The adulteration concentration of palm kernel oil(PKO)in virgin coconut oil(VCO)was quantified using near-infrared(NIR)hyperspectral imaging.Nowadays,some VCO is adulterated with lower-priced PKO to reduce production ...The adulteration concentration of palm kernel oil(PKO)in virgin coconut oil(VCO)was quantified using near-infrared(NIR)hyperspectral imaging.Nowadays,some VCO is adulterated with lower-priced PKO to reduce production costs,which diminishes the quality of the VCO.This study used NIR hyperspectral imaging in the wavelength region 900-1,650 nm to create a quantitative model for the detection of PKO contaminants(0-100%)in VCO and to develop predictive mapping.The prediction equation for the adulteration of VCO with PKO was constructed using the partial least squares regression method.The best predictive model was pre-processed using the standard normal variate method,and the coefficient of determination of prediction was 0.991,the root mean square error of prediction was 2.93%,and the residual prediction deviation was 10.37.The results showed that this model could be applied for quantifying the adulteration concentration of PKO in VCO.The prediction adulteration concentration mapping of VCO with PKO was created from a calibration model that showed the color level according to the adulteration concentration in the range of 0-100%.NIR hyperspectral imaging could be clearly used to quantify the adulteration of VCO with a color level map that provides a quick,accurate,and non-destructive detection method.展开更多
By automatically learning the priors embedded in images with powerful modelling ca-pabilities,deep learning-based algorithms have recently made considerable progress in reconstructing the high-resolution hyperspectral...By automatically learning the priors embedded in images with powerful modelling ca-pabilities,deep learning-based algorithms have recently made considerable progress in reconstructing the high-resolution hyperspectral(HR-HS)image.With previously collected large-amount of external data,these methods are intuitively realised under the full supervision of the ground-truth data.Thus,the database construction in merging the low-resolution(LR)HS(LR-HS)and HR multispectral(MS)or RGB image research paradigm,commonly named as HSI SR,requires collecting corresponding training triplets:HR-MS(RGB),LR-HS and HR-HS image simultaneously,and often faces dif-ficulties in reality.The learned models with the training datasets collected simultaneously under controlled conditions may significantly degrade the HSI super-resolved perfor-mance to the real images captured under diverse environments.To handle the above-mentioned limitations,the authors propose to leverage the deep internal and self-supervised learning to solve the HSI SR problem.The authors advocate that it is possible to train a specific CNN model at test time,called as deep internal learning(DIL),by on-line preparing the training triplet samples from the observed LR-HS/HR-MS(or RGB)images and the down-sampled LR-HS version.However,the number of the training triplets extracted solely from the transformed data of the observation itself is extremely few particularly for the HSI SR tasks with large spatial upscale factors,which would result in limited reconstruction performance.To solve this problem,the authors further exploit deep self-supervised learning(DSL)by considering the observations as the unlabelled training samples.Specifically,the degradation modules inside the network were elaborated to realise the spatial and spectral down-sampling procedures for transforming the generated HR-HS estimation to the high-resolution RGB/LR-HS approximation,and then the reconstruction errors of the observations were formulated for measuring the network modelling performance.By consolidating the DIL and DSL into a unified deep framework,the authors construct a more robust HSI SR method without any prior training and have great potential of flexible adaptation to different settings per obser-vation.To verify the effectiveness of the proposed approach,extensive experiments have been conducted on two benchmark HS datasets,including the CAVE and Harvard datasets,and demonstrate the great performance gain of the proposed method over the state-of-the-art methods.展开更多
With limited number of labeled samples,hyperspectral image(HSI)classification is a difficult Problem in current research.The graph neural network(GNN)has emerged as an approach to semi-supervised classification,and th...With limited number of labeled samples,hyperspectral image(HSI)classification is a difficult Problem in current research.The graph neural network(GNN)has emerged as an approach to semi-supervised classification,and the application of GNN to hyperspectral images has attracted much attention.However,in the existing GNN-based methods a single graph neural network or graph filter is mainly used to extract HSI features,which does not take full advantage of various graph neural networks(graph filters).Moreover,the traditional GNNs have the problem of oversmoothing.To alleviate these shortcomings,we introduce a deep hybrid multi-graph neural network(DHMG),where two different graph filters,i.e.,the spectral filter and the autoregressive moving average(ARMA)filter,are utilized in two branches.The former can well extract the spectral features of the nodes,and the latter has a good suppression effect on graph noise.The network realizes information interaction between the two branches and takes good advantage of different graph filters.In addition,to address the problem of oversmoothing,a dense network is proposed,where the local graph features are preserved.The dense structure satisfies the needs of different classification targets presenting different features.Finally,we introduce a GraphSAGEbased network to refine the graph features produced by the deep hybrid network.Extensive experiments on three public HSI datasets strongly demonstrate that the DHMG dramatically outperforms the state-ofthe-art models.展开更多
Although airborne hyperspectral data with detailed spatial and spectral information has demonstrated significant potential for tree species classification,it has not been widely used over large areas.A comprehensive p...Although airborne hyperspectral data with detailed spatial and spectral information has demonstrated significant potential for tree species classification,it has not been widely used over large areas.A comprehensive process based on multi-flightline airborne hyperspectral data is lacking over large,forested areas influenced by both the effects of bidirectional reflectance distribution function(BRDF)and cloud shadow contamination.In this study,hyperspectral data were collected over the Mengjiagang Forest Farm in Northeast China in the summer of 2017 using the Chinese Academy of Forestry's LiDAR,CCD,and hyperspectral systems(CAF-LiCHy).After BRDF correction and cloud shadow detection processing,a tree species classification workflow was developed for sunlit and cloud-shaded forest areas with input features of minimum noise fraction reduced bands,spectral vegetation indices,and texture information.Results indicate that BRDF-corrected sunlit hyperspectral data can provide a stable and high classification accuracy based on representative training data.Cloud-shaded pixels also have good spectral separability for species classification.The red-edge spectral information and ratio-based spectral indices with high importance scores are recommended as input features for species classification under varying light conditions.According to the classification accuracies through field survey data at multiple spatial scales,it was found that species classification within an extensive forest area using airborne hyperspectral data under various illuminations can be successfully carried out using the effective radiometric consistency process and feature selection strategy.展开更多
Hyperspectral image super-resolution,which refers to reconstructing the high-resolution hyperspectral image from the input low-resolution observation,aims to improve the spatial resolution of the hyperspectral image,w...Hyperspectral image super-resolution,which refers to reconstructing the high-resolution hyperspectral image from the input low-resolution observation,aims to improve the spatial resolution of the hyperspectral image,which is beneficial for subsequent applications.The development of deep learning has promoted significant progress in hyperspectral image super-resolution,and the powerful expression capabilities of deep neural networks make the predicted results more reliable.Recently,several latest deep learning technologies have made the hyperspectral image super-resolution method explode.However,a comprehensive review and analysis of the latest deep learning methods from the hyperspectral image super-resolution perspective is absent.To this end,in this survey,we first introduce the concept of hyperspectral image super-resolution and classify the methods from the perspectives with or without auxiliary information.Then,we review the learning-based methods in three categories,including single hyperspectral image super-resolution,panchromatic-based hyperspectral image super-resolution,and multispectral-based hyperspectral image super-resolution.Subsequently,we summarize the commonly used hyperspectral dataset,and the evaluations for some representative methods in three categories are performed qualitatively and quantitatively.Moreover,we briefly introduce several typical applications of hyperspectral image super-resolution,including ground object classification,urban change detection,and ecosystem monitoring.Finally,we provide the conclusion and challenges in existing learning-based methods,looking forward to potential future research directions.展开更多
Sanxingdui cultural relics are the precious cultural heritage of humanity with high values of history,science,culture,art and research.However,mainstream analytical methods are contacting and detrimental,which is unfa...Sanxingdui cultural relics are the precious cultural heritage of humanity with high values of history,science,culture,art and research.However,mainstream analytical methods are contacting and detrimental,which is unfavorable to the protection of cultural relics.This paper improves the accuracy of the extraction,location,and analysis of artifacts using hyperspectral methods.To improve the accuracy of cultural relic mining,positioning,and analysis,the segmentation algorithm of Sanxingdui cultural relics based on the spatial spectrum integrated network is proposed with the support of hyperspectral techniques.Firstly,region stitching algorithm based on the relative position of hyper spectrally collected data is proposed to improve stitching efficiency.Secondly,given the prominence of traditional HRNet(High-Resolution Net)models in high-resolution data processing,the spatial attention mechanism is put forward to obtain spatial dimension information.Thirdly,in view of the prominence of 3D networks in spectral information acquisition,the pyramid 3D residual network model is proposed to obtain internal spectral dimensional information.Fourthly,four kinds of fusion methods at the level of data and decision are presented to achieve cultural relic labeling.As shown by the experiment results,the proposed network adopts an integrated method of data-level and decision-level,which achieves the optimal average accuracy of identification 0.84,realizes shallow coverage of cultural relics labeling,and effectively supports the mining and protection of cultural relics.展开更多
Rapid and large area acquisition of nitrogen(N)deficiency status is important for achieving the optimal fertilization of rice.Most existing studies,however,focus on the use of unmanned aerial vehicle(UAV)remote sensin...Rapid and large area acquisition of nitrogen(N)deficiency status is important for achieving the optimal fertilization of rice.Most existing studies,however,focus on the use of unmanned aerial vehicle(UAV)remote sensing to diagnose N nutrition in rice,while there are fewer studies on the quantitative description of the degree of N deficiency in rice,and the effects of the critical N concentration on the spectral changes in rice have rarely been explored.Therefore,based on the canopy spectral data obtained by remotely-sensed UAV hyperspectral images,the N content in rice was obtained through field sampling.The construction method of the rice curve for the northeastern critical N concentration was studied,and on this basis,N deficiency was determined.Taking the spectrum of the critical N concentration state as the standard spectrum,the spectral reflectivity data were transformed by the ratios and differences,and the feature extraction of the spectral data was carried out by the successive projections algorithm(SPA).Finally,by taking the characteristic band as the input variable and N deficiency as the output variable,a set of multivariate linear regression(MLR),long short-term memory(LSTM)inversion models based on extreme learning machine(ELM),and the nondominated sorting genetic algorithmⅢextreme learning machine(NSGA-Ⅲ-ELM)were constructed.The results showed two key aspects of this system:1)The correlation between the N deficiency data and original spectrum was poor,but the correlation between the N deficiency data and N deficiency could be improved by a difference change and ratio transformation;2)The inversion results based on the ratio spectrum and NSGA-Ⅲ-ELM algorithm were the best,as the R2values of the training set and validation set were 0.852 and 0.810,and the root mean square error(RMSE)values were 0.291 and 0.308,respectively.From the perspective of the spectral data,the inversion accuracy of the ratio spectrum was better than the accuracy of the original spectrum or difference spectrum.At the algorithm level,the model inversion results based on LSTM algorithms showed a serious overfitting phenomenon and poor inversion effect.The inversion accuracy based on the NSGA-Ⅲ-ELM algorithm was better than the accuracy of the MLR algorithm or the ELM algorithm.Therefore,the inversion model based on the ratio spectrum and NSGA-Ⅲ-ELM algorithm could effectively invert the N deficiency in rice and provide critical technical support for accurate topdressing based on the N status in the rice.展开更多
Most unsupervised or semisupervised hyperspectral anomaly detection(HAD)methods train background reconstruction models in the original spectral domain.However,due to the noise and spatial resolution limitations,there ...Most unsupervised or semisupervised hyperspectral anomaly detection(HAD)methods train background reconstruction models in the original spectral domain.However,due to the noise and spatial resolution limitations,there may be a lack of discrimination between backgrounds and anomalies.This makes it easy for the autoencoder to capture the lowlevel features shared between the two,thereby increasing the difficulty of separating anomalies from the backgrounds,which runs counter to the purpose of HAD.To this end,the authors map the original spectrums to the fractional Fourier domain(FrFD)and reformulate it as a mapping task in which restoration errors are employed to distinguish background and anomaly.This study proposes a novel frequency‐to‐spectrum mapping generative adversarial network for HAD.Specifically,the depth separable features of backgrounds and anomalies are enhanced in the FrFD.Due to the semisupervised approach,FTSGAN needs to learn the embedded features of the backgrounds,thus mapping and restoring them from the FrFD to the original spectral domain.This strategy effectively prevents the model from focussing on the numerical equivalence of input and output,and restricts the ability of FTSGAN to restore anomalies.The comparison and analysis of the experiments verify that the proposed method is competitive.展开更多
Hyperspectral remote sensing/imaging spectroscopy is a novel approach to reaching a spectrum from all the places of a huge array of spatial places so that several spectral wavelengths are utilized for making coherent ...Hyperspectral remote sensing/imaging spectroscopy is a novel approach to reaching a spectrum from all the places of a huge array of spatial places so that several spectral wavelengths are utilized for making coherent images.Hyperspectral remote sensing contains acquisition of digital images from several narrow,contiguous spectral bands throughout the visible,Thermal Infrared(TIR),Near Infrared(NIR),and Mid-Infrared(MIR)regions of the electromagnetic spectrum.In order to the application of agricultural regions,remote sensing approaches are studied and executed to their benefit of continuous and quantitativemonitoring.Particularly,hyperspectral images(HSI)are considered the precise for agriculture as they can offer chemical and physical data on vegetation.With this motivation,this article presents a novel Hurricane Optimization Algorithm with Deep Transfer Learning Driven Crop Classification(HOADTL-CC)model onHyperspectralRemote Sensing Images.The presentedHOADTL-CC model focuses on the identification and categorization of crops on hyperspectral remote sensing images.To accomplish this,the presentedHOADTL-CC model involves the design ofHOAwith capsule network(CapsNet)model for generating a set of useful feature vectors.Besides,Elman neural network(ENN)model is applied to allot proper class labels into the input HSI.Finally,glowworm swarm optimization(GSO)algorithm is exploited to fine tune the ENNparameters involved in this article.The experimental result scrutiny of the HOADTL-CC method can be tested with the help of benchmark dataset and the results are assessed under distinct aspects.Extensive comparative studies stated the enhanced performance of the HOADTL-CC model over recent approaches with maximum accuracy of 99.51%.展开更多
Compressed sensing(CS),as an efficient data transmission method,has achieved great success in the field of data transmission such as image,video and text.It can robustly recover signals from fewer Measurements,effecti...Compressed sensing(CS),as an efficient data transmission method,has achieved great success in the field of data transmission such as image,video and text.It can robustly recover signals from fewer Measurements,effectively alleviating the bandwidth pressure during data transmission.However,CS has many shortcomings in the transmission of hyperspectral image(HSI)data.This work aims to consider the application of CS in the transmission of hyperspectral image(HSI)data,and provides a feasible research scheme for CS of HSI data.HSI has rich spectral information and spatial information in bands,which can reflect the physical properties of the target.Most of the hyperspectral image compressed sensing(HSICS)algorithms cannot effectively use the inter-band information of HSI,resulting in poor reconstruction effects.In this paper,A three-stage hyperspectral image compression sensing algorithm(Three-stages HSICS)is proposed to obtain intra-band and inter-band characteristics of HSI,which can improve the reconstruction accuracy of HSI.Here,we establish a multi-objective band selection(Mop-BS)model,amulti-hypothesis prediction(MHP)model and a residual sparse(ReWSR)model for HSI,and use a staged reconstruction method to restore the compressed HSI.The simulation results show that the three-stage HSICS successfully improves the reconstruction accuracy of HSICS,and it performs best among all comparison algorithms.展开更多
Recently,deep learning has achieved considerable results in the hyperspectral image(HSI)classification.However,most available deep networks require ample and authentic samples to better train the models,which is expen...Recently,deep learning has achieved considerable results in the hyperspectral image(HSI)classification.However,most available deep networks require ample and authentic samples to better train the models,which is expensive and inefficient in practical tasks.Existing few‐shot learning(FSL)methods generally ignore the potential relationships between non‐local spatial samples that would better represent the underlying features of HSI.To solve the above issues,a novel deep transformer and few‐shot learning(DTFSL)classification framework is proposed,attempting to realize fine‐grained classification of HSI with only a few‐shot instances.Specifically,the spatial attention and spectral query modules are introduced to overcome the constraint of the convolution kernel and consider the information between long‐distance location(non‐local)samples to reduce the uncertainty of classes.Next,the network is trained with episodes and task‐based learning strategies to learn a metric space,which can continuously enhance its modelling capability.Furthermore,the developed approach combines the advantages of domain adaptation to reduce the variation in inter‐domain distribution and realize distribution alignment.On three publicly available HSI data,extensive experiments have indicated that the proposed DT‐FSL yields better results concerning state‐of‐the‐art algorithms.展开更多
Background Fiber maturity is a key cotton quality property,and its variability in a sample impacts fiber processing and dyeing performance.Currently,the maturity is determined by using established protocols in laborat...Background Fiber maturity is a key cotton quality property,and its variability in a sample impacts fiber processing and dyeing performance.Currently,the maturity is determined by using established protocols in laboratories under a controlled environment.There is an increasing need to measure fiber maturity using low-cost(in general less than $20000)and small portable systems.In this study,a laboratory feasibility was performed to assess the ability of the shortwave infrared hyperspectral imaging(SWIR HSI)technique for determining the conditioned fiber maturity,and as a comparison,a bench-top commercial and expensive(in general greater than $60000)near infrared(NIR)instrument was used.Results Although SWIR HSI and NIR represent different measurement technologies,consistent spectral characteristics were observed between the two instruments when they were used to measure the maturity of the locule fiber samples in seed cotton and of the well-defined fiber samples,respectively.Partial least squares(PLS)models were established using different spectral preprocessing parameters to predict fiber maturity.The high prediction precision was observed by a lower root mean square error of prediction(RMSEP)(<0.046),higher R_(p)^(2)(>0.518),and greater percentage(97.0%)of samples within the 95% agreement range in the entire NIR region(1000-2500 nm)without the moisture band at 1940 nm.Conclusion SWIR HSI has a good potential for assessing cotton fiber maturity in a laboratory environment.展开更多
Hyperspectral imaging technique is known as a promising non-destructive way for detecting plants diseases and pests.In most previous studies,the utilization of the whole spectrum or a large number of bands as well as ...Hyperspectral imaging technique is known as a promising non-destructive way for detecting plants diseases and pests.In most previous studies,the utilization of the whole spectrum or a large number of bands as well as the complexity of model structure severely hampers the application of the technique in practice.If a detection system can be established with a few bands and a relatively simple logic,it would be of great significance for application.This study established a method for identifying and discriminating three commonly occurring diseases and pests of wheat,i.e.,powdery mildew,yellow rust and aphid with a few specific bands.Through a comprehensive spectral analysis,only three bands at 570,680 and 750 nm were selected.A novel vegetation index namely Ratio Triangular Vegetation Index(RTVI)was developed for detecting anomalous areas on leaves.Then,the Support Vector Machine(SVM)method was applied to construct the discrimination model based on the spectral ratio analysis.The validating results suggested that the proposed method with only three spectral bands achieved a promising accuracy with the Overall Accuracy(OA)of 83%.With three bands from the hyperspectral imaging data,the three wheat diseases and pests were successfully detected and discriminated.A stepwise strategy including background removal,damage lesions recognition and stresses discrimination was proposed.The present work can provide a basis for the design of low cost and smart instruments for disease and pest detection.展开更多
Peach aphid is a common pest and hard to detect.This study employs hyperspectral imaging technology to identify early damage in green cabbage caused by peach aphid.Through principal component transformation and multip...Peach aphid is a common pest and hard to detect.This study employs hyperspectral imaging technology to identify early damage in green cabbage caused by peach aphid.Through principal component transformation and multiple linear regression analysis,the correlation relation between spectral characteristics and infestation stage is analyzed.Then,four characteristic wavelength selection methods are compared and optimal characteristic wavelengths subset is determined to be input for modelling.One linear algorithm and two nonlinear modelling algorithms are compared.Finally,support vector machine(SVM)model based on the characteristic wavelengths selected by multi-cluster feature selection(MCFS)acquires the highest identification accuracy,which is 98.97%.These results indicate that hyperspectral imaging technology have the ability to identify early peach aphid infestation stages on green cabbages.展开更多
Most methods for classifying hyperspectral data only consider the local spatial relation-ship among samples,ignoring the important non-local topological relationship.However,the non-local topological relationship is b...Most methods for classifying hyperspectral data only consider the local spatial relation-ship among samples,ignoring the important non-local topological relationship.However,the non-local topological relationship is better at representing the structure of hyperspectral data.This paper proposes a deep learning model called Topology and semantic information fusion classification network(TSFnet)that incorporates a topology structure and semantic information transmis-sion network to accurately classify traditional Chinese medicine in hyperspectral images.TSFnet uses a convolutional neural network(CNN)to extract features and a graph convolution network(GCN)to capture potential topological relationships among different types of Chinese herbal medicines.The results show that TSFnet outperforms other state-of-the-art deep learning classification algorithms in two different scenarios of herbal medicine datasets.Additionally,the proposed TSFnet model is lightweight and can be easily deployed for mobile herbal medicine classification.展开更多
Spectral unmixing helps to identify different components present in the spectral mixtures which occur in the uppermost layer of the area owing to the low spatial resolution of hyperspectral images.Most spectral unmixi...Spectral unmixing helps to identify different components present in the spectral mixtures which occur in the uppermost layer of the area owing to the low spatial resolution of hyperspectral images.Most spectral unmixing methods are globally based and do not consider the spectral variability among its endmembers that occur due to illumination,atmospheric,and environmental conditions.Here,endmember bundle extraction plays a major role in overcoming the above-mentioned limitations leading to more accurate abundance fractions.Accordingly,a two-stage approach is proposed to extract endmembers through endmember bundles in hyperspectral images.The divide and conquer method is applied as the first step in subset images with only the non-redundant bands to extract endmembers using the Vertex Component Analysis(VCA)and N-FINDR algorithms.A fuzzy rule-based inference system utilizing spectral matching parameters is proposed in the second step to categorize endmembers.The endmember with the minimum error is chosen as the final endmember in each specific category.The proposed method is simple and automatically considers endmember variability in hyperspectral images.The efficiency of the proposed method is evaluated using two real hyperspectral datasets.The average spectral angle and abundance angle are used to analyze the performance measures.展开更多
Accurate histopathology classification is a crucial factor in the diagnosis and treatment of Cholangiocarcinoma(CCA).Hyperspectral images(HSI)provide rich spectral information than ordinary RGB images,making them more...Accurate histopathology classification is a crucial factor in the diagnosis and treatment of Cholangiocarcinoma(CCA).Hyperspectral images(HSI)provide rich spectral information than ordinary RGB images,making them more useful for medical diagnosis.The Convolutional Neural Network(CNN)is commonly employed in hyperspectral image classification due to its remarkable capacity for feature extraction and image classification.However,many existing CNN-based HSI classification methods tend to ignore the importance of image spatial context information and the interdependence between spectral channels,leading to unsatisfied classification performance.Thus,to address these issues,this paper proposes a Spatial-Spectral Joint Network(SSJN)model for hyperspectral image classification that utilizes spatial self-attention and spectral feature extraction.The SSJN model is derived from the ResNet18 network and implemented with the non-local and Coordinate Attention(CA)modules,which extract long-range dependencies on image space and enhance spatial features through the Branch Attention(BA)module to emphasize the region of interest.Furthermore,the SSJN model employs Conv-LSTM modules to extract long-range depen-dencies in the image spectral domain.This addresses the gradient disappearance/explosion phenom-ena and enhances the model classification accuracy.The experimental results show that the pro-posed SSJN model is more efficient in leveraging the spatial and spectral information of hyperspec-tral images on multidimensional microspectral datasets of CCA,leading to higher classification accuracy,and may have useful references for medical diagnosis of CCA.展开更多
Hyperspectral images can easily discriminate different materials due to their fine spectral resolution.However,obtaining a hyperspectral image(HSI)with a high spatial resolution is still a challenge as we are limited ...Hyperspectral images can easily discriminate different materials due to their fine spectral resolution.However,obtaining a hyperspectral image(HSI)with a high spatial resolution is still a challenge as we are limited by the high computing requirements.The spatial resolution of HSI can be enhanced by utilizing Deep Learning(DL)based Super-resolution(SR).A 3D-CNNHSR model is developed in the present investigation for 3D spatial super-resolution for HSI,without losing the spectral content.The 3DCNNHSR model was tested for the Hyperion HSI.The pre-processing of the HSI was done before applying the SR model so that the full advantage of hyperspectral data can be utilized with minimizing the errors.The key innovation of the present investigation is that it used 3D convolution as it simultaneously applies convolution in both the spatial and spectral dimensions and captures spatial-spectral features.By clustering contiguous spectral content together,a cube is formed and by convolving the cube with the 3D kernel a 3D convolution is realized.The 3D-CNNHSR model was compared with a 2D-CNN model,additionally,the assessment was based on higherresolution data from the Sentinel-2 satellite.Based on the evaluation metrics it was observed that the 3D-CNNHSR model yields better results for the SR of HSI with efficient computational speed,which is significantly less than previous studies.展开更多
Recently,the autoencoder(AE)based method plays a critical role in the hyperspectral anomaly detection domain.However,due to the strong generalised capacity of AE,the abnormal samples are usually reconstructed well alo...Recently,the autoencoder(AE)based method plays a critical role in the hyperspectral anomaly detection domain.However,due to the strong generalised capacity of AE,the abnormal samples are usually reconstructed well along with the normal background samples.Thus,in order to separate anomalies from the background by calculating reconstruction errors,it can be greatly beneficial to reduce the AE capability for abnormal sample reconstruction while maintaining the background reconstruction performance.A memory‐augmented autoencoder for hyperspectral anomaly detection(MAENet)is proposed to address this challenging problem.Specifically,the proposed MAENet mainly consists of an encoder,a memory module,and a decoder.First,the encoder transforms the original hyperspectral data into the low‐dimensional latent representation.Then,the latent representation is utilised to retrieve the most relevant matrix items in the memory matrix,and the retrieved matrix items will be used to replace the latent representation from the encoder.Finally,the decoder is used to reconstruct the input hyperspectral data using the retrieved memory items.With this strategy,the background can still be reconstructed well while the abnormal samples cannot.Experiments conducted on five real hyperspectral anomaly data sets demonstrate the superiority of the proposed method.展开更多
基金supported by the Thailand Research Fund through the Royal Golden Jubilee Ph.D.Program(PHD/0225/2561)the Faculty of Engineering,Kamphaeng Saen Campus,Kasetsart University,Thailand。
文摘The adulteration concentration of palm kernel oil(PKO)in virgin coconut oil(VCO)was quantified using near-infrared(NIR)hyperspectral imaging.Nowadays,some VCO is adulterated with lower-priced PKO to reduce production costs,which diminishes the quality of the VCO.This study used NIR hyperspectral imaging in the wavelength region 900-1,650 nm to create a quantitative model for the detection of PKO contaminants(0-100%)in VCO and to develop predictive mapping.The prediction equation for the adulteration of VCO with PKO was constructed using the partial least squares regression method.The best predictive model was pre-processed using the standard normal variate method,and the coefficient of determination of prediction was 0.991,the root mean square error of prediction was 2.93%,and the residual prediction deviation was 10.37.The results showed that this model could be applied for quantifying the adulteration concentration of PKO in VCO.The prediction adulteration concentration mapping of VCO with PKO was created from a calibration model that showed the color level according to the adulteration concentration in the range of 0-100%.NIR hyperspectral imaging could be clearly used to quantify the adulteration of VCO with a color level map that provides a quick,accurate,and non-destructive detection method.
基金Ministry of Education,Culture,Sports,Science and Technology,Grant/Award Number:20K11867。
文摘By automatically learning the priors embedded in images with powerful modelling ca-pabilities,deep learning-based algorithms have recently made considerable progress in reconstructing the high-resolution hyperspectral(HR-HS)image.With previously collected large-amount of external data,these methods are intuitively realised under the full supervision of the ground-truth data.Thus,the database construction in merging the low-resolution(LR)HS(LR-HS)and HR multispectral(MS)or RGB image research paradigm,commonly named as HSI SR,requires collecting corresponding training triplets:HR-MS(RGB),LR-HS and HR-HS image simultaneously,and often faces dif-ficulties in reality.The learned models with the training datasets collected simultaneously under controlled conditions may significantly degrade the HSI super-resolved perfor-mance to the real images captured under diverse environments.To handle the above-mentioned limitations,the authors propose to leverage the deep internal and self-supervised learning to solve the HSI SR problem.The authors advocate that it is possible to train a specific CNN model at test time,called as deep internal learning(DIL),by on-line preparing the training triplet samples from the observed LR-HS/HR-MS(or RGB)images and the down-sampled LR-HS version.However,the number of the training triplets extracted solely from the transformed data of the observation itself is extremely few particularly for the HSI SR tasks with large spatial upscale factors,which would result in limited reconstruction performance.To solve this problem,the authors further exploit deep self-supervised learning(DSL)by considering the observations as the unlabelled training samples.Specifically,the degradation modules inside the network were elaborated to realise the spatial and spectral down-sampling procedures for transforming the generated HR-HS estimation to the high-resolution RGB/LR-HS approximation,and then the reconstruction errors of the observations were formulated for measuring the network modelling performance.By consolidating the DIL and DSL into a unified deep framework,the authors construct a more robust HSI SR method without any prior training and have great potential of flexible adaptation to different settings per obser-vation.To verify the effectiveness of the proposed approach,extensive experiments have been conducted on two benchmark HS datasets,including the CAVE and Harvard datasets,and demonstrate the great performance gain of the proposed method over the state-of-the-art methods.
文摘With limited number of labeled samples,hyperspectral image(HSI)classification is a difficult Problem in current research.The graph neural network(GNN)has emerged as an approach to semi-supervised classification,and the application of GNN to hyperspectral images has attracted much attention.However,in the existing GNN-based methods a single graph neural network or graph filter is mainly used to extract HSI features,which does not take full advantage of various graph neural networks(graph filters).Moreover,the traditional GNNs have the problem of oversmoothing.To alleviate these shortcomings,we introduce a deep hybrid multi-graph neural network(DHMG),where two different graph filters,i.e.,the spectral filter and the autoregressive moving average(ARMA)filter,are utilized in two branches.The former can well extract the spectral features of the nodes,and the latter has a good suppression effect on graph noise.The network realizes information interaction between the two branches and takes good advantage of different graph filters.In addition,to address the problem of oversmoothing,a dense network is proposed,where the local graph features are preserved.The dense structure satisfies the needs of different classification targets presenting different features.Finally,we introduce a GraphSAGEbased network to refine the graph features produced by the deep hybrid network.Extensive experiments on three public HSI datasets strongly demonstrate that the DHMG dramatically outperforms the state-ofthe-art models.
基金supported by the National Natural Science Foundation of China (Grant No.42101403)the National Key Researchand Development Program of China (Grant No.2017YFD0600404)。
文摘Although airborne hyperspectral data with detailed spatial and spectral information has demonstrated significant potential for tree species classification,it has not been widely used over large areas.A comprehensive process based on multi-flightline airborne hyperspectral data is lacking over large,forested areas influenced by both the effects of bidirectional reflectance distribution function(BRDF)and cloud shadow contamination.In this study,hyperspectral data were collected over the Mengjiagang Forest Farm in Northeast China in the summer of 2017 using the Chinese Academy of Forestry's LiDAR,CCD,and hyperspectral systems(CAF-LiCHy).After BRDF correction and cloud shadow detection processing,a tree species classification workflow was developed for sunlit and cloud-shaded forest areas with input features of minimum noise fraction reduced bands,spectral vegetation indices,and texture information.Results indicate that BRDF-corrected sunlit hyperspectral data can provide a stable and high classification accuracy based on representative training data.Cloud-shaded pixels also have good spectral separability for species classification.The red-edge spectral information and ratio-based spectral indices with high importance scores are recommended as input features for species classification under varying light conditions.According to the classification accuracies through field survey data at multiple spatial scales,it was found that species classification within an extensive forest area using airborne hyperspectral data under various illuminations can be successfully carried out using the effective radiometric consistency process and feature selection strategy.
基金supported in part by the National Natural Science Foundation of China(62276192)。
文摘Hyperspectral image super-resolution,which refers to reconstructing the high-resolution hyperspectral image from the input low-resolution observation,aims to improve the spatial resolution of the hyperspectral image,which is beneficial for subsequent applications.The development of deep learning has promoted significant progress in hyperspectral image super-resolution,and the powerful expression capabilities of deep neural networks make the predicted results more reliable.Recently,several latest deep learning technologies have made the hyperspectral image super-resolution method explode.However,a comprehensive review and analysis of the latest deep learning methods from the hyperspectral image super-resolution perspective is absent.To this end,in this survey,we first introduce the concept of hyperspectral image super-resolution and classify the methods from the perspectives with or without auxiliary information.Then,we review the learning-based methods in three categories,including single hyperspectral image super-resolution,panchromatic-based hyperspectral image super-resolution,and multispectral-based hyperspectral image super-resolution.Subsequently,we summarize the commonly used hyperspectral dataset,and the evaluations for some representative methods in three categories are performed qualitatively and quantitatively.Moreover,we briefly introduce several typical applications of hyperspectral image super-resolution,including ground object classification,urban change detection,and ecosystem monitoring.Finally,we provide the conclusion and challenges in existing learning-based methods,looking forward to potential future research directions.
基金supported by Light of West China(No.XAB2022YN10)Shaanxi Key Rsearch and Development Plan(No.2018ZDXM-SF-093)Shaanxi Province Key Industrial Innovation Chain(Nos.S2022-YF-ZDCXL-ZDLGY-0093,2023-ZDLGY-45).
文摘Sanxingdui cultural relics are the precious cultural heritage of humanity with high values of history,science,culture,art and research.However,mainstream analytical methods are contacting and detrimental,which is unfavorable to the protection of cultural relics.This paper improves the accuracy of the extraction,location,and analysis of artifacts using hyperspectral methods.To improve the accuracy of cultural relic mining,positioning,and analysis,the segmentation algorithm of Sanxingdui cultural relics based on the spatial spectrum integrated network is proposed with the support of hyperspectral techniques.Firstly,region stitching algorithm based on the relative position of hyper spectrally collected data is proposed to improve stitching efficiency.Secondly,given the prominence of traditional HRNet(High-Resolution Net)models in high-resolution data processing,the spatial attention mechanism is put forward to obtain spatial dimension information.Thirdly,in view of the prominence of 3D networks in spectral information acquisition,the pyramid 3D residual network model is proposed to obtain internal spectral dimensional information.Fourthly,four kinds of fusion methods at the level of data and decision are presented to achieve cultural relic labeling.As shown by the experiment results,the proposed network adopts an integrated method of data-level and decision-level,which achieves the optimal average accuracy of identification 0.84,realizes shallow coverage of cultural relics labeling,and effectively supports the mining and protection of cultural relics.
基金supported by grants from the Key Project of Liaoning Provincial Department of Education,China(LSNZD202005)。
文摘Rapid and large area acquisition of nitrogen(N)deficiency status is important for achieving the optimal fertilization of rice.Most existing studies,however,focus on the use of unmanned aerial vehicle(UAV)remote sensing to diagnose N nutrition in rice,while there are fewer studies on the quantitative description of the degree of N deficiency in rice,and the effects of the critical N concentration on the spectral changes in rice have rarely been explored.Therefore,based on the canopy spectral data obtained by remotely-sensed UAV hyperspectral images,the N content in rice was obtained through field sampling.The construction method of the rice curve for the northeastern critical N concentration was studied,and on this basis,N deficiency was determined.Taking the spectrum of the critical N concentration state as the standard spectrum,the spectral reflectivity data were transformed by the ratios and differences,and the feature extraction of the spectral data was carried out by the successive projections algorithm(SPA).Finally,by taking the characteristic band as the input variable and N deficiency as the output variable,a set of multivariate linear regression(MLR),long short-term memory(LSTM)inversion models based on extreme learning machine(ELM),and the nondominated sorting genetic algorithmⅢextreme learning machine(NSGA-Ⅲ-ELM)were constructed.The results showed two key aspects of this system:1)The correlation between the N deficiency data and original spectrum was poor,but the correlation between the N deficiency data and N deficiency could be improved by a difference change and ratio transformation;2)The inversion results based on the ratio spectrum and NSGA-Ⅲ-ELM algorithm were the best,as the R2values of the training set and validation set were 0.852 and 0.810,and the root mean square error(RMSE)values were 0.291 and 0.308,respectively.From the perspective of the spectral data,the inversion accuracy of the ratio spectrum was better than the accuracy of the original spectrum or difference spectrum.At the algorithm level,the model inversion results based on LSTM algorithms showed a serious overfitting phenomenon and poor inversion effect.The inversion accuracy based on the NSGA-Ⅲ-ELM algorithm was better than the accuracy of the MLR algorithm or the ELM algorithm.Therefore,the inversion model based on the ratio spectrum and NSGA-Ⅲ-ELM algorithm could effectively invert the N deficiency in rice and provide critical technical support for accurate topdressing based on the N status in the rice.
基金supported by the National Natural Science Foundation of China under Grant 62161160336Grant 41871245in part by the Belgium Vlaio project(AI ICON‐2021‐0599:Smart industrial spectral cameras via artificial intelligence).
文摘Most unsupervised or semisupervised hyperspectral anomaly detection(HAD)methods train background reconstruction models in the original spectral domain.However,due to the noise and spatial resolution limitations,there may be a lack of discrimination between backgrounds and anomalies.This makes it easy for the autoencoder to capture the lowlevel features shared between the two,thereby increasing the difficulty of separating anomalies from the backgrounds,which runs counter to the purpose of HAD.To this end,the authors map the original spectrums to the fractional Fourier domain(FrFD)and reformulate it as a mapping task in which restoration errors are employed to distinguish background and anomaly.This study proposes a novel frequency‐to‐spectrum mapping generative adversarial network for HAD.Specifically,the depth separable features of backgrounds and anomalies are enhanced in the FrFD.Due to the semisupervised approach,FTSGAN needs to learn the embedded features of the backgrounds,thus mapping and restoring them from the FrFD to the original spectral domain.This strategy effectively prevents the model from focussing on the numerical equivalence of input and output,and restricts the ability of FTSGAN to restore anomalies.The comparison and analysis of the experiments verify that the proposed method is competitive.
基金the Deanship of Scientific Research at King Khalid University for funding this work through Large Groups Project under Grant Number(25/43)Princess Nourah bint Abdulrahman University Researchers Supporting Project Number(PNURSP2022R303)Princess Nourah bint Abdulrahman University,Riyadh,Saudi Arabia.The authors would like to thank the Deanship of Scientific Research at Umm Al-Qura University for supporting this work by Grant Code:22UQU4340237DSR28.
文摘Hyperspectral remote sensing/imaging spectroscopy is a novel approach to reaching a spectrum from all the places of a huge array of spatial places so that several spectral wavelengths are utilized for making coherent images.Hyperspectral remote sensing contains acquisition of digital images from several narrow,contiguous spectral bands throughout the visible,Thermal Infrared(TIR),Near Infrared(NIR),and Mid-Infrared(MIR)regions of the electromagnetic spectrum.In order to the application of agricultural regions,remote sensing approaches are studied and executed to their benefit of continuous and quantitativemonitoring.Particularly,hyperspectral images(HSI)are considered the precise for agriculture as they can offer chemical and physical data on vegetation.With this motivation,this article presents a novel Hurricane Optimization Algorithm with Deep Transfer Learning Driven Crop Classification(HOADTL-CC)model onHyperspectralRemote Sensing Images.The presentedHOADTL-CC model focuses on the identification and categorization of crops on hyperspectral remote sensing images.To accomplish this,the presentedHOADTL-CC model involves the design ofHOAwith capsule network(CapsNet)model for generating a set of useful feature vectors.Besides,Elman neural network(ENN)model is applied to allot proper class labels into the input HSI.Finally,glowworm swarm optimization(GSO)algorithm is exploited to fine tune the ENNparameters involved in this article.The experimental result scrutiny of the HOADTL-CC method can be tested with the help of benchmark dataset and the results are assessed under distinct aspects.Extensive comparative studies stated the enhanced performance of the HOADTL-CC model over recent approaches with maximum accuracy of 99.51%.
基金supported by the National Natural Science Foundation of China under Grant No.61806138Key R&D program of Shanxi Province(High Technology)under Grant No.201903D121119Science and Technology Development Foundation of the Central Guiding Local under Grant No.YDZJSX2021A038.
文摘Compressed sensing(CS),as an efficient data transmission method,has achieved great success in the field of data transmission such as image,video and text.It can robustly recover signals from fewer Measurements,effectively alleviating the bandwidth pressure during data transmission.However,CS has many shortcomings in the transmission of hyperspectral image(HSI)data.This work aims to consider the application of CS in the transmission of hyperspectral image(HSI)data,and provides a feasible research scheme for CS of HSI data.HSI has rich spectral information and spatial information in bands,which can reflect the physical properties of the target.Most of the hyperspectral image compressed sensing(HSICS)algorithms cannot effectively use the inter-band information of HSI,resulting in poor reconstruction effects.In this paper,A three-stage hyperspectral image compression sensing algorithm(Three-stages HSICS)is proposed to obtain intra-band and inter-band characteristics of HSI,which can improve the reconstruction accuracy of HSI.Here,we establish a multi-objective band selection(Mop-BS)model,amulti-hypothesis prediction(MHP)model and a residual sparse(ReWSR)model for HSI,and use a staged reconstruction method to restore the compressed HSI.The simulation results show that the three-stage HSICS successfully improves the reconstruction accuracy of HSICS,and it performs best among all comparison algorithms.
基金supported by the National Natural Science Foundation of China under Grant 62161160336 and Grant 42030111.
文摘Recently,deep learning has achieved considerable results in the hyperspectral image(HSI)classification.However,most available deep networks require ample and authentic samples to better train the models,which is expensive and inefficient in practical tasks.Existing few‐shot learning(FSL)methods generally ignore the potential relationships between non‐local spatial samples that would better represent the underlying features of HSI.To solve the above issues,a novel deep transformer and few‐shot learning(DTFSL)classification framework is proposed,attempting to realize fine‐grained classification of HSI with only a few‐shot instances.Specifically,the spatial attention and spectral query modules are introduced to overcome the constraint of the convolution kernel and consider the information between long‐distance location(non‐local)samples to reduce the uncertainty of classes.Next,the network is trained with episodes and task‐based learning strategies to learn a metric space,which can continuously enhance its modelling capability.Furthermore,the developed approach combines the advantages of domain adaptation to reduce the variation in inter‐domain distribution and realize distribution alignment.On three publicly available HSI data,extensive experiments have indicated that the proposed DT‐FSL yields better results concerning state‐of‐the‐art algorithms.
基金supported partially by the USDA-ARS Research Project#6054-44000-080-00D.
文摘Background Fiber maturity is a key cotton quality property,and its variability in a sample impacts fiber processing and dyeing performance.Currently,the maturity is determined by using established protocols in laboratories under a controlled environment.There is an increasing need to measure fiber maturity using low-cost(in general less than $20000)and small portable systems.In this study,a laboratory feasibility was performed to assess the ability of the shortwave infrared hyperspectral imaging(SWIR HSI)technique for determining the conditioned fiber maturity,and as a comparison,a bench-top commercial and expensive(in general greater than $60000)near infrared(NIR)instrument was used.Results Although SWIR HSI and NIR represent different measurement technologies,consistent spectral characteristics were observed between the two instruments when they were used to measure the maturity of the locule fiber samples in seed cotton and of the well-defined fiber samples,respectively.Partial least squares(PLS)models were established using different spectral preprocessing parameters to predict fiber maturity.The high prediction precision was observed by a lower root mean square error of prediction(RMSEP)(<0.046),higher R_(p)^(2)(>0.518),and greater percentage(97.0%)of samples within the 95% agreement range in the entire NIR region(1000-2500 nm)without the moisture band at 1940 nm.Conclusion SWIR HSI has a good potential for assessing cotton fiber maturity in a laboratory environment.
基金subsidized by National Natural Science Foundation of China(Grant No.42071420)External Cooperation Program of the Chinese Academy of Sciences(183611KYSB20200080)+1 种基金National Key R&D Program of China(2019YFE0125300)Beijing Nova Program of Science and Technology(Z191100001119089).
文摘Hyperspectral imaging technique is known as a promising non-destructive way for detecting plants diseases and pests.In most previous studies,the utilization of the whole spectrum or a large number of bands as well as the complexity of model structure severely hampers the application of the technique in practice.If a detection system can be established with a few bands and a relatively simple logic,it would be of great significance for application.This study established a method for identifying and discriminating three commonly occurring diseases and pests of wheat,i.e.,powdery mildew,yellow rust and aphid with a few specific bands.Through a comprehensive spectral analysis,only three bands at 570,680 and 750 nm were selected.A novel vegetation index namely Ratio Triangular Vegetation Index(RTVI)was developed for detecting anomalous areas on leaves.Then,the Support Vector Machine(SVM)method was applied to construct the discrimination model based on the spectral ratio analysis.The validating results suggested that the proposed method with only three spectral bands achieved a promising accuracy with the Overall Accuracy(OA)of 83%.With three bands from the hyperspectral imaging data,the three wheat diseases and pests were successfully detected and discriminated.A stepwise strategy including background removal,damage lesions recognition and stresses discrimination was proposed.The present work can provide a basis for the design of low cost and smart instruments for disease and pest detection.
基金supported by China National Key Research and Development Program(No.2016YFD0700304)Shandong Natural Science Foundation Youth Program(No.ZR2021QC216)Agricultural Scientific and Technological Innovation Project of Shandong Academy of Agricultural Science(No.CXGC2023A34)。
文摘Peach aphid is a common pest and hard to detect.This study employs hyperspectral imaging technology to identify early damage in green cabbage caused by peach aphid.Through principal component transformation and multiple linear regression analysis,the correlation relation between spectral characteristics and infestation stage is analyzed.Then,four characteristic wavelength selection methods are compared and optimal characteristic wavelengths subset is determined to be input for modelling.One linear algorithm and two nonlinear modelling algorithms are compared.Finally,support vector machine(SVM)model based on the characteristic wavelengths selected by multi-cluster feature selection(MCFS)acquires the highest identification accuracy,which is 98.97%.These results indicate that hyperspectral imaging technology have the ability to identify early peach aphid infestation stages on green cabbages.
基金supported by the National Natural Science Foundation of China(No.62001023)Beijing Natural Science Foundation(No.JQ20021)。
文摘Most methods for classifying hyperspectral data only consider the local spatial relation-ship among samples,ignoring the important non-local topological relationship.However,the non-local topological relationship is better at representing the structure of hyperspectral data.This paper proposes a deep learning model called Topology and semantic information fusion classification network(TSFnet)that incorporates a topology structure and semantic information transmis-sion network to accurately classify traditional Chinese medicine in hyperspectral images.TSFnet uses a convolutional neural network(CNN)to extract features and a graph convolution network(GCN)to capture potential topological relationships among different types of Chinese herbal medicines.The results show that TSFnet outperforms other state-of-the-art deep learning classification algorithms in two different scenarios of herbal medicine datasets.Additionally,the proposed TSFnet model is lightweight and can be easily deployed for mobile herbal medicine classification.
文摘Spectral unmixing helps to identify different components present in the spectral mixtures which occur in the uppermost layer of the area owing to the low spatial resolution of hyperspectral images.Most spectral unmixing methods are globally based and do not consider the spectral variability among its endmembers that occur due to illumination,atmospheric,and environmental conditions.Here,endmember bundle extraction plays a major role in overcoming the above-mentioned limitations leading to more accurate abundance fractions.Accordingly,a two-stage approach is proposed to extract endmembers through endmember bundles in hyperspectral images.The divide and conquer method is applied as the first step in subset images with only the non-redundant bands to extract endmembers using the Vertex Component Analysis(VCA)and N-FINDR algorithms.A fuzzy rule-based inference system utilizing spectral matching parameters is proposed in the second step to categorize endmembers.The endmember with the minimum error is chosen as the final endmember in each specific category.The proposed method is simple and automatically considers endmember variability in hyperspectral images.The efficiency of the proposed method is evaluated using two real hyperspectral datasets.The average spectral angle and abundance angle are used to analyze the performance measures.
基金supported by National Natural Science Foundation of China(No.62101040).
文摘Accurate histopathology classification is a crucial factor in the diagnosis and treatment of Cholangiocarcinoma(CCA).Hyperspectral images(HSI)provide rich spectral information than ordinary RGB images,making them more useful for medical diagnosis.The Convolutional Neural Network(CNN)is commonly employed in hyperspectral image classification due to its remarkable capacity for feature extraction and image classification.However,many existing CNN-based HSI classification methods tend to ignore the importance of image spatial context information and the interdependence between spectral channels,leading to unsatisfied classification performance.Thus,to address these issues,this paper proposes a Spatial-Spectral Joint Network(SSJN)model for hyperspectral image classification that utilizes spatial self-attention and spectral feature extraction.The SSJN model is derived from the ResNet18 network and implemented with the non-local and Coordinate Attention(CA)modules,which extract long-range dependencies on image space and enhance spatial features through the Branch Attention(BA)module to emphasize the region of interest.Furthermore,the SSJN model employs Conv-LSTM modules to extract long-range depen-dencies in the image spectral domain.This addresses the gradient disappearance/explosion phenom-ena and enhances the model classification accuracy.The experimental results show that the pro-posed SSJN model is more efficient in leveraging the spatial and spectral information of hyperspec-tral images on multidimensional microspectral datasets of CCA,leading to higher classification accuracy,and may have useful references for medical diagnosis of CCA.
基金Deanship of Scientific Research at King Khalid University for funding this work through large group Research Project under Grant Number RGP2/80/44.
文摘Hyperspectral images can easily discriminate different materials due to their fine spectral resolution.However,obtaining a hyperspectral image(HSI)with a high spatial resolution is still a challenge as we are limited by the high computing requirements.The spatial resolution of HSI can be enhanced by utilizing Deep Learning(DL)based Super-resolution(SR).A 3D-CNNHSR model is developed in the present investigation for 3D spatial super-resolution for HSI,without losing the spectral content.The 3DCNNHSR model was tested for the Hyperion HSI.The pre-processing of the HSI was done before applying the SR model so that the full advantage of hyperspectral data can be utilized with minimizing the errors.The key innovation of the present investigation is that it used 3D convolution as it simultaneously applies convolution in both the spatial and spectral dimensions and captures spatial-spectral features.By clustering contiguous spectral content together,a cube is formed and by convolving the cube with the 3D kernel a 3D convolution is realized.The 3D-CNNHSR model was compared with a 2D-CNN model,additionally,the assessment was based on higherresolution data from the Sentinel-2 satellite.Based on the evaluation metrics it was observed that the 3D-CNNHSR model yields better results for the SR of HSI with efficient computational speed,which is significantly less than previous studies.
基金supported in part by the National Natural Science Foundation of China under Grant 62076199in part by the Open Research Fund of Beijing Key Laboratory of Big Data Technology for Food Safety under Grant BTBD‐2020KF08Beijing Technology and Business University,and in part by the Key R&D project of Shaan'xi Province under Grant 2021GY‐027 and 2022ZDLGY01‐03.
文摘Recently,the autoencoder(AE)based method plays a critical role in the hyperspectral anomaly detection domain.However,due to the strong generalised capacity of AE,the abnormal samples are usually reconstructed well along with the normal background samples.Thus,in order to separate anomalies from the background by calculating reconstruction errors,it can be greatly beneficial to reduce the AE capability for abnormal sample reconstruction while maintaining the background reconstruction performance.A memory‐augmented autoencoder for hyperspectral anomaly detection(MAENet)is proposed to address this challenging problem.Specifically,the proposed MAENet mainly consists of an encoder,a memory module,and a decoder.First,the encoder transforms the original hyperspectral data into the low‐dimensional latent representation.Then,the latent representation is utilised to retrieve the most relevant matrix items in the memory matrix,and the retrieved matrix items will be used to replace the latent representation from the encoder.Finally,the decoder is used to reconstruct the input hyperspectral data using the retrieved memory items.With this strategy,the background can still be reconstructed well while the abnormal samples cannot.Experiments conducted on five real hyperspectral anomaly data sets demonstrate the superiority of the proposed method.
