral reconstruction technology can improve the accuracy of the inversion model.However,appropriate feature selection and simple models are more suitable for the remote sensing inversion task of traditional agronomic plot experiments.To strengthen the application of hyperspectral reconstruction technology in agricultural remote sensing,further development is necessary with broader wavelength ranges and more diverse agricultural scenarios.展开更多
Potassium(K)is a highly mobile nutrient element that continuously adjusts its demand strategy among and within cotton leaves through redistribution,indirectly leading to variations in the leaf potassium content(LKC,%)...Potassium(K)is a highly mobile nutrient element that continuously adjusts its demand strategy among and within cotton leaves through redistribution,indirectly leading to variations in the leaf potassium content(LKC,%)at different leaf positions.However,due to the interaction between light and leaf age,leaf sensitivity to this change varies at different positions,including the reflection and absorption of the spectrum.Selecting the optimal leaf position for monitoring is a crucial factor in the rapid and accurate evaluation of cotton LKC using spectral remote sensing technology.Therefore,this study proposes a comprehensive multi-leaf position estimation model based on the vertical distribution characteristics of LKC from top to bottom,aiming to achieve an accurate estimation of cotton LKC and optimize the strategy for selecting the monitored leaf position.Between 2020 and 2021,we collected hyperspectral imaging data of the main stem leaves at different positions from top to bottom(Li,i=1,2,3,...,n)during the cotton budding,flowering,and boll-setting stages.Vertical distribution characteristics,sensitivity differences,and spectral correlations of LKC at different leaf positions were investigated.Additionally,the optimal range of the dominant leaf position for monitoring was determined.Partial least squares regression(PLSR),random forest regression(RFR),support vector machine regression(SVR),and the entropy weight method(EWM)were employed to develop LKC estimation models for single-and multi-leaf positions.The results showed a vertical heterogeneous distribution of cotton LKC,with LKC initially increasing and then gradually decreasing from top to bottom;the average LKC of cotton reached its maximum value at the flowering stage.The upper leaf position demonstrated greater sensitivity to K and exhibited a stronger correlation with the spectrum.The selected dominant leaf positions for the three growth stages were L1-L5,L1-L4,and L1-L2,respectively.Based on the dominant leaf position monitoring range,the optimal single leaf position models for estimating LKC during the three growth stages were PLSR-L4,PLSR-L1,and SVR-L2,with the coefficient of determination of the validation set(R2val)being 0.786,0.580,and 0.768,and the root-mean-square error of the validation set(RMSEval)being 0.168,0.197,and 0.191,respectively.The multi-leaf position LKC estimation model was constructed by EWM with R2val being 0.887,0.728,and 0.703,and RMSEval being 0.134,0.172,and 0.209,respectively.In contrast,the newly developed multi-leaf position comprehensive estimation model yielded superior results,improving the model’s stability based on high accuracy,especially during the budding and flowering stages.These findings hold significant importance for investigating cotton LKC spectral models and selecting suitable leaf positions for field monitoring.展开更多
基金supported by the Innovation and Development Special Project of the China Meteorological Administration(Grant No.CXFZ2024J058)the Guangdong Province Basic and Applied Basic Research Foundation Meteorological Joint Fund Project(Grant No.2024A1515510036)+1 种基金the National Key R&D Program of China(Grant No.2022YFC3004101)the Technical Innovation Team Project of Guangzhou Meteorological Satellite Ground Station(Grant No.CXTD202401).
文摘FY-3G is the first polar-orbiting satellite equipped with a precipitation measurement radar(PMR)operating at Ku-andKa-band frequencies in China.In this study,we compare the reflectivity data from the FY-3G PMR Ku product and groundbasedradars(GRs)during 2024.Also,the FY-3G PMR is used as a third-party reference to evaluate the reflectivityconsistency among different GRs.The FY-3G PMR and GRs share similarities in their general distribution,characteristics,and intensity of reflectivity in strong precipitation cloud systems,though the former presents less detailed system structure.Systematic deviations between the FY-3G PMR and GRs and between GRs are comparable,albeit the reflectivity of the FY-3G PMR is generally slightly stronger than that of GRs(especially X-band GRs),with a mean bias ranging from 0.7 to 1.7dB.S-band GRs exhibit the smallest systematic deviation(STD=3.09 dB)from the FY-3G PMR,whereas the X-band GRsshow the largest(STD=3.61 dB),indirectly indicating the highest internal consistency among S-band GRs and the lowestamong X-band GRs.Besides,both S-and C-band GRs display similar deviations when paired with the FY-3G PMR as wellas when paired with their adjacent S/C-band GRs,suggesting good consistency between these two bands.In contrast,XbandGRs exhibit relatively poor consistency with S-band GRs and the FY-3G PMR,showing a deviation ranging from 3.0to 4.6 dB.
基金funded by the National Natural Science Foundation of China(Nos.52474165 and 52522404)。
文摘Cemented paste backfill(CPB)is a technology that achieves safe mining by filling the goaf with waste rocks,tailings,and other materials.It is an inevitable choice to deal with the development of deep and highly difficult mines and meet the requirements of environmental protection and safety regulations.It promotes the development of a circular economy in mines through the development of lowgrade resources and the resource utilization of waste,and extends the service life of mines.The mass concentration of solid content(abbreviated as“concentration”)is a critical parameter for CPB.However,discrepancies often arise between the on-site measurements and the pre-designed values due to factors such as groundwater inflow and segregation within the goaf,which cannot be evaluated after the solidification of CPB.This paper innovatively provides an in-situ non-destructive approach to identify the real concentration of CPB after curing for certain days using hyperspectral imaging(HSI)technology.Initially,the spectral variation patterns under different concentration conditions were investigated through hyperspectral scanning experiments on CPB samples.The results demonstrate that as the CPB concentration increases from 61wt%to 73wt%,the overall spectral reflectance gradually increases,with two distinct absorption peaks observed at 1407 and 1917 nm.Notably,the reflectance at 1407 nm exhibited a strong linear relationship with the concentration.Subsequently,the K-nearest neighbors(KNN)and support vector machine(SVM)algorithms were employed to classify and identify different concentrations.The study revealed that,with the KNN algorithm,the highest accuracy was achieved when K(number of nearest neighbors)was 1,although this resulted in overfitting.When K=3,the model displayed the optimal balance between accuracy and stability,with an accuracy of 95.03%.In the SVM algorithm,the highest accuracy of 98.24%was attained with parameters C(regularization parameter)=200 and Gamma(kernel coefficient)=10.A comparative analysis of precision,accuracy,and recall further highlighted that the SVM provided superior stability and precision for identifying CPB concentration.Thus,HSI technology offers an effective solution for the in-situ,non-destructive monitoring of CPB concentration,presenting a promising approach for optimizing and controlling CPB characteristic parameters.
基金supported by the Jiangsu Province Frontier Leading Technology Basic Research Special Project-Research on the New Optoelectronic Imaging and Information Processing Basic Theory and Method(No:BK20192003).
文摘This review summarizes studies of hydrothermal alteration minerals at the Qiucun gold deposit in southeastern China and focuses on characterization and mapping of the deposit using hyperspectral remote sensing.The deposit exhibits multistage fluid-rock interaction,as evidenced by systematic alteration assemblages,including silicification,sericitization by white micas,the development of argillaceous clays,variable chloritization,and locally significant carbonate alteration.We describe the genetic importance of such mineral groups and emphasize their diagnostic Visible and Near-Infrared to Short-Wave Infrared(VNIR-SWIR)spectral signatures,especially Al-OH,Mg-OH/Fe-OH,and CO3 absorption bands,which make it possible to distinguish between minerals,not to mention the fact that,in some instances,compositional trends may be predicted.This review’s methodological advances are discussed beginning with data collection at satellite,airborne,and ground levels,proceeding to processing procedures,such as atmospheric and topographic correction,and culminating in spectral analysis,including continuum removal,spectral matching,and unmixing/classification techniques.An integrated study of hyperspectral findings reveals that alteration minerals develop spatially coherent zones that are strongly controlled by fault/fracture structures and host-rock reactivity,producing proximal silicification/sericitization cores and larger silicified/larcenies of argillaceous rocks owing to diverse apex coverings of carbonate.This should be combined with petrography and geochemistry to address overprinting,mixed pixels,and surface weathering,and to couple mineral maps with ore-forming processes.The review finds that hyperspectral remote sensing offers a solid modeling platform for the deposit-scale alteration at Qiucun and other hydrothermal gold systems,and outlines the directions for future research to integrate quantitatively and more threedimensional alteration characterization.
文摘Also known as imaging spectroscopy,hyperspectral remote sensing is becoming a key technology for ecosystem and natural resource management sustainability.Hyperspectral observations can be used to measure tens to hundreds of narrow bands of reflected radiation to resolve diagnostic absorption bands and spectral shape variations associated with vegetation pigments,water status of the canopy,biochemical composition,mineralogies,and organic matter of the soil,and water quality constituents of aquatic water.These abilities allow one to make a transition between the descriptive mapping and the functional monitoring,the anticipation of stress and disturbance early,and the more accurate attribution of environmental change.This summary encompasses improvements on the entire sensor-to-product pipeline,including field and UAV(Unmanned Aerial Vehicle)system platform developments,airborne campaign and spaceborne mission developments,calibration and analysis-ready preprocessing improvements,empirical learning methodology improvements,radiative transfer-based inversion method,spectral unmixing,deep learning,and hybrid physics-machine learning.We underline the increased importance of the combination of data with LiDAR(Light Detection and Ranging),SAR(Synthetic Aperture Radar),and thermal features aimed at decreasing the level of ambiguity and enhancing operational resilience.Applications based on decision are evaluated in terms of biodiversity and habitat evaluation,vegetation functionality and restoration,stress and disturbance,sustainable agricultural production,inland water quality and coastal water quality,land degradation and soil status,and environmental impact assessment.Inhibiting factors to operational adoption have always been perceived to be domain shift by region,season,and sensor,ground truth and validation,mixed pixels and scale mismatch,preprocessing sensitivities,and desirable uncertainty quantification and product output that is interpretable.We conclude with the scalability,sustainability,service priorities,such as harmonization standards,representative benchmarking,uncertainty-aware delivery,and co-design of stakeholders.
基金supported by the Shandong Province Higher Education Institutions New Technology R&D Platform—Spatiotemporal IoT Cloud Application New Technology R&D Center,Shandong Vocational Education Skill Master Studio—Zhao Yaqian Skill Master Studio,and Shandong University of Engineering and Vocational Technology.
文摘Hyperspectral remote sensing has emerged as a transformative technology for sustainable natural resource management by providing unprecedented insight into the biochemical,biophysical,and compositional properties of Earth’s surface.The high spectral resolution of hyperspectral sensors allows a very specific discrimination of materials,monitoring of environmental stress at a very early stage,and provides quantitative retrieval of ecological and geochemical parameters in a wide range of landscapes.The booming technology in sensor design,machine learning,spectral unmixing,and multi-sensor data fusion has further improved the analysis potential and application of imaging spectroscopy to a large extent.This paper involves a discussion of the oversight of such technological advances and the manner in which they are utilized in the principal fields that include forestry,agriculture,water,mineral exploration,and coastal ecosystems.Case studies allow us to identify the potential practical consequences of both spaceborne and unmanned aerial vehicles(UAV)-based hyperspectral systems and AI-based workflows that can be used to aid in more efficient and accurate environmental review.Even though the issues associated with data volume,atmospheric impacts,lack of uniformity in the calibration process,and socioeconomic limits continue to exist,the new technology in sensor miniaturization,cloud computing,and artificial intelligence indicates a fast-changing environment.All these developments make hyperspectral remote sensing a key instrument in solving global sustainability problems and evidence-based management of natural resources in an evolving world.
文摘Hyperspectral images(HSIs)are susceptible to various noise interferences during the imaging process,leading to degraded image quality and affecting the accuracy of information extraction.Efficient denoising methods are crucial for ensuring the accuracy of subsequent remote sensing analysis and applications.In view of the characteristics of hyperspectral image data,such as high dimensionality,strong spectral correlation,and high computational complexity,a threedimensional visual state space U-Net(VSSU3D)was proposed in this paper.By introducing a visual state space module into the traditional U-Net,and combining the spatial-spectral characteristics of hyperspectral images with the core idea of the Mamba model,targeted optimizations wereachieved to effectively model global information dependencies while reducing computational complexity.Additionally,a simplified channel attention module was embedded between the encoder and decoder to enhance cross-scale feature fusion capabilities.Experimental results on multiple publicly available hyperspectral image datasets demonstrated that VSSU3D achieved denoising performance comparable to or superior to existing advanced methods,which verified its effectiveness.
文摘Agricultural remote sensing has been developed and applied in monitoring soil,crop growth,weed infestation,insects,diseases and water status in farm fields to provide data and information to guide agricultural management practices.Precision agriculture has been implemented through prescription mapping of crop fields at different scales with the data remotely sensed from space-borne,airborne and ground-based platforms.Ground-based remote sensing techniques offer portability,flexibility and controllability in applications for precision agriculture.In weed management,crop injury from off-target herbicide spray drift and herbicide resistance in weeds are two important issues.For precision weed management,ground-based hyperspectral remote sensing techniques were developed for detection of crop injury from dicamba and differentiation between glyphosate resistant and sensitive weeds.This research presents the techniques for ground-based hyperspectral remote sensing for these two applications.Results illustrate the advantages of ground-based hyperspectral remote sensing for precision weed management.
基金funded by the National Key Research and Development Program of China(Grant No.2023YFB3907500)the National Natural Science Foundation(Grant No.42330602)the“Fengyun Satellite Remote Sensing Product Validation and Verification”Youth Innovation Team of the China Meteorological Administration(Grant No.CMA2023QN12)。
文摘The Global Precipitation Measurement(GPM)dual-frequency precipitation radar(DPR)products(Version 07A)are employed for a rigorous comparative analysis with ground-based operational weather radar(GR)networks.The reflectivity observed by GPM Ku PR is compared quantitatively against GR networks from CINRAD of China and NEXRAD of the United States,and the volume matching method is used for spatial matching.Additionally,a novel frequency correction method for all phases as well as precipitation types is used to correct the GPM Ku PR radar frequency to the GR frequency.A total of 20 GRs(including 10 from CINRAD and 10 from NEXRAD)are included in this comparative analysis.The results indicate that,compared with CINRAD matched data,NEXRAD exhibits larger biases in reflectivity when compared with the frequency-corrected Ku PR.The root-mean-square difference for CINRAD is calculated at 2.38 d B,whereas for NEXRAD it is 3.23 d B.The mean bias of CINRAD matched data is-0.16 d B,while the mean bias of NEXRAD is-2.10 d B.The mean standard deviation of bias for CINRAD is 2.15 d B,while for NEXRAD it is 2.29 d B.This study effectively assesses weather radar data in both the United States and China,which is crucial for improving the overall consistency of global precipitation estimates.
基金supported in part by the Six Talent Peaks Project in Jiangsu Province under Grant 013040315in part by the China Textile Industry Federation Science and Technology Guidance Project under Grant 2017107+1 种基金in part by the National Natural Science Foundation of China under Grant 31570714in part by the China Scholarship Council under Grant 202108320290。
文摘The cleanliness of seed cotton plays a critical role in the pre-treatment of cotton textiles,and the removal of impurity during the harvesting process directly determines the quality and market value of cotton textiles.By fusing band combination optimization with deep learning,this study aims to achieve more efficient and accurate detection of film impurities in seed cotton on the production line.By applying hyperspectral imaging and a one-dimensional deep learning algorithm,we detect and classify impurities in seed cotton after harvest.The main categories detected include pure cotton,conveyor belt,film covering seed cotton,and film adhered to the conveyor belt.The proposed method achieves an impurity detection rate of 99.698%.To further ensure the feasibility and practical application potential of this strategy,we compare our results against existing mainstream methods.In addition,the model shows excellent recognition performance on pseudo-color images of real samples.With a processing time of 11.764μs per pixel from experimental data,it shows a much improved speed requirement while maintaining the accuracy of real production lines.This strategy provides an accurate and efficient method for removing impurities during cotton processing.
基金National Natural Science Foundation of China(No. 42571463)Macao Young Scholars Program(No. AM2023033)+1 种基金Shaanxi Province Youth Science and Technology Star Program(No. 2024ZCKJXX-115)Natural Science Foundation of Shaanxi Province(No.2025JC-YBMS-257)。
文摘Hyper Spectral Image Super-Resolution(HSI-SR) has gained significant attention in recent years due to its potential applications.However,the challenge of obtaining high-resolution hyperspectral images is compounded by limitations in sensor resolution and the high dimensionality of spectral data.Traditional approaches,including interpolation-based methods and sparse representation techniques,often struggle to capture the intricate spectral-spatial dependencies in hyperspectral images.To address these limitations,this study proposes a Hadamard Self-Attention Network(HSAN) for fusing a High-resolution Multispectral Image(Hr-MSI) and a Low-resolution Hyper Spectral Image(Lr-HSI),achieving HSI-SR for obtaining a High-resolution Hyper Spectral Image(Hr-HSI).The core of HSAN is a new Hadamard self-attention mechanism that can be more efficient than traditional dot-product attention because it avoids matrix multiplications and softmax operations.Considering that deep learning-based data fusion typically entails a significant computational and storage burden,this new approach can be integrated with convolutional layers to form an unsupervised lightweight network,which significantly reduces dependence on computational resources.Experimental results across four datasets validate the effectiveness and advantages of HSAN,compared with state-of-the-art approaches.The source code will be available at https://github.com/zxnhkdm/HSAN.
基金supported by the Science and Technology Development Plan Project of Jilin Provincial Department of Science and Technology (No.20220203112S)the Jilin Provincial Department of Education Science and Technology Research Project (No.JJKH20210039KJ)。
文摘In this study,eight different varieties of maize seeds were used as the research objects.Conduct 81 types of combined preprocessing on the original spectra.Through comparison,Savitzky-Golay(SG)-multivariate scattering correction(MSC)-maximum-minimum normalization(MN)was identified as the optimal preprocessing technique.The competitive adaptive reweighted sampling(CARS),successive projections algorithm(SPA),and their combined methods were employed to extract feature wavelengths.Classification models based on back propagation(BP),support vector machine(SVM),random forest(RF),and partial least squares(PLS)were established using full-band data and feature wavelengths.Among all models,the(CARS-SPA)-BP model achieved the highest accuracy rate of 98.44%.This study offers novel insights and methodologies for the rapid and accurate identification of corn seeds as well as other crop seeds.
基金support from the National Natural Science Foundation of China(Grant Nos.52379103,52279103)the Natural Science Foundation of Shandong Province(Grant No.ZR2023YQ049).
文摘Imaging hyperspectral technology has distinctive advantages of non-destructive and non-contact measurement,and the integration of spectral and spatial data.These characteristics present new methodologies for intelligent geological sensing in tunnels and other underground engineering projects.However,the in situ acquisition and rapid classification of hyperspectral images in underground still faces great challenges,including the difficulty in obtaining uniform hyperspectral images and the complexity of deploying sophisticated models on mobile platforms.This study proposes an intelligent lithology identification method based on partition feature extraction of hyperspectral images.Firstly,pixel-level hyperspectral information from representative lithological regions is extracted and fused to obtain rock hyperspectral image partition features.Subsequently,an SG-SNV-PCA-DNN(SSPD)model specifically designed for optimizing rock hyperspectral data,performing spectral dimensionality reduction,and identifying lithology is integrated.In an experimental study involving 3420 hyperspectral images,the SSPD identification model achieved the highest accuracy in the testing set,reaching 98.77%.Moreover,the speed of the SSPD model was found to be 18.5%faster than that of the unprocessed model,with an accuracy improvement of 5.22%.In contrast,the ResNet-101 model,used for point-by-point identification based on non-partitioned features,achieved a maximum accuracy of 97.86%in the testing set.In addition,the partition feature extraction methods significantly reduce computational complexity.An objective evaluation of various models demonstrated that the SSPD model exhibited superior performance,achieving a precision(P)of 99.46%,a recall(R)of 99.44%,and F1 score(F1)of 99.45%.Additionally,a pioneering in situ detection work was carried out in a tunnel using underground hyperspectral imaging technology.
文摘In recent years,camouflage technology has evolved from single-spectral-band applications to multifunctional and multispectral implementations.Hyperspectral imaging has emerged as a powerful technique for target identification due to its capacity to capture both spectral and spatial information.The advancement of imaging spectroscopy technology has significantly enhanced reconnaissance capabilities,offering substantial advantages in camouflaged target classification and detection.However,the increasing spectral similarity between camouflaged targets and their backgrounds has significantly compromised detection performance in specific scenarios.Conventional feature extraction methods are often limited to single,shallow spectral or spatial features,failing to extract deep features and consequently yielding suboptimal classification accuracy.To address these limitations,this study proposes an innovative 3D-2D convolutional neural networks architecture incorporating depthwise separable convolution(DSC)and attention mechanisms(AM).The framework first applies dimensionality reduction to hyperspectral images and extracts preliminary spectral-spatial features.It then employs an alternating combination of 3D and 2D convolutions for deep feature extraction.For target classification,the LogSoftmax function is implemented.The integration of depthwise separable convolution not only enhances classification accuracy but also substantially reduces model parameters.Furthermore,the attention mechanisms significantly improve the network's ability to represent multidimensional features.Extensive experiments were conducted on a custom land-based hyperspectral image dataset.The results demonstrate remarkable classification accuracy:98.74%for grassland camouflage,99.13%for dead leaf camouflage and 98.94%for wild grass camouflage.Comparative analysis shows that the proposed framework is outstanding in terms of classification accuracy and robustness for camouflage target classification.
基金supported by the National Natural Science Foundation of China(No.62071323).
文摘In hyperspectral image classification(HSIC),accurately extracting spatial and spectral information from hyperspectral images(HSI)is crucial for achieving precise classification.However,due to low spatial resolution and complex category boundary,mixed pixels containing features from multiple classes are inevitable in HSIs.Additionally,the spectral similarity among different classes challenge for extracting distinctive spectral features essential for HSIC.To address the impact of mixed pixels and spectral similarity for HSIC,we propose a central-pixel guiding sub-pixel and sub-channel convolution network(CP-SPSC)to extract more precise spatial and spectral features.Firstly,we designed spatial attention(CP-SPA)and spectral attention(CP-SPE)informed by the central pixel to effectively reduce spectral interference of irrelevant categories in the same patch.Furthermore,we use CP-SPA to guide 2D sub-pixel convolution(SPConv2d)to capture spatial features finer than the pixel level.Meanwhile,CP-SPE is also utilized to guide 1D sub-channel con-volution(SCConv1d)in selecting more precise spectral channels.For fusing spatial and spectral information at the feature-level,the spectral feature extension transformation module(SFET)adopts mirror-padding and snake permutation to transform 1D spectral information of the center pixel into 2D spectral features.Experiments on three popular datasets demonstrate that ours out-performs several state-of-the-art methods in accuracy.
基金Supported by the Natural Science Foundation of Heilongjiang Province(LH2020C003)。
文摘Taoren and Xingren are commonly used herbs in East Asian medicine with different medication functions but huge economic differences,and there are cases of adulterated sales in market transactions.An effective adulteration recognition based on hyperspectral technology and machine learning was designed as a non-destructive testing method in this paper.A hyperspectral dataset comprising 500 Taoren and 500 Xingren samples was established;six feature selection methods were considered in the modeling of radial basis function-support vector machine(RBF-SVM),whose interaction between the two optimization methods was further researched.Two mixed metaheuristics modeling methods,Mixed-PSO and Mixed-SA,were designed,which fused both band selection and hyperparameter optimization from two-stage into one with detailed process analysis.The metrics of this mixed model were improved by comparing with traditional two-stage method.The accuracy of Mixed-PSO was 89.2%in five-floods crossvalidation that increased 4.818%than vanilla RBF-SVM;the accuracy of Mixed-SA was 88.7%which could reach the same as the traditional two-stage method,but it only relied on 48 crux bands in full 100 bands in RBF-SVM model fitting.
基金funded by Innovation and Development Special Project of China Meteorological Administration(CXFZ2022J038,CXFZ2024J035)Sichuan Science and Technology Program(No.2023YFQ0072)+1 种基金Key Laboratory of Smart Earth(No.KF2023YB03-07)Automatic Software Generation and Intelligent Service Key Laboratory of Sichuan Province(CUIT-SAG202210).
文摘Accurate cloud classification plays a crucial role in aviation safety,climate monitoring,and localized weather forecasting.Current research has been focusing on machine learning techniques,particularly deep learning based model,for the types identification.However,traditional approaches such as convolutional neural networks(CNNs)encounter difficulties in capturing global contextual information.In addition,they are computationally expensive,which restricts their usability in resource-limited environments.To tackle these issues,we present the Cloud Vision Transformer(CloudViT),a lightweight model that integrates CNNs with Transformers.The integration enables an effective balance between local and global feature extraction.To be specific,CloudViT comprises two innovative modules:Feature Extraction(E_Module)and Downsampling(D_Module).These modules are able to significantly reduce the number of model parameters and computational complexity while maintaining translation invariance and enhancing contextual comprehension.Overall,the CloudViT includes 0.93×10^(6)parameters,which decreases more than ten times compared to the SOTA(State-of-the-Art)model CloudNet.Comprehensive evaluations conducted on the HBMCD and SWIMCAT datasets showcase the outstanding performance of CloudViT.It achieves classification accuracies of 98.45%and 100%,respectively.Moreover,the efficiency and scalability of CloudViT make it an ideal candidate for deployment inmobile cloud observation systems,enabling real-time cloud image classification.The proposed hybrid architecture of CloudViT offers a promising approach for advancing ground-based cloud image classification.It holds significant potential for both optimizing performance and facilitating practical deployment scenarios.
文摘Among hyperspectral imaging technologies, interferometric spectral imaging is widely used in remote sening due to advantages of large luminous flux and high resolution. However, with complicated mechanism, interferometric imaging faces the impact of multi-stage degradation. Most exsiting interferometric spectrum reconstruction methods are based on tradition model-based framework with multiple steps, showing poor efficiency and restricted performance. Thus, we propose an interferometric spectrum reconstruction method based on degradation synthesis and deep learning.Firstly, based on imaging mechanism, we proposed an mathematical model of interferometric imaging to analyse the degradation components as noises and trends during imaging. The model consists of three stages, namely instrument degradation, sensing degradation, and signal-independent degradation process. Then, we designed calibration-based method to estimate parameters in the model, of which the results are used for synthesizing realistic dataset for learning-based algorithms.In addition, we proposed a dual-stage interferogram spectrum reconstruction framework, which supports pre-training and integration of denoising DNNs. Experiments exhibits the reliability of our degradation model and synthesized data, and the effectiveness of the proposed reconstruction method.
文摘Images obtained from hyperspectral sensors provide information about the target area that extends beyond the visible portions of the electromagnetic spectrum.However,due to sensor limitations and imperfections during the image acquisition and transmission phases,noise is introduced into the acquired image,which can have a negative impact on downstream analyses such as classification,target tracking,and spectral unmixing.Noise in hyperspectral images(HSI)is modelled as a combination from several sources,including Gaussian/impulse noise,stripes,and deadlines.An HSI restoration method for such a mixed noise model is proposed.First,a joint optimisation framework is proposed for recovering hyperspectral data corrupted by mixed Gaussian-impulse noise by estimating both the clean data as well as the sparse/impulse noise levels.Second,a hyper-Laplacian prior is used along both the spatial and spectral dimensions to express sparsity in clean image gradients.Third,to model the sparse nature of impulse noise,anℓ_(1)−norm over the impulse noise gradient is used.Because the proposed methodology employs two distinct priors,the authors refer to it as the hyperspectral dual prior(HySpDualP)denoiser.To the best of authors'knowledge,this joint optimisation framework is the first attempt in this direction.To handle the non-smooth and nonconvex nature of the generalℓ_(p)−norm-based regularisation term,a generalised shrinkage/thresholding(GST)solver is employed.Finally,an efficient split-Bregman approach is used to solve the resulting optimisation problem.Experimental results on synthetic data and real HSI datacube obtained from hyperspectral sensors demonstrate that the authors’proposed model outperforms state-of-the-art methods,both visually and in terms of various image quality assessment metrics.
基金supported by the China Agriculture Research System(CARS-15-22)the Laboratory of Lingnan Modern Agriculture Project,China(NT2021009)+2 种基金the Key-Area Research and Development Program ofGuangdong Province,China(2019B020214003)the Guangdong Technical System of Peanut and Soybean Industry,China(2019KJ136-05)the 111 Project,China(D18019)。
文摘Quantitative inversion is a major topic in remote sensing science.The development of visible light-based hyperspectral reconstruction techniques has opened novel prospects for low-cost,high-precision remote sensing inversion in agriculture.The aim of this study was to assess the effectiveness of hyperspectral reconstruction technology in agricultural remote sensing applications.Hyperspectral images were reconstructed using the MST++hyperspectral reconstruction model and compared with the original visible light images in terms of their correlations with physiological parameters,the accuracy of single-feature modeling,and the accuracy of combined feature modeling.The results showed that compared to the visible light image,the reconstructed data exhibited a stronger correlation with the measured physiological parameters,and the accuracy was improved for both the single feature and combined feature inversion modes.However,compared to multispectral sensors,hyperspectral reconstruction provided limited improvement of the inversion model accuracy.The results suggest that for physiological parameters that are not easy to observe directly,deep mining of features in visible light data through hyperspectral reconstruction technology can improve the accuracy of the inversion model.However,appropriate feature selection and simple models are more suitable for the remote sensing inversion task of traditional agronomic plot experiments.To strengthen the application of hyperspectral reconstruction technology in agricultural remote sensing,further development is necessary with broader wavelength ranges and more diverse agricultural scenarios.
基金supported by the Corps Leading Talents Program,China(2023YZ01)the Tianshan Talent Training Program,China(2023TS05)the Crop Smart Production Innovation Team,China(2023TD01).
文摘Potassium(K)is a highly mobile nutrient element that continuously adjusts its demand strategy among and within cotton leaves through redistribution,indirectly leading to variations in the leaf potassium content(LKC,%)at different leaf positions.However,due to the interaction between light and leaf age,leaf sensitivity to this change varies at different positions,including the reflection and absorption of the spectrum.Selecting the optimal leaf position for monitoring is a crucial factor in the rapid and accurate evaluation of cotton LKC using spectral remote sensing technology.Therefore,this study proposes a comprehensive multi-leaf position estimation model based on the vertical distribution characteristics of LKC from top to bottom,aiming to achieve an accurate estimation of cotton LKC and optimize the strategy for selecting the monitored leaf position.Between 2020 and 2021,we collected hyperspectral imaging data of the main stem leaves at different positions from top to bottom(Li,i=1,2,3,...,n)during the cotton budding,flowering,and boll-setting stages.Vertical distribution characteristics,sensitivity differences,and spectral correlations of LKC at different leaf positions were investigated.Additionally,the optimal range of the dominant leaf position for monitoring was determined.Partial least squares regression(PLSR),random forest regression(RFR),support vector machine regression(SVR),and the entropy weight method(EWM)were employed to develop LKC estimation models for single-and multi-leaf positions.The results showed a vertical heterogeneous distribution of cotton LKC,with LKC initially increasing and then gradually decreasing from top to bottom;the average LKC of cotton reached its maximum value at the flowering stage.The upper leaf position demonstrated greater sensitivity to K and exhibited a stronger correlation with the spectrum.The selected dominant leaf positions for the three growth stages were L1-L5,L1-L4,and L1-L2,respectively.Based on the dominant leaf position monitoring range,the optimal single leaf position models for estimating LKC during the three growth stages were PLSR-L4,PLSR-L1,and SVR-L2,with the coefficient of determination of the validation set(R2val)being 0.786,0.580,and 0.768,and the root-mean-square error of the validation set(RMSEval)being 0.168,0.197,and 0.191,respectively.The multi-leaf position LKC estimation model was constructed by EWM with R2val being 0.887,0.728,and 0.703,and RMSEval being 0.134,0.172,and 0.209,respectively.In contrast,the newly developed multi-leaf position comprehensive estimation model yielded superior results,improving the model’s stability based on high accuracy,especially during the budding and flowering stages.These findings hold significant importance for investigating cotton LKC spectral models and selecting suitable leaf positions for field monitoring.
