Hyperspectral Applications Research Articles

Hyperspectral application for early diagnosis of American foulbrood disease in the honeybee (Apis mellifera L.) larvae

Non-destructive methods are of utmost importance for early diagnosis of American foulbrood (AFB) disease in honeybees’ larvae. This study discusses the potential application of spectroradiometric techniques for spectral discrimination of AFB disease in honeybees. AFB is a honeybees’ brood fully destructive bacterial disease caused by the spore forming bacterium; Paenibacillus larvae (P. larvae). It is important to detect the disease early as routine apiary administration and interchange of coloney components could spread it easily to healthy ones. Hyperspectral measurements of honeybee larval samples from 11 colonies were taken in range (350–2500 nm). The two spectral bands, Red and the Short-Wave Infrared (SWIR2), were of the promising results for differentiation between healthy and the different degree of honeybees’ larval infection. Correlation coefficient between Red and SWIR2 spectral reflectance data indicates that SWIR2 is the best spectral band to be used for discriminating between healthy and infected honeybee larvae. Confirmation of AFB was carried out by both cultivating smears of suspected larvae on agar plates and KAT-PCR technique for bacterial DNA confirmation. Both cultivation and PCR results agreed with hyperspectral results indicating that 6 colonies were of high loads of P. larvae, 2 of moderate bacterial loads and 1 was weak. Only 2 colonies appeared free from bacteria. Data analysed, discriminatory signatures the diseased samples in their various pathological stages. this research is an attempt to determine the feasibility of early diagnosis of AFB disease in honeybee brood through the application of hyperspectral technology.

Three-dimensional modeling of alteration information with hyperspectral core imaging and application to uranium exploration in the Heyuanbei uranium deposit, Xiangshan, Jiangxi, China

Hyperspectral technology is particularly good at identifying hydrothermally altered minerals and can offer support for three-dimensional (3-D) alteration modeling of hydrothermal deposit. We propose a 3-D alteration modeling method with hyperspectral core imaging to deepen hydrothermal alteration research for uranium exploration of the Heyuanbei uranium deposit in the western Xiangshan ore field, Jiangxi, South China. We use the following three steps: (1) identify and map the altered minerals of drill cores, (2) quantify all types of altered minerals using a pixel classification-based statistical algorithm, and (3) establish 3-D alteration model using a parallel section and a triangulated irregular network. Hyperspectral core imaging data from 14 drill cores from the Heyuanbei uranium deposit are used to establish a 3-D alteration model of the superimposed stratigraphy and structural information by this method. The modeling results indicate that large-scale illite and kaolinite–dickite alterations have developed in the Heyuanbei deposit. The illite alteration has the following obvious zoning characteristics: (1) the spatial correlation of short-wavelength illite and kaolinite–dickite, which developed in two volcanic formations, is high, and the alteration is stronger in the upper part of the drill cores than in the lower part of the drill cores; (2) the long-wavelength illite is mainly distributed in the lower volcanic formations. The zoning characteristics reveal that the hydrothermal fluid is acidic in the upper cores, whereas in the lower cores, alkaline fluid is present, and the fluid environment has the characteristics of a transition from early alkaline to late acidic conditions. Moreover, the short-wavelength illite is more closely related to fluorite-hydromica-type uranium mineralization than the long-wavelength illite, and the conversion action from long-wavelength illite to short-wavelength illite is favorable for uranium enrichment. Thus, short-wavelength illite, kaolinite, dickite, and fluorite can be used as prospective indicators for uranium exploration in the Heyuanbei mining area and nearby areas.

Loss of target information in full pixel and subpixel target detection in hyperspectral data with and without dimensionality reduction

In most hyperspectral target detection applications, targets are usually small and require both spatial as well as spectral detection. Hyperspectral imaging facilitates target detection (TD) applications greatly, however, due to large spectral content, hyperspectral data requires dimensionality reduction (DR) which also leads to loss of target information both at full pixel and subpixel level. Literature reports many DR and TD algorithms in practice. Several studies have focussed on assessing the loss of target information in DR, however, not much work seems to have been done to assess loss of target information in full pixel and subpixel TD in hyperspectral data with and without DR. This paper seeks to study various combinations of DR techniques combined with full pixel and subpixel TD algorithms. The results indicate that in the case of full pixel targets, both DR and TD contribute to the loss of target information, however, there is more loss of target information in the case when DR precedes TD in comparison to a case where TD is applied without DR. In the case of subpixel TD, however, there appears to be loss of subpixel target information in the case where TD alone is performed in comparison to a case where DR precedes TD.

Guest Editorial: Hyperspectral Imaging and Applications

Guest Editorial: Hyperspectral Imaging and Applications

Subspace-based multitask learning framework for hyperspectral imagery classification

Subspace-based models have been widely applied for hyperspectral imagery applications, especially for classification. The main principle of these methods is based on the fact that the original image can approximately lie on a lower-dimensional subspace. However, due to the existence of mixed samples, the subspace projection is unstable and affected by the selection of training samples, such that may lead to poor characterization and classification performances. In order to improve the robustness and characterization ability of the subspace-based classification models, this paper proposes a novel subspace-based multitask learning framework. In particular, the original image is first projected to the multiple subspaces in different branches. Then, the support vector machine (SVM) classifier is applied in each branch to deal with the projected data sets. With a consideration of integrating the spatial information, an extended step is provided including the process of a Markov Random Field (MRF) based on the result of SVM. Finally, the classification result is obtained by a decision fusion process. Experimental results on three real hyperspectral data sets demonstrate the improvements on classification performance of the proposed methods over other related methods.

Sum-Product Unmixing for Hyperspectral Analysis With Endmember Variability

Models of endmember variability capture the notion that multiple spectra may represent a single class or material, and while these models are physically realistic, they often give rise to excessive computational complexity during the spectral unmixing process. In this letter, we present a computationally tractable supervised unmixing method that uses Gauss–Markov processes to model endmember variability and interband correlation properties within endmembers. We use a probabilistic graphical model over multiple Gauss–Markov processes to capture the mixing effects of a spectral sensor and employ sum-product message passing to develop an accelerated statistical unmixing algorithm. The computational complexity of the proposed unmixing algorithm is only linear in the number of bands, making it suitable for both hyperspectral (hundreds of bands) and ultraspectral (thousands of bands) applications. Unmixing examples with measured reflectance spectra show sizable performance improvements when accounting for interband correlation using the proposed method, and empirical results quantify orders of magnitude reduction in complexity compared to alternative methods.

Unsupervised classification-based hyperspectral data processing: lossy compression

The existing lossy compression algorithms play an important role in reducing the cost of storage equipment and bandwidth for hyperspectral (HS) application. However, none of the lossy compression algorithms considers the real-time classification of HS data. In this paper, we present a new lossy compression method for HS data that aims to optimally compress in both spatial and spectral domains and simultaneously maximize classification performance. For this target, Harsanyi–Farrand–Chang (HFC) and k-means++ algorithms are applied to achieve a spectral library and an index matrix for HS image. Spectral angle mapping and Euclidean distance are used to update the spectral library and the index matrix. The experiment results indicate that the proposed method has a good classification performance. The results also reveal that the proposed method works well in real-time classification and compression of HS data with a large volume and achieves a high compression ratio. It is noteworthy to mention that the superiority of our method in compression becomes more apparent as the volume of HS data grows. Consequently, the proposed method has a strong advantage in HS applications that require both compression and classification.

HICF: A MATLAB PACKAGE FOR HYPERSPECTRAL IMAGE CLASSIFICATION AND FUSION FOR EDUCATIONAL LEARNING AND RESEARCH

Abstract. A significant surge has been observed with the development and research in remote sensing in recent years for hyperspectral applications in Earth observation. Subsequently, the development of software and tools have also experienced an unprecedented rise, both in research as well as in academia. Although commercial software and tools such as ENVI by ITT Visual Information Solutions, Boulder, CO, USA are available for visualizing and analyzing the hyperspectral images, such software are expensive. Some open source toolboxes such as the MATLAB-based Hyperspectral Image Analysis Toolbox (HIAT) are also available. However, mostly these toolboxes have not been packaged for dissemination and operation without the MATLAB software which is commercial. In this paper, we introduce the Hyperspectral Image Classification and Fusion (HICF) package which is being developed at the Geoinformatics laboratory, Department of Civil Engineering, Indian Institute of Technology Kanpur (IITK) in MATLAB that can be used by standalone installation with an open source supplementary MATLAB compiler. This software is intended to provide a collection of algorithms both conventional and those developed at the Geoinformatics laboratory that utilizes the numerical computing capability of MATLAB for the processing of hyperspectral and multispectral imagery. The HICF software comprises a simple design of the graphical user interface which can be efficiently used particularly for academic purposes.

Hyperspectral Unmixing Using Sparsity-Constrained Deep Nonnegative Matrix Factorization With Total Variation

Hyperspectral unmixing is an important processing step for many hyperspectral applications, mainly including: 1) estimation of pure spectral signatures (endmembers) and 2) estimation of the abundance of each endmember in each pixel of the image. In recent years, nonnegative matrix factorization (NMF) has been highly attractive for this purpose due to the nonnegativity constraint that is often imposed in the abundance estimation step. However, most of the existing NMF-based methods only consider the information in a single layer while neglecting the hierarchical features with hidden information. To alleviate such limitation, in this paper, we propose a new sparsity-constrained deep NMF with total variation (SDNMF-TV) technique for hyperspectral unmixing. First, by adopting the concept of deep learning, the NMF algorithm is extended to deep NMF model. The proposed model consists of pretraining stage and fine-tuning stage , where the former pretrains all factors layer by layer and the latter is used to reduce the total reconstruction error. Second, in order to exploit adequately the spectral and spatial information included in the original hyperspectral image, we enforce two constraints on the abundance matrix. Specifically, the $L_{1/2}$ constraint is adopted, since the distribution of each endmember is sparse in the 2-D space. The TV regularizer is further introduced to promote piecewise smoothness in abundance maps. For the optimization of the proposed model, multiplicative update rules are derived using the gradient descent method. The effectiveness and superiority of the SDNMF-TV algorithm are demonstrated by comparing with other unmixing methods on both synthetic and real data sets.

Sparse Unmixing of Hyperspectral Data with Noise Level Estimation

Recently, sparse unmixing has received particular attention in the analysis of hyperspectral images (HSIs). However, traditional sparse unmixing ignores the different noise levels in different bands of HSIs, making such methods sensitive to different noise levels. To overcome this problem, the noise levels at different bands are assumed to be different in this paper, and a general sparse unmixing method based on noise level estimation (SU-NLE) under the sparse regression framework is proposed. First, the noise in each band is estimated on the basis of the multiple regression theory in hyperspectral applications, given that neighboring spectral bands are usually highly correlated. Second, the noise weighting matrix can be obtained from the estimated noise. Third, the noise weighting matrix is integrated into the sparse regression unmixing framework, which can alleviate the impact of different noise levels at different bands. Finally, the proposed SU-NLE is solved by the alternative direction method of multipliers. Experiments on synthetic datasets show that the signal-to-reconstruction error of the proposed SU-NLE is considerably higher than those of the corresponding traditional sparse regression unmixing methods without noise level estimation, which demonstrates the efficiency of integrating noise level estimation into the sparse regression unmixing framework. The proposed SU-NLE also shows promising results in real HSIs.

Learning Sensor-Specific Spatial-Spectral Features of Hyperspectral Images via Convolutional Neural Networks

Convolutional neural network (CNN) is well known for its capability of feature learning and has made revolutionary achievements in many applications, such as scene recognition and target detection. In this paper, its capability of feature learning in hyperspectral images is explored by constructing a five-layer CNN for classification (C-CNN). The proposed C-CNN is constructed by including recent advances in deep learning area, such as batch normalization, dropout, and parametric rectified linear unit (PReLU) activation function. In addition, both spatial context and spectral information are elegantly integrated into the C-CNN such that spatial-spectral features are learned for hyperspectral images. A companion feature-learning CNN (FL-CNN) is constructed by extracting fully connected feature layers in this C-CNN. Both supervised and unsupervised modes are designed for the proposed FL-CNN to learn sensor-specific spatial-spectral features. Extensive experimental results on four benchmark data sets from two well-known hyperspectral sensors, namely airborne visible/infrared imaging spectrometer (AVIRIS) and reflective optics system imaging spectrometer (ROSIS) sensors, demonstrate that our proposed C-CNN outperforms the state-of-the-art CNN-based classification methods, and its corresponding FL-CNN is very effective to extract sensor-specific spatial-spectral features for hyperspectral applications under both supervised and unsupervised modes.

One-Dimensional Convolutional Neural Network Land-Cover Classification of Multi-Seasonal Hyperspectral Imagery in the San Francisco Bay Area, California

In this study, a 1-D Convolutional Neural Network (CNN) architecture was developed, trained and utilized to classify single (summer) and three seasons (spring, summer, fall) of hyperspectral imagery over the San Francisco Bay Area, California for the year 2015. For comparison, the Random Forests (RF) and Support Vector Machine (SVM) classifiers were trained and tested with the same data. In order to support space-based hyperspectral applications, all analyses were performed with simulated Hyperspectral Infrared Imager (HyspIRI) imagery. Three-season data improved classifier overall accuracy by 2.0% (SVM), 1.9% (CNN) to 3.5% (RF) over single-season data. The three-season CNN provided an overall classification accuracy of 89.9%, which was comparable to overall accuracy of 89.5% for SVM. Both three-season CNN and SVM outperformed RF by over 7% overall accuracy. Analysis and visualization of the inner products for the CNN provided insight to distinctive features within the spectral-temporal domain. A method for CNN kernel tuning was presented to assess the importance of learned features. We concluded that CNN is a promising candidate for hyperspectral remote sensing applications because of the high classification accuracy and interpretability of its inner products.

Estimation of the Number of Endmembers in a Hyperspectral Image via the Hubness Phenomenon

Estimation of the number of endmembers (NOE) is an important first step in many hyperspectral unmixing applications. We present a new method for solving this problem, based on the statistics of the indegree distribution (IDD) of the data nearest neighbor graph. It is known that this IDD shows a high dependence on the intrinsic dimensionality (ID) of the data, and becomes skewed for increasing dimensionality. This effect is known as the hubness phenomenon, and we propose a technique that exploits this effect to derive an estimate for the NOE in a hyperspectral data set. While this number should have a trivial relation with the ID of the data set, this relation is often obscured by the large correlations that exist between endmember spectra and adjacent spectral bands. The proposed technique circumvents this problem by building representative statistics based on simulated hyperspectral data sets, and therefore performs much better than alternative techniques. Also several types of nonlinearly mixed data sets can be treated by the proposed technique, which is illustrated with bilinear data sets.

FPGA implementation of a maximum simplex volume algorithm for endmember extraction from remotely sensed hyperspectral images

Spectral unmixing is a very important technique for remotely sensed hyperspectral unmixing. Since more hyperspectral applications now require real or near real-time processing capabilities, fast spectral unmixing using field-programmable gate arrays (FPGAs) has received considerable interest in recent years. FPGAs can provide onboard, high computing performance at low power consumption. Another important characteristic of FPGA-based systems is reconfigurability, which makes them more flexible to process different kind of scenes. Pure signature (endmember) extraction is a fundamental step in spectral unmixing, which has been tackled using the maximum volume principle by several algorithms, most notably N-FINDR and simplex growing algorithm (SGA). These algorithms find out the simplex with maximum volume as a mechanism to extract endmembers. However, a previous dimensionality reduction step is generally required, which introduces information loss and additional computational burden. To address these issues, in this work we introduce a new volume calculation formula and further develop a new real-time implementation of a maximum simplex volume algorithm (called RT-MSVA). The proposed RT-MSVA does not need dimensionality reduction, so all spectral bands can be used without losing any information to ensure robust endmember extraction accuracy. Experiments with synthetic and real hyperspectral images have been conducted to evaluate the accuracy and computational performance of our proposed method. Our experimental results indicate that proposed FPGA-based implementation significantly outperforms the corresponding software version and achieves real-time processing performance in the considered problem. It also exhibits better endmember extraction accuracy and comparable performance to other available techniques, such as a real-time implementation of a simplex growing algorithm (RT-FSGA).

Ultranarrow spectral response of InGaAs QDIPs through the optimization of strain-coupled stacks and capping layer composition

The effect of the capping layer and the number of strain-coupled stacks on the optoelectrical properties of In0.5Ga0.5As quantum dot infrared photodetectors (QDIPs) are reported in this paper. GaAs-capped and InGaAs-capped bilayer, trilayer, pentalayer, and heptalayer QDIPs were grown for the first time and analyzed. The ground-state photoluminescence emission of all coupled QDIPs redshifted relative to that of the uncoupled QDIPs. All coupled InGaAs-capped QDIPs exhibited monomodal spectral response, whereas among GaAs-capped QDIPs, only the bilayer QDIP exhibited monomodal response. The optimal activation energy (339.87meV) and full width at half maximum (FWHM; 16meV) as well as the lowest dark current density (6.5E−8A/cm2) at 100K, and −1V bias were observed for the InGaAs-capped trilayer QDIP among all due to better dot confinement, homogeneity, and also better strain coupling between the vertically stacks QD layers. The peak spectral response at 7.08µm obtained from this QDIP had an ultranarrow FWHM of 8.67meV, making it useful for hyperspectral applications in the infrared region.

Discovering Diverse Subset for Unsupervised Hyperspectral Band Selection.

Band selection, as a special case of the feature selection problem, tries to remove redundant bands and select a few important bands to represent the whole image cube. This has attracted much attention, since the selected bands provide discriminative information for further applications and reduce the computational burden. Though hyperspectral band selection has gained rapid development in recent years, it is still a challenging task because of the following requirements: 1) an effective model can capture the underlying relations between different high-dimensional spectral bands; 2) a fast and robust measure function can adapt to general hyperspectral tasks; and 3) an efficient search strategy can find the desired selected bands in reasonable computational time. To satisfy these requirements, a multigraph determinantal point process (MDPP) model is proposed to capture the full structure between different bands and efficiently find the optimal band subset in extensive hyperspectral applications. There are three main contributions: 1) graphical model is naturally transferred to address band selection problem by the proposed MDPP; 2) multiple graphs are designed to capture the intrinsic relationships between hyperspectral bands; and 3) mixture DPP is proposed to model the multiple dependencies in the proposed multiple graphs, and offers an efficient search strategy to select the optimal bands. To verify the superiority of the proposed method, experiments have been conducted on three hyperspectral applications, such as hyperspectral classification, anomaly detection, and target detection. The reliability of the proposed method in generic hyperspectral tasks is experimentally proved on four real-world hyperspectral data sets.

Sparsity-Regularized Robust Non-Negative Matrix Factorization for Hyperspectral Unmixing

Hyperspectral unmixing (HU) is one of the crucial steps for many hyperspectral applications, including material classification and recognition. In the last decade, non-negative matrix factorization (NMF) and its extensions have been widely studied and have achieved advanced performances in HU. Unfortunately, most of the existing NMF-based methods make the assumption that the hyperspectral data are only corrupted by Gaussian noise. In real applications, the hyperspectral data are inevitably corrupted by sparse noise, which includes impulse noise, stripes, deadlines, and others types of noise. By separately modeling the sparse noise and Gaussian noise, a robust NMF (RNMF) model is subsequently introduced to unmix the hyperspectral data. The proposed RNMF model is able to simultaneously handle Gaussian noise and sparse noise, and can be efficiently learned with elegant update rules. In addition, sparsity regularizers are added to restrict the abundance maps in the RNMF, with the consideration of the sparse property of the material types within the hyperspectral scene. The experimental results with simulated and real data confirm the superiority of the proposed sparsity-regularized RNMF methods compared to the traditional NMF methods.

GPU-Based Parallel Design of the Hyperspectral Signal Subspace Identification by Minimum Error (HySime)

Signal subspace identification provides a performance improvement in hyperspectral applications, such as target detection, spectral unmixing, and classification. The HySime method is a well-known unsupervised approach for hyperspectral signal subspace identification. It computes the estimated noise and signal correlation matrices from which a subset of eigenvectors is selected to best represent the signal subspace in the least square sense. Depending on the complexity and dimensionality of the hyperspectral scene, the HySime algorithm may be computationally expensive. In this paper, we propose a massively parallel design of the HySime method for acceleration on NVIDIA's graphics processing units (GPUs). Our pure GPU-based implementation includes the optimal use of the page-locked host memory, block size, and the number of registers per thread. The proposed implementation was validated in terms of accuracy and performance using the NASA AVIRIS hyperspectral data. The benchmark with the NVIDIA GeForce GTX 580 and Tesla K20 GPUs shows significant speedups with regards to the optimized CPU-based serial counterpart. This new fast implementation of the HySime method demonstrates good potential for real-time hyperspectral applications.

Semisupervised Hyperspectral Image Classification via Discriminant Analysis and Robust Regression

In recent years, hyperspectral image classification (HSIC) has received increasing attention in a wide range of hyperspectral applications. It is still very challenging due to the following factors: 1) there are not enough labeled samples; 2) the images are easy to be polluted by outlier channels; and 3) different objects may have similar spectra. Considering these three factors, we propose a novel semisupervised HSIC method, which is constructed on discriminant analysis and robust regression (DARR). Specifically, a regression-based semisupervised technique is employed by not only exploiting the rich information in labeled samples, but also taking advantage of abundant unlabeled ones. In this way, the true data distribution can be obtained accurately. Then, we introduce a robust adaptive loss function to measure the representation loss. As a result, it can greatly relieve the side effects of outlier channels. Finally, to increase discriminating power of our approach for different objects, we utilize the pairwise constraints to incorporate the discriminant information among labeled samples. Through these constraints, the same-category samples are projected to be close to each other, while the different-category samples are as far apart as possible. The above three components can be integrated into a graph-based objective function, whose optimization is systematically provided. Extensive experiments on four data sets demonstrate that our method achieves higher quantitative results, as well as more satisfactory visual performances by comparing with state-of-the-art methods and using different parameter settings.

Hyperspectral unmixing based on iteratively re-weighted smoothed l<SUB align="right">0 regularisation

Hyperspectral unmixing (HU) plays a fundamental role in a wide range of hyperspectral applications. Sparse-based approach has recently received much attention in hyperspectral unmixing area. Sparse unmixing is based on the assumption that each mixed pixel in the hyperspectral image can be expressed in the form of linear combination of a number of pure spectral signatures that are known in advance. Despite the success of sparse unmixing based on the L0, L1/2 or L0 regulariser, the limitation of this approach on its computational complexity or sparsity affects the efficiency or accuracy. In this paper, as iteratively re-weighted algorithm can help to find the much sparser solution in optimal problems, we proposed a sparse unmixing method based on iteratively re-weighted smoothed L0 regularisation, named RSL0SU. Then a variable splitting and augmented Lagrangian algorithm is introduced to solve the optimisation problem. Comparing to SUnSAL method and SL0SU method, our experimental results with both simulated and real hyperspectral data sets demonstrate that the RSL0SU method is an effective and accurate spectral unmixing algorithm for hyperspectral remote sensing imagery.