最新刊期
卷 24 ， 期 4 ， 2020
-
摘要：Hyperspectral imaging technology provides high spectral resolution information of objects on the Earth. Hyperion and Compact High-Resolution Imaging Spectrometer have been the main sources of spaceborne hyperspectral data in the past few decades. However, the quality and quantity of the data cannot totally meet the challenging requirements of various applications. In this study, we design and present a visible and short-wave infrared hyperspectral imager called the Advanced Hyperspectral Imager (AHSI), which is one of the six payloads on China's GF-5 satellite launched on May 9, 2018. It is the first spaceborne hyperspectral sensor that utilizes convex grating spectrophotometry and an improved three concentric-mirror (Offner) configuration. We introduce the basic structure and imaging principle of AHSI.Ground object lights are reflected into an off-axis three-mirror telescope using pointing mirror. The lights are focused on a Field Of View (FOV) separator and then split into two parts. One part enters a visible/near-infrared (VNIR) spectrometer, and the other goes into a short-wave infrared (SWIR) spectrometer. Compound light is dispersed into a series of monochromatic light through convex grating and focused on a detector using a mirror with an improved Offner structure. A VNIR detector is a back-illuminated frame transfer charge-coupled device with a size of 2048×300 pixels. A SWIR detector is a HgCdTe focal plane cooled at 110 K, and has a size of 2048×512 pixels formed by four infrared focal plane arrays of the same size (512×512), with a staggering spatial arrangement. During calibration, sunlight is reflected into an optical system using a diffuse reflectance panel. The absolute radiation response of AHSI is calibrated with a solar diffuse reflection signal, and the degradation of the diffuse reflectance panel is corrected with a ratioing radiometer. The central wavelength and bandwidth are calibrated with onboard LED calibration component and solar atmospheric absorption profile, respectively (O2 adsorption peaks at 760 and 1260 nm).The main performance metrics of the AHSI are obtained and validated via well-designed on-orbit experiments. It has a spectral range of 400–2500 nm, with spectral resolutions of higher than 5 nm in the VNIR and 10 nm in the SWIR, respectively. The swath width is 60 km, and the spatial resolution is 30 m. Compared with Hyperion, which is a classical hyperspectral instrument, the imaging width of the AHSI is superior (eight times), the number of spectral bands has increased by hundreds, and the Signal-to-Noise Ratio (SNR) has improved by four times. Compared with the instruments recently developed/planned by Germany, Italy, Indian, and Japan, the AHSI has competitive performance and has evident advantages in terms of swath width and number of spectral channels. The AHSI has achieved the widest swath width and broadest spectral range to shorten the revisit time, improve the observation and monitoring efficiency, and refine the Earth’s observation.As one of the six main payloads on GF-5 satellite, the AHSI demonstrates the development of spaceborne hyperspectral imaging technology in China. It features a large FOV telescope, a low-distortion large flat-field fine spectrometer, a large-size infrared focal plane detector, a long-life large cooling capacity cryocooler, a high-precision calibration system, and a high-precision image compensation mechanism. The AHSI has outstanding capability of detecting and identifying ground objects, making it suitable to precision applications, such as ecological environment monitoring, land and resource survey, and oil/gas exploration.关键词：GF-5 satellite;hyperspectral remote sensing;hyperspectral imager;wide swath width;low distortion- 1039
- |
- 293
- |
- 17
发布时间：2020-08-31 -
摘要：The advanced hyperspectral imager of GF-5 is a high spectral resolution camera, and it has outstanding recognition and classification ability for complex features and environments. The high accuracy geometric calibration is the key factor for the geometrical quality of satellite imagery. The geometric calibration model based on the pointing angle of the probe is used in this paper. The internal and external calibration parameters are solved step by step and some typical images are selected for experimental verification. The results indicated that the absolute positioning accuracy of AHSI can be stably better than 60 m (2 pixels) and the internal positioning accuracy is better than 0.5 pixels, the bands registration accuracy can be better than 0.3 pixels.关键词：GF-5 satellite;hyperspectral camera;strict geometric model;geometric calibration;positioning accuracy- 774
- |
- 192
- |
- 4
发布时间：2020-08-31 -
摘要：The Advanced Hyperspectral Imager (AHSI), one of the six payloads on China’s GF-5 satellite that was launched on May 9, 2018, is a visible/short-wave infrared hyperspectral imager built by Shanghai Institute of Technical Physics (SITP), Chinese Academy of Sciences (CAS). In order to verify and validate the on-orbit radiometric performance of AHSI, and improve quantitative application of the AHSI data, the instrument conducts on-orbit radiometric characterizations from June 2018 to April 2019. Solar calibration and vicarious calibration were both utilized.Observational and calibration data of well-known calibration sites, such as, Dunhuang and Baotou calibration field in China, were acquired and analyzed to determine whether the pre-flight characterization was still applicable to on-orbit operations. These data was used to measure specific instrument parameters, such as, in corresponding order, (1) Signal to Noise Ratio (SNR), (2) relative and absolute radiometric calibration accuracy, (3) uncertainty of on orbit absolute radiometric calibration, (4) validation accuracy of site absolute radiometric calibration, (5) dynamic range and (6) radiometric performance stability. The definition of these parameters as well as the acquisition of the coefficients of relative and absolute radiometric calibration has been illustrated.Results show that the SNR of AHSI reaches up to 700 at a wavelength of about 500 nm, and 500 at a wavelength of about 1500 nm, the error of relative and absolute radiometric calibration are smaller than 0.5% and 3%, respectively, the uncertainty of on orbit absolute radiometric calibration is within 2.59%(VNIR) and 2.68%(SWIR), the error of the validation of site absolute radiometric calibration is smaller than 5%, and the dynamic range can be fined with 256 gain. In particular, with the image motion compensating, the integration time can be prolonged and the SNR can be further increased, therefore, object/region with weak signal, such as, the sea/lake, the Antarctic and the Arctic can also be observed. In addition, the variation of the dark level, SNR, and the coefficients of relative radiometric calibration from September 2018 to March 2019 are all much less than 1%, meaning that the radiometric response and noise value are stable, which in turn reflects that the radiometric performance of AHSI is stable.The on orbit characterization of GF-5/AHSI lasted for seven months and with continuing assessment of the instrument throughout the first year of operations. As it can be seen, the AHSI has good and stable on-orbit radiometric performance, i.e., high SNR, high calibration accuracy, and high consistency. Therefore, the data of AHSI is highly reliable and can be used in quantitative applications.关键词：remote sensing;GF-5 satellite;hyperspectral imager;on-orbit measurements;instrument radiometric performance;signal to noise ratio;radiometric calibration;validation experiment;dynamic range- 789
- |
- 241
- |
- 12
发布时间：2020-08-31 -
摘要：GF-5 satellite is an important scientific research satellite in China’s high-resolution projects. It is also the first full-spectrum hyperspectral satellite in the world to simultaneously observe the atmosphere and land. GF-5 satellite can meet the urgent needs of China’s environmental monitoring, resource exploration, disaster prevention and mitigation, and other industries. However, similar to many hyperspectral satellite data, random band noises are found in some of its imaging data, thereby reducing the quality of data to a certain extent. The large data width, high spatial resolution, rich detail texture, and heterogeneity of the terrain also establish high requirements for strip removal. In this study, a method of image strip removal considering land–water differences is proposed to robustly remove strip noise in data, restore the real radiation information of the surface, and improve the application value of the GF-5 hyperspectral data.In the proposed algorithm, we use a computationally efficient moment matching algorithm as the basic framework, with the idea of separate treatment between water and land, and combined it with an optimized statistical strategy to obtain many reliable statistical results, and then use 1D variational filtering to obtain a reliable statistical reference value. The moment matching algorithm is used to correct the water and land areas for effectively removing the complex strip noise in the image.Experiments on the L1 data of the GF-5 hyperspectral data without strip preprocessing show that the proposed method can be robustly removed compared with the traditional moment matching, histogram matching, and wavelet Fourier joint filtering methods. Stripe noise in the data, especially in complex scenes, has improved removal. Therefore, the proposed method can effectively solve the high-resolution GF-5 hyperspectral data radiation degradation caused by strip noise, improve the data quality, and utilize its advantages in resource, environment, and ecological applications.This study proposes a global moment matching method based on 1D variational filtering guidance to address the band noise in GF-5 hyperspectral image data. This algorithm is based on the moment matching algorithm and uses different statistics between water and land to overcome the unreliable statistical characteristics of the bands in heterogeneous regions. 1D variational filtering technology is used to obtain water and land regions. Domains have accurate statistical reference values. The experimental results show that the proposed method can robustly remove band noise in data, effectively solve the problem of radiation degradation caused by high-score band noise, and improve the data quality. In the next step, the abundant spectral dimension information of GF-5 hyperspectral image data is utilized to correct the image data polluted by noise and improve the strip noise removal result.关键词：stripe noise;one-dimensional filtering;variation;moment matching method;GF-5 hyperspectral image;image restoration;remote sensing of hyperspectral imagery- 649
- |
- 117
- |
- 1
发布时间：2020-08-31 -
摘要：Thin clouds widely exist in the visible bands of GaoFen-5 visible-shortwave infrared Advanced HyperSpectral Imager (AHSI),thereby degrading the data quality. In this study, a thin cloud correction method based on statistical information and scattering model was proposed to correct the clouds and restore the surface information with high fidelity.The proposed method combines the statistical information between two adjacent visible bands and the atmospheric scattering model among visible bands. Two steps are included, that is estimating the Relative Cloud Radiance (RCR) and calculating the absolute radiance for different visible bands.For a pair of adjacent visible bands, the ground radiance of clear pixels is highly and linearly correlated. However, the linear relationship deviates when the pixel is contaminated by clouds. The stronger the cloud contamination is, the larger the deviation will be. Therefore, the RCR can be accurately estimated using the two adjacent visible bands of the AHSI. Relying on the RCR, different strategies were utilized to calculate the absolute cloud radiance for different visible bands in accordance with their scatter properties. A hierarchical dark object strategy was used to calculate the cloud radiance for the bands in blue and green spectral regions. A scattering model was adopted for the bands in red spectral region. The cloud-free results can be obtained by subtracting the band-varied cloud radiance from cloudy images when the absolute cloud radiance of different visible bands were calculated.Synthetic and real experiments were conducted to validate the effectiveness of the proposed method through qualitative and quantitative analyses. In the synthetic experiments, the cloud contamination in visible bands can be completely cleared , and the results are closest to the ground truth compared with the two other methods. The values of root-mean-square error, mean absolute error, and spectral angle values are only 1.9891, 1.6822, and 0.4901,respectively, which are smaller than those of the compared methods. In the real experiments, the thin clouds in various scenes can be completely corrected, whereas the spectral characteristics of the clear regions are relatively maintained. The scores of the Q-index, structural similarity index, and universal quality index showed that the proposed method achieve the best performance in most cases compared with the two other methods. All the comparisons indicated the proposed method has superior cloud correction ability for different scenes.This study proposed a thin cloud correction method for the GF-5 AHSI visible data. The highly linear correlation between the two adjacent visible bands can be used to accurately estimate the RCR. Various strategies were utilized to calculate cloud radiance in accordance with the scattering properties of different bands, wherea hierarchical dark object strategy was used for the bands in blue and green spectral regions, whereas a scattering model strategy was used for the bands in red spectral region. The thin clouds in the visible bands of AHSI data can be completely corrected by combining the statistical information and scattering model. For various scenes, the results achieve the best performance compared with the compared methods in terms ofqualitative and quantitative aspects.关键词：remote sensing;GF-5 AHSI data;visible images;statistical information;scattering model;thin cloud correction- 551
- |
- 200
- |
- 2
发布时间：2020-08-31 -
摘要：This study proposed a multisensor image fusion solution for the GF-5/GF-1 spatial-spectral fusion with large spatial resolution ratio. We aimed to obtain the fused image through step-by-step fusion of multisensor remote sensing images. A unified fusion framework for multisensor image fusion was derived on the basis of step-by-step fusion theory. An integrated multisensor image fusion method based on multiresolution analysis theory was proposed in accordance with the unified framework. The proposed method can overcome the difficulty of integrating complementary high spatial and spectral information of multisource images under high spatial resolution ratio. In the proposed method, a modulation transfer function was applied to separate the spatial (high frequency) and spectral components (low frequency) of multisource images. The fusion weight was constructed by comprehensively considering the relationship between multisensor high spatial resolution images and high spectral resolution images and the relationship among the spectral bands of the high spectral resolution image. Fused images with the highest spatial and spectral resolutions can be obtained. The GF-1 panchromatic, GF-1 multispectral, and GF-5 hyperspectral images were used in the experiments. Experimental results show that the proposed multisensor spatial–spectral fusion can effectively integrate the complementary spatial and spectral information to obtain the comparative fused results.关键词：remote sensing;GF-5 satellite;spatial-spectral fusion;large spatial resolution difference;multi-sensor- 717
- |
- 258
- |
- 11
发布时间：2020-08-31 -
摘要：The core of unmixing in hyperspectral images is to determine the mathematical form of the spectral mixing model in accordance with the radiative transmission characteristics of ground features and then obtain the endmember spectrum and abundance results. A nonlinear model is suitable for real mixed scenes because of the complexity of feature scenes. The accuracy of unmixing can be significantly improved by combining the advantages of autoencoder in internal structure data mining and feature extraction. However, this method cannot consider the collinearity of the model, resultingin overfittingand is sensitive to noise. This study proposed an unsupervised Enhanced Nonlinear AutoEncoder (ENAE) method. The introduction of endmember regularization reduces the collinearity between endmembers, thereby improving the accuracy of hyperspectral unmixing.The implementation steps of the ENAE method include two phases, where the first phaseis the initialization of the network structure, and the second phase is the nonlinear unmixing. The initialization phase determinesthe number of nodes of the encoder and initial value of endmember and abundance, and the nonlinear unmixing phase mainly realizesthe minimization of the loss function. The initialization of endmember and abundance aims to rapidly make the loss function converge. The objective function of the ENAE method includes the mean square error between the reconstructed and original images and endmember regularization. L2 regularization is used to constrain the weight of endmembers for enabling the ENAE to learn the nonlinear effect in the nonlinear mixing model. In the entire network iteration, the ENAE method is a self-learning process and does not require the participation of prior knowledge. Therefore, the ENAE method is an unsupervised nonlinear unmixing method.Experiments are conducted to validate the effectiveness of the proposed method. The experimental dataset includes simulation, urban real, and GF-5 satellite data. Three accuracy evaluation indices, namely, spectral angle distance, root mean square error of abundance, and image reconstruction error, are used to evaluate the effect of unmixing performance. Compared with the traditional nonlinear unmixing method, the deep learning method is superior in terms of endmember extraction and abundance estimation. The ENAE method can obtain high unmixing accuracy in the deep learning method, thereby provingthe effectiveness and robustness of the proposed method.The collinearity problem between endmembers is reduced by introducing the endmember regularization constraint in the autoencoder, thereby improving the accuracy of unmixing in hyperspectral images. In future work, we will introduce noise reduction, sparsity, and spatial information to improve the method and focus on obtaining the actual value of the pure ground spectrum to study the unmixing algorithm in hyperspectral images, which will be valuable in improving the application capabilities of hyperspectral remote sensing satellites in China. With the development of deep learning method interpretability research, exploring nonlinear unmixing methods withmany explanatory and physical meanings will be investigated forhyperspectral images withmixed pixel.关键词：hyperspectral remote sensing;GF-5 satellite;autoencoder;mixed pixel;nonlinear unmixing- 574
- |
- 137
- |
- 6
发布时间：2020-08-31 -
摘要：Hyperspectral unmixing, as a crucial preprocessing step for many hyperspectral applications, refers to the process of decomposing an image into a set of endmembers and corresponding abundance matrices. Nonnegative Matrix Factorization (NMF) has been widely utilized in hyperspectral unmixing because of its simplicity and effectiveness, whereas traditional NMF exits the local minimum problem. Various modified NMFs have been proposed to address such problem. Deep NMF has shown good performance in feature extraction using a multilayer NMF model.A novel deep NMF algorithm called reweighted sparsity regularized deep NMF with total variation (RSDNMF-TV) is proposed in this study by integrating the total variation and reweighted sparsity with deep layers. First, the deep NMF is obtained by extending the traditional single-layer NMF to the multilayer with pretraining and fine-tuning stages. The former stage pretrains all factors layer by layer, and the latter reduces the decomposition error. Second, a weighted sparse regularizer is integrated into the deep NMF model by sparing the abundance matrix, and its weights are adaptively updated in accordance with the abundance matrix. Finally, the total variation is introduced to improve the piecewise smoothness of abundance maps. In this study, gradient descent method is implemented for the multiplicative update.The experimental results obtained on simulated and real data sets confirm the effectiveness of the proposed method. For real data sets, we utilized the well-known AVIRIS Cuprite and GF-5 data sets. Six other algorithms, namely, total variation regularized reweighted sparse NMF, spatial group sparsity regularized NMF, minimum-volume CNMF, L1/2 sparsity CNMF, multilayer NMF, and VCA-FCLS, were used for comparison. The results on the three hyperspectral datasets show that the proposed unmixing method can outperform other algorithms.In summary, the proposed algorithm achieves improved performance with strong robustness and denoising ability, especially on images with low signal-to-noise ratio. However, RSDNMF-TV has several limitations, which are summarized as follows: (1) the deep NMF has higher computational complexity compared with the traditional NMF. (2) The proposed method only utilizes the prior knowledge of abundance and ignores the constraints on endmembers, such as spectral variability. (3) Parameter β is relevant to image’s spatial autocorrelation, and the adaptive selection of the parameter remains challenging. Future work will focus on solving these problems.关键词：remote sensing;hyperspectral unmixing;deep learning;deep nonnegative matrix factorization;reweighted sparsity;total variation- 715
- |
- 151
- |
- 3
发布时间：2020-08-31 -
摘要：GF-5 is the first full-spectrum hyperspectral satellite used to achieve comprehensive observations of the atmosphere and land. The hyperspectral sensors on the GF-5 have high spectral resolution and wide coverage. However, labeling all the materials on these wide ranges is extremely difficult in practical applications. Hyperspectral classification is extremely difficult when the number of labeled samples is limited or no labeled sample is available. In this study, we aim to present an effective unsupervised domain adaptation technique that uses labeled pixels in the existing old domain(source domain)to classify the scenes with limited or no labeled pixels (target domain).Correlation alignment (CORAL) is a simple and effective domain adaptation method. However, the covariance computation in CORAL may be inaccurate in the case of limited training samples.We propose a new CORAL algorithm on the basis of a sparse matrix transform technique (CORAL-SMT) to solve this problem. The proposed method first uses the sparse matrix transform technique to estimate the covariance matrices of the source and target domains and then performs the CORAL between the estimated covariance matrices. The SMT method can obtain an accurate covariance estimator, which is constantly positive and definite.In the experiment, we compare the proposed CORAL-SMT with some classical domain adaptation methods, such as subspace alignment, principal component analysis, CORAL, transfer component analysis, geodesic flow kernel, and information the oretical learning. After domain adaptation, we use the nearest neighbor and support vector machine as classifiers to classify the unlabeled data in the target domain. Two GF-5 hyperspectral datasets, namely, Huanghekou and Yancheng datasets, are used to evaluate the performance of different methods. Experimental results demonstrate the effectiveness of the proposed method compared with subspace-based alignment methods and CORAL.The GF-5 data have good spectral discriminative ability. In the case of bias sampling, the performance of classifying the target samples on the basis of the source model is acceptable. The distribution difference between source and target domains is decreased, and the classification performance is intensively improved using the domain adaptation technique. The SMT technique can improve the covariance estimation,thereby benefiting the following domain adaptation.关键词：remote sensing;GF-5 satellite;hyperspectral image;domain adaptation;classification;correlation alignment;sparse matrix transform- 670
- |
- 278
- |
- 5
发布时间：2020-08-31 -
摘要：Hyperspectral images with high spectral resolution contain hundreds and even thousands of spectral bands and convey abundant spectral information to distinguish subtle spectral differences, especially between similar materials, thereby providing unique advantage for target detection. At the same time, hyperspectral images cause large number of bands and redundant information between adjacent bands. The high-dimensional data structure frequently reduces the separability between the anomaly and background classes of hyperspectral images. This study proposed a Band Selection-based Collaborative Representation (BSCR) method for hyperspectral anomaly detection to overcome these shortcomings.In BSCR, we first selected hyperspectral bands via an optimal clustering framework and obtained a set of representative bands. The separability between the anomaly and background classes enhanced. Then, we reconstructed each pixel in the image through collaborative representation. We obtained a large residual when reconstructing the anomaly pixel via collaborative representation and achieved a large output value for an anomaly pixel because of the enhanced separability between the anomaly and background classes, thereby improving the separation of the anomaly class from the background class.The proposed algorithm was tested on synthetic and real hyperspectral images. The experimental results of three hyperspectral images show that the proposed BSCR demonstrates outstanding detection performance in the receiver operating characteristic curve, area under the curve value, and separability map compared with other state-of-the-art detectors.BSCR has an improved discriminative ability to separate the target from the background. Compared with several traditional anomaly detection algorithms, BSCR enhances the separability between the target and background through band selection and can effectively separate the anomaly class from the background class in hyperspectral images. Algorithms, such as PCAroCRD, which also conduct collaborative representation after the dimensionality reduction of hyperspectral images, can remove certain anomaly pixels and make the background construction stable. However, the dimensionality reduction mode of such algorithms will change the original signal of the image, thereby making it smooth and increasing the difficulty to distinguish between anomaly and background classes. The dimensionality reduction method of the band selection in BSCR can enhance the separability between the target and background without changing the original signal of the image, thereby making it easy to distinguish when detecting anomaly pixels and enabling sufficient discriminative ability to separate the target from the background.关键词：remote sensing;anomaly detection;band selection;collaborative representation;hyperspectral images;GF-5 satellite- 699
- |
- 264
- |
- 4
发布时间：2020-08-31 -
摘要：With the release of GF-5 hyperspectral data, hyperspectral remote sensing plays an increasingly important role in environmental monitoring, emergency management, and object extraction. Classification is the primary problem of hyperspectral image applications. Some cascade forest models have been proposed to overcome the limitations of traditional deep neural networks, such as requiring excessive training samples and optimization of a large number of hyperparameters. Traditional cascade forest models have several disadvantages, such as (1) high model complexity, (2) homogeneous base classifiers, and (3) inability to discriminate the similar spectrum. In this study, a novel classification approach based on cascade forest is proposed to solve the above drawbacks.The proposed improved cascade forest is an accumulation of layers, and each layer consists of two decision tree forests and a logistic regression classifier. Compared with traditional models, the number of forests in the improved method is reduced from four to two with the same accuracy and efficiency. Meanwhile, the original completely random tree forest is replaced by an efficient rotation forest to improve the diversity. The logistic regression classifier is added to determine the separating hyperplane among similar spectra. The number of layers is determined by the accuracy of validation set.The proposed method is implemented on three hyperspectral datasets (GF-5, Indian Pines, and Pavia University datasets). DBN, SVM, RoF, RF, and original cascade forest are selected as the contrast methods. Experimental results on three different real hyperspectral datasets confirms the superiority of the proposed method, especially on the Indian Pines dataset, which has a similar spectrum. The improved cascade model can combine multiple classification results from different base classifiers to obtain the final results through the logistic regression classifier, and the quadratic discriminant process of the logistic regression classifier can effectively improve the classification accuracy. The impacts of the number of trees on the final results are discussed. The proposed model obtains the best performance with optimal parameters. Although the single training time is long, the insensitivity of model parameters immensely improves the training efficiency.Compared with DNN, the improved cascade forest has the following advantages: (1) adaptive to determine the number of layers on the basis of the classification accuracy, (2) few hyperparameters are required in the improved cascade forest, making it easy to optimize the structure, and (3) each forest is independently trained because the training process do not have backpropagation, thereby accelerating the improved cascade forest by the CPU.关键词：remote sensing;hyperspectral classification;cascade forest;rotation forest;ensemble learning;deep learning- 904
- |
- 283
- |
- 9
发布时间：2020-08-31 -
摘要：Mineral identification, which is a feature of hyperspectral remote sensing technology, has been widely applied in geoscience and has achieved remarkable application results in geological and mineral fields. With the improvement of spectral resolution, mineral identification has gradually developed from the identification of mineral species to the identification of fine information, such as mineral subclasses and mineral components. Fine mineral information is extremely important in applications, such as the prediction and evaluation of mineral resources and geological environment indication. It directly affects the breadth and depth of hyperspectral remote sensing geological application. Spectral resolution and mineral identification methods are the main factors in fine mineral identification. GF-5 has 330 bands at the spectral range of 350—2500 nm, and its spectral resolution is higher than 10 nm. Its ultrahigh spectral resolution provides the possibility for fine mineral identification.In this study, a mineral identification method was presented on the basis of spectral characteristic enhancement matching degree and characteristic parameters by summarizing and analyzing the advantages and disadvantages of two commonly used mineral identification methods, namely, spectral matching and characteristic parameters, and combining the characteristics of GF-5 hyperspectral data. The proposed method was applied to conduct mineral identification in Liuyuan, Gansu, and Cuprite, USA. The mineral types and subclasses were first identified, and then the information on sericite composition was reversed. The airborne hyperspectral data were compared with the mapping results of GF-5.The results show that the GF-5 mineral identification information distribution has a good consistency with airborne HyMap and AVIRIS, and the average accuracy of GF-5 mineral identification is 90% higher compared with the airborne data. The accuracy rate, as a holistic evaluation, only serves as a reference because of the relatively limited statistical data, uneven distribution of mineral information, and the difference in original spatial resolution. The comparison results show that the proposed mineral identification method can meet the requirements of GF-5 mineral fine identification.Ultrahigh spectral resolution makes GF-5 advantageous in the identification of mineral composition information and distinguishing minerals with high spectral similarity. The proposed mineral identification method based on spectral characteristic enhancement matching degree and characteristic parameters can provide technical support for subsequent operational applications.关键词：GF-5;hyperspectral;shortwave infrared;mineral mapping;remote sensing geology- 1392
- |
- 539
- |
- 26
发布时间：2020-08-31 -
摘要：With the increasing of population, the world has experienced unprecedented urbanization, especially in developing countries. Rapid urbanization inevitably brings serious ecological and environmental problems. In order to better monitor the urbanization process of mega-cities, this study, which focus on north of the Yangtze river of Wuhan metropolitan region, attempts aimed to analyze the urban landscape pattern characteristics and the trends of urbanization trends by using the north of the Yangtze River of Wuhan metropolitan region as the case study.In this study, we proposed a modified Spectral–Spatial Unified Network (SSUN-)-CRF method, which revised from a Spectral-Spatial Unified Networks classification (SSUN) method, based on deep-learning, to extract the urban land use classification on a fine scale and recover detailed structure with GF-5 hyperspectral data in 2018. Then, we assessed seven key landscape pattern indices to understand the landscape pattern features of seven regions in Wuhan, which are separated into main urban area and suburban area. Then, in this paper we built the urban expansion model, from class metrics to landscape metrics.The resulst show that: (1) The SSUN-CRF classification algorithm can achieve as little loss of spectral information as possible while taking into account spatial information. Our method can effectively achieve edge refinement and overcome the misclassification of semantic segmentation. The Overall Accuracy (OA) and Kappa were higher than 0.9048 and 0.8807, respectively. (2) According to the landscape analysis, on the class metrics level: the PLAND (Percentage of Landscape) indices of different classes in different districts demonstrate that Wuhan has abundant water resource, the main urban area of Wuhan is more urbanized than that of suburban area. Comparing three kinds of residential areas, Residential Two is the principal part with high-rise buildings. It indicate that Wuhan has a higher per capita living standard. However, Hanyang District develops slower than other main urban areas with the problem of “villages in the city”. Furthermore, the advantages of agriculture and water area in the suburban area are obvious, the ecological environment is good, but the urbanization degree is low. The the ecological virescence of suburban area needs to be improved. The PD (Patch Density) and AI (Aggregation Index) indices illustrate that Jianghan District has the most evenly distributed commercial area, and has the center of business in Wuhan. While, the commercial area in the suburban district is less and concentrated. (3) On the landscape metrics level: We choose 4 indices to analysis the landscape in Wuhan including PAFRAC (Perimeter Area Fractal Dimension), CONTAG (Contagion Index), SHDI (Shannon's Diversity Index) and SHEI (Shannon's Evenness Index). Wuhan shows the stable and diverse landscape ecological with balanced land use distribution. What’s more, after using moving window to evaluate the state of Wuhan’s urbanization, it is clear to see the tendency in line from the main urban area to the southeast and southwest of the suburban districts. The main urban area is more evenly distributed and urbanized than the suburban area.The results of this research of serve as guide for the overall development plan, optimizing land use optimization, and sustainable development construction in Wuhan. What’s more, this model can further provide decision-making reference for the development of the Yangtze River Economic Belt and the “Yangtze River Spindle.”关键词：remote sensing;urbanization;landscape pattern;land use;deep learning;fully connected conditional random field;Fragstats;GF-5 hyperspectral image;Wuhan- 1191
- |
- 280
- |
- 14
发布时间：2020-08-31
- Postal code：100190
- Tel：010-58887052 Email：nrsb@aircas.ac.cn
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备20021838号-9
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0