Hyperspectral Imaging Applications Research Articles

The Application of Hyperspectral Imaging to the Measurement of Pressure Injury Area.

Wound size measurement is an important indicator of wound healing. Nurses measure wound size in terms of length × width in wound healing assessment, but it is easy to overestimate the extent of the wound due to irregularities around it. Using hyperspectral imaging (HIS) to measure the area of a pressure injury could provide more accurate data than manual measurement, ensure that the same tool is used for standardized assessment of wounds, and reduce the measurement time. This study was a pilot cross-sectional study, and a total of 30 patients with coccyx sacral pressure injuries were recruited to the rehabilitation ward after approval by the human subjects research committee. We used hyperspectral images to collect pressure injury images and machine learning (k-means) to automatically classify wound areas in combination with the length × width rule (LW rule) and image morphology algorithm for wound judgment and area calculation. The results calculated from the data were compared with the calculations made by the nursing staff using the length × width rule. The use of hyperspectral images, machine learning, the length × width rule (LW rule), and an image morphology algorithm to calculate the wound area yielded more accurate measurements than did nurses, effectively reduced the chance of human error, reduced the measurement time, and produced real-time data. HIS can be used by nursing staff to assess wounds with a standardized approach so as to ensure that proper wound care can be provided.

Clinical applications of non-invasive multi and hyperspectral imaging of cell and tissue autofluorescence beyond oncology.

Hyperspectral and multispectral imaging of cell and tissue autofluorescence employs fluorescence imaging, without exogenous fluorophores, across multiple excitation/emission combinations (spectral channels). This produces an image stack where each pixel (matched by location) contains unique information about the sample's spectral properties. Analysis of this data enables access to a rich, molecularly specific data set from a broad range of cell-native fluorophores (autofluorophores) directly reflective of biochemical status, without use of fixation or stains. This non-invasive, non-destructive technology has great potential to spare the collection of biopsies from sensitive regions. As both staining and biopsy may be impossible, or undesirable, depending on the context, this technology great diagnostic potential for clinical decision making. The main research focus has been on the identification of neoplastic tissues. However, advances have been made in diverse applications-including ophthalmology, cardiovascular health, neurology, infection, assisted reproduction technology and organ transplantation.

Graph-Based Structural Deep Spectral-Spatial Clustering for Hyperspectral Image

As a fundamental task in hyperspectral image (HSI) applications, HSI clustering has been extensively researched over the years. Inspired by the capability of graphs in structure modeling, some recent works have exploited the local graph-structured information among pixels to boost the HSI clustering performance. However, these works only explore shallow graph structures for representation learning, while ignoring the beneficial high-order as well as global structure characteristics of HSI pixels, thus resulting in limited clustering performance. In this paper, a novel graph-based structural deep spectral-spatial clustering network (GSDSSC) is proposed to sufficiently explore the structure information among pixels. Specifically, a self-expression embedded multi-graph auto-encoder is proposed to explore high-order structure associations among pixels, thereby capturing robust spectral-spatial features and global clustering structure. Besides, to learn more compact spectral-spatial features for clustering, a global structure-guided optimization mechanism is further designed to constrain the local feature distribution with global clustering structure. Extensive experiments on three public HSI benchmark datasets illustrate the superior performance of the proposed GSDSSC over state-of-the-art HSI clustering methods.

Combined Deep Priors With Low-Rank Tensor Factorization for Hyperspectral Image Restoration

Mixed noise pollution severely disturbs hyperspectral image (HSI) processing and applications. Plenty of algorithms have been developed to address this issue via two strategies: model-driven or data-driven strategy. However, model-driven methods exist in the highly time-consuming weakness of iterative optimization and unstable sensitivity of setting parameters. Data-driven methods usually perform poor due to the overfitting effects. To solve these issues, we combine both the deep denoising priors with low-rank tensor factorization (DP-LRTF) for HSI restoration. The proposed method uses Tucker tensor factorization to depict the global spectral low-rank constraint. Then the spectral orthogonal basis and spatial reduced factor are optimized by two deep denoising priors, respectively. Through this integrated strategy, we can simultaneously exploit the intrinsic low-rank property of HSI, and utilize the powerful feature extraction ability by deep learning for HSI restoration. Compared with model-driven and data-driven methods, DP-LRTF outperforms on HSI mixed noise removal and execution efficiency for various simulated/real experiments.

Application of Hyperspectral Imaging for Rapid and Nondestructive Detection of Paraffine-Contaminated Rice

The emergence of paraffin-coated rice in China, aimed at enhancing its market appeal and achieving a translucent appearance, has given rise to a significant global food safety concern. This situation poses substantial health risks to consumers. Hyperspectral analysis, recognized as a powerful and nondestructive technique for assessing food quality and safety, offers a potential solution. This study conducted a comprehensive investigation using Visible-Near Infrared (VIS-NIR) hyperspectral imaging systems operating within the 400-1000 nm range to identify paraffin-contaminated rice. Various rice varieties from diverse regions were obtained and intentionally tainted with varying levels of paraffin. Imaged samples were further preprocessed for spectral data extraction from individual rice seeds’ regions of interest (ROI). The dataset encompassed 3000 spectral records obtained from both non-contaminated and contaminated samples. The obtained spectral data were employed to develop partial least squares discriminant analysis (PLS-DA) and principal component linear discriminant analysis. The primary goal was to discriminate between contaminated and non-contaminated rice samples effectively. Notably, the results indicated that PLS-DA consistently achieved an accuracy exceeding 94% across various preprocessing techniques. Overall, this study showcased the potential of combining hyperspectral imaging with chemometrics to detect paraffin-contaminated rice seeds, providing a valuable contribution to food safety assessment in the industry.

Prediction of Honeydew Contaminations on Cotton Samples by In-Line UV Hyperspectral Imaging

UV hyperspectral imaging (225 nm-410 nm) was used to identify and quantify the honeydew content of real cotton samples. Honeydew contamination causes losses of millions of dollars annually. This study presents the implementation and application of UV hyperspectral imaging as a non-destructive, high-resolution, and fast imaging modality. For this novel approach, a reference sample set, which consists of sugar and protein solutions that were adapted to honeydew, was set-up. In total, 21 samples with different amounts of added sugars/proteins were measured to calculate multivariate models at each pixel of a hyperspectral image to predict and classify the amount of sugar and honeydew. The principal component analysis models (PCA) enabled a general differentiation between different concentrations of sugar and honeydew. A partial least squares regression (PLS-R) model was built based on the cotton samples soaked in different sugar and protein concentrations. The result showed a reliable performance with R2cv = 0.80 and low RMSECV = 0.01 g for the validation. The PLS-R reference model was able to predict the honeydew content laterally resolved in grams on real cotton samples for each pixel with light, strong, and very strong honeydew contaminations. Therefore, inline UV hyperspectral imaging combined with chemometric models can be an effective tool in the future for the quality control of industrial processing of cotton fibers.

Application of Near-Infrared Spectroscopy and Hyperspectral Imaging Combined with Machine Learning Algorithms for Quality Inspection of Grape: A Review.

Grape is a fruit rich in various vitamins, and grape quality is increasingly highly concerned with by consumers. Traditional quality inspection methods are time-consuming, laborious and destructive. Near-infrared spectroscopy (NIRS) and hyperspectral imaging (HSI) are rapid, non-destructive and accurate techniques for quality inspection and safety assessment of agricultural products, which have great potential in recent years. The review summarized the applications and achievements of NIRS and HSI for the quality inspection of grapes for the last ten years. The review introduces basic principles, signal mode, data acquisition, analysis and processing of NIRS and HSI data. Qualitative and quantitative analysis were involved and compared, respectively, based on spectral features, image features and fusion data. The advantages, disadvantages and development trends of NIRS and HSI techniques in grape quality and safety inspection are summarized and discussed. The successful application of NIRS and HSI in grape quality inspection shows that many fruit inspection tasks could be assisted with NIRS and HSI.

Robust statistical analysis to predict and estimate the concentration of the cannabidiolic acid in Cannabis sativa L.: A comparative study

Growing awareness of the medicinal and therapeutic benefits of cannabis has resulted in extensive research, increasing numbers of commercial growers, and the provision of commercial testing. For commercial production, the importance of monitoring for consistent cannabinoid composition is essential. Gas chromatography and/or liquid chromatography methods are effective methods. Unfortunately, they are not easily scaled up for frequent or large-scale testing. Precision cultivation for medicinal cannabis requires the industry to test large volumes of plants frequently during different stages of the growth cycle. This cannot be achieved with destructive testing alone. A statistical method with 1H nuclear magnetic resonance (NMR) is proposed and investigated to explore its potential for a cost-effective mass-screening method of plant material samples. In addition, hyperspectral imaging (HSI) is employed for frequent non-destructive measurements. This allows increasing the frequency of observations during the growth of the plant. The cannabidiolic acid (CBDA) concentration is determined by the application of both HSI and NMR spectroscopy and validated against liquid chromatography-mass spectrometry (LCMS). The paper proposes a multivariate statistical regression algorithm that automatically determines the CBDA concentration directly from bucket integration of the NMR spectrum. The algorithm successfully predicted the CBDA concentration of 7 unknown samples from data selected from 4 known samples, with an average R2 value of 0.98. The proposed statistical method was applied to the data collected for HSI. It showed that while the hyperspectral dataset can be correlated with CBDA concentration, it was subject to high variance. However, HSI retains the spatial information of the actual plant structure, allowing the CBDA prediction to be mapped back to the original spatial location in the plant while providing visual information on CBDA concentration within a flower or leaf without requiring the plant component to be destroyed in the process.

Cryogenic mid-wave infrared hyperspectral Fabry-Perot filter based on a tensile-strained single-layer subwavelength grating mirror.

We report here the first demonstration of a cryogenic mid-wave infrared (MWIR) hyperspectral fixed-cavity Fabry-Perot filter based on a suspended tensile-strained single-layer 2-D subwavelength grating (SWG) mirror. Optical design optimization of the 2-D SWG mirror and parameter tolerance study are performed. For the first time, process control of grating air-hole sidewall angle and the grating air-hole fill-factor fabrication error caused by e-beam lithography electron-scattering effect is reported. At 80 K, namely the operating temperature of MWIR photodetectors, the as-fabricated suspended 2-D SWG mirror has achieved excellent surface flatness with a slight center-to-edge bowing of 15 nm over a 1-mm2 large mirror area and a high average reflectivity of 0.97 across a wavelength range of 3.72-5 µm, which represents an unprecedentedly wide fractional bandwidth Δλ/λc of 30%. The cryogenically cooled Fabry-Perot filter exhibits an unrivaled high spectral resolution of 10 nm that far exceeds the optical requirement for MWIR hyperspectral imaging applications.

Consideration on the Cleaning Effect of Black Contaminants using Visible-Near-Infrared (VNIR) Hyperspectral Images

VNIR hyperspectral images can be analyzed in situations involving color changes, such as before and after cleaning, without pre-treatment. In this study, to analyze the application of the VNIR hyperspectral system, we compared the black contaminants on the surface of the Jigwangguksa Pagoda upper stereobate at Beopcheonsa Temple, Wonju before and after cleaning with laser. After a preliminary experiment to evaluate the applicability of VNIR hyperspectral imaging, we took images of the four sides (north, south, east, and west). The 823 nm band image was used for comparing before and after cleaning. As per the result of the image differencing analysis about before and after cleaning, the spectral reflectance of the area where the cleaning was performed showed a tendency to increase. Additionally, it was found that not only the washed area but also the extent of change could be calculated through monochrome for quantitative comparison, and imaging analysis using the VNIR hyperspectral system was sufficiently effective for comparison before and after cleaning.

What Lies Beyond Sight? Applications of Ultraportable Hyperspectral Imaging (VIS-NIR) for Archaeological Fieldwork

ABSTRACT This paper discusses the use of a small ultraportable hyperspectral camera in the VIS-NIR (Visible-Near Infrared) range for archaeological fieldwork and its hardware, data processing workflows, and spectral information that can be used for in situ screening. Hyperspectral imaging is a widespread, non-destructive analytical technique used in various disciplines for highlighting invisible patterns and mapping the spectral signatures of selected targets. In archaeology, it has mostly been applied for remote sensing satellite imagery to uncover information about features that are hidden underground. Targeted applications of hyperspectral imaging have been developed in the last few years, opening new perspectives for material analysis based on spectral mapping. Recent advances in portable instrumentation have led to the development of small and rugged cameras that can be used directly in the field for investigating different materials.

Tree Species Classification Based on Fusion Images by GF-5 and Sentinel-2A

Forest ecosystem detection and assessment usually requires accurate spatial distribution information of forest tree species. Remote sensing technology has been confirmed as the most important method for tree species acquisition, and space-borne hyperspectral imagery, with the advantages of high spectral resolution, provides a better possibility for tree species classification. However, the present in-orbit hyperspectral imager has proved to be too low in spatial resolution to meet the accuracy needs of tree species classification. In this study, we firstly explored and evaluated the effectiveness of the Gram-Schmidt (GS) Harmonic analysis fusion (HAF) method for image fusion of GaoFen-5 (GF-5) and Sentinel-2A. Then, the Integrated Forest Z-Score (IFZ) was used to extract forest information from the fused image. Next, the spectral and textural features of the fused image, and topographic features extracted from DEM were selected according to random forest importance ranking (Mean Decreasing Gini (MDG) and Mean Decreasing Accuracy (MDA)), and imported into the random forest classifier to complete tree species classification. The results showed that: comparing some evaluation factors such as information entropy, average gradient and standard deviation of the fused images, the GS fusion image was proven to have a higher degree of spatial integration and spectral fidelity. The random forest importance ranking showed that WBI, Aspect, NDNI, ARI2, FRI were more important for tree species classification. Both the classification accuracy and kappa coefficients of the fused images were significantly greatly improved when compared to those of original GF-5 images. The overall classification accuracy ranged from 61.17% to 86.93% for different feature combination scenarios, and accuracy of the selected method based on MDA achieved higher results (OA = 86.93%, Kappa = 0.85). This study demonstrated the feasibility of fusion of GF-5 and Sentinel-2A images for tree species classification, which further provides good reference for application of in-orbit hyperspectral images.

Spatial Sampling and Grouping Information Entropy Strategy Based on Kernel Fuzzy C-Means Clustering Method for Hyperspectral Band Selection

The high spectral resolution of hyperspectral images (HSIs) provides rich information but causes data redundancy, which imposes a computational burden on practical applications. Band selection methods can select a subset of HSI without changing the main information to reduce the spectral dimension. Clustering-based methods can reduce band correlation significantly, but traditional clustering methods are mostly hard clustering and are not accurate enough to partition the bands. An unsupervised band selection method based on fuzzy c-means clustering (FCM) was introduced to tackle this problem. However, FCM can easily obtain the local optimal solution and take a long time to process high-dimensional data. Hence, this work applies kernel function and a sampling strategy to reduce calculation time, and information entropy is used to initialize the cluster center. A kernel FCM algorithm based on spatial sampling and a grouping information entropy strategy is proposed and called SSGIE-KFCM. This method not only optimizes the calculation process and reduces the amount of computation data, accelerating the calculation efficiency, but also adopts grouping information entropy to improve the probability of obtaining a global optimal solution. Classification experiments on two public HSI datasets show that: (1) The classification performance of the whole band can be achieved or even exceeded by using only a small number of bands to achieve the purpose of dimensionality reduction. (2) The classification accuracy can be improved compared with the FCM method. (3) With the introduction of sampling strategy and kernel function, the computational speed is at least 24 times faster than that of FCM. It has been proven that the SSGIE-KFCM method can significantly reduce the amount of HSI while retaining the primary information of the original data, which further promotes the research and application of HSI in the remote sensing area.

Application of hyperspectral imaging assisted with integrated deep learning approaches in identifying geographical origins and predicting nutrient contents of Coix seeds

Coix seed (CS, Coix lachryma-jobi L. var. ma-yuen (Roman.) Stapf) has rich nutrients, including starch, protein and oil. The geographical origin with a protected geographical indication and high levels of nutrient contents ensures the quality of CS, but non-destructive and rapid methods for predicting these quality indicators remain to be explored. This paper proposed hyperspectral imaging (HSI) assisted with the integrated deep learning models of attention mechanism (AM), convolutional neural networks, and long short-term memory. The method achieved the effective wavelengths selection, the highest prediction accuracy for production region discrimination and the lowest mean absolute error and root mean squared error for nutrient contents prediction. Moreover, the wavelengths selected via the AM model were explicable and reliable for predicting the geographical origins and nutrient contents. The proposed combination of HSI with integrated deep learning models has great potential in the quality evaluation of CS.

Application of near-infrared hyperspectral imaging coupled with chemometrics for rapid and non-destructive prediction of protein content in single chickpea seed

Evaluating the protein content of a single chickpea seed in a rapid, non-destructive, and precise manner is crucial for facilitating the breeding of high-protein chickpeas. This study explored the potential of near-infrared (NIR) hyperspectral imaging (HSI) to predict the protein content in a single chickpea seed. Eight varieties of chickpeas with different protein contents were subjected to NIR reflectance hyperspectral imaging in the spectral range of 900–2500 nm at two different positions of chickpea seed (micropyle down and micropyle up). The spectral data was correlated with the measured reference protein content of chickpea seed for building the partial least square regression (PLSR) and support vector machine regression (SVMR) models based on different spectral preprocessing techniques, with full spectrum and effective wavelengths selected using competitive adaptive reweighted sampling (CARS) and iteratively retaining informative variables (IRIV) algorithms. When using the full spectrum, the optimal protein prediction model was obtained using PLSR, which yielded correlation coefficient of prediction (R2p) and root mean square error of prediction (RMSEP) values of 0.935 and 0.987, respectively, with external parameter orthogonalization (EPO)+standard normal variate (SNV) preprocessing for micropyle down position of chickpea seed. The IRIV selected wavelength with PLSR yielded the best model with R2p and RMSEP of 0.947 and 0.861, respectively, at the micropyle down position of chickpea seed. Hence, the optimal prediction models were obtained using PLSR with EPO+SNV at the micropyle down position of chickpea seed.

Reweighted sparse unmixing for hyperspectral images with noise level estimation

Spectral unmixing plays an important role in hyperspectral image applications and an amount of work has been devoted to hyperspectral unmixing (HU) in recent years. Hyperspectral images (HSIs) are typically corrupted by mixed types of noise and the noise level usually varies among different bands. The unmixing results could be misleading without individualized noise estimations for each band. In addition, there exist conspicuous sparsity of endmembers and spatial correlation in hyperspectral scenes, which can be utilized to improve the hyperspectral unmixing performance. In this paper, we propose a novel model that integrates weighted sparse unmixing with bandwise noise characterization. To deal with various noise present in HSIs, we introduce a weighting matrix derived from the Gaussian noise estimation in each band and a l1 regularizer for sparse noise. Then we incorporate a weighted sparse regularizer, which includes both spatial and spectral signatures weighting factors to take advantage of spatial correlation among fractional abundances and enhance the sparsity of endmembers simultaneously. An efficient algorithm based on alternating direction method of multipliers (ADMM) is utilized to solve the proposed model. The experimental results on synthetic and real hyperspectral data suggest that the consolidation of noise estimation and weighted regularizer could achieve superior unmixing results compared with other advanced unmixing methods.

Towards explainable AI for hyperspectral image classification in Edge Computing environments

Several Remote Sensing applications in Hyperspectral Imaging rely on Artificial Intelligence techniques, particularly Deep Neural Networks, which often outperform other algorithms. However, despite their effectiveness, these techniques are considered opaque for the non-linearity of the underlying model, making a conscious results interpretation difficult even for experts in the application field. The present work aims to describe the results of our experiments toward Explainable Artificial Intelligence techniques for hyperspectral remote sensing image classification in Edge Computing environments. The proposed technique extends the traditional 2D Gradient-weighted Class Activation Mapping (Grad-CAM) to 3D Convolutional Neural Networks. Moreover, we use spectral-cumulative Grad-CAM and class probability as complementary methods. The experimental results confirm that the observer can interpret the choices made within neural network layers by visualizing the activation volumes provided by the proposed method.

Compressive full-Stokes polarization and flexible hyperspectral imaging with efficient reconstruction

Polarization hyperspectral compressive imaging has wide application prospects due to reflecting rich information of the target from the spatial, spectral and polarization dimensions. However, compressive imaging methods are limited in system complexity and reconstruction efficiency. In this paper, the combination of a quarter-wave plate, a linear polarizer and a liquid crystal tunable filter enables compressive full-Stokes polarization and flexible hyperspectral imaging. Then, the four Stokes parameters can be quickly reconstructed with high precision through numerical operations and compressive sensing. Both simulations and experiments verify the effectiveness of this method. Compared with the feature scaling method, this work shortens the reconstruction time by 83.73%, and greatly improves the reconstruction result of the first Stokes parameter from limited reconstruction to accurate solution. This work not only gives insights from efficient reconstruction to system design, but also creates conditions for the application of polarization hyperspectral imaging in specific fields.

Development and comparison of classification models on VIS-NIR hyperspectral imaging spectra for qualitative detection of the Staphylococcus aureus in fresh chicken breast

Chicken is at risk of contaminated foodborne pathogens in the production process. Timely and nondestructive detection of foodborne pathogens in chicken is essential for food security. The study aims to explore the feasibility of developing efficient classification models for qualitative detection of Staphylococcus aureus in chicken breast using the hyperspectral imaging technique. Principal component analysis was used to process the full spectral information and three wavelength selection methods (competitive adaptive reweighted sampling, genetic algorithm, and successive projections algorithm) were applied to extract effective wavelengths. These methods were combined with the support vector machine algorithm to develop conventional classification models, respectively. In addition, a convolutional neural network model based on deep learning was designed and trained for comparison. The performance of the convolutional neural network model was significantly better than that of conventional classification models. The overall accuracy for chicken sample classifications was improved from 83.88% to 91.38%. The results demonstrated that deep learning can effectively extract spectral features and promote the application of hyperspectral imaging in foodborne pathogens detection of chicken products.

Advances in Machine Learning and Hyperspectral Imaging in the Food Supply Chain

Food quality and safety are the essential hot issues of social concern. In recent years, there has been a growing demand for real-time food information, and non-destructive testing is gradually replacing traditional manual sensory testing and chemical analysis methods with lagging and destructive effects and has strong potential for application in the food supply chain. With the maturity and development of computer science and spectroscopic techniques, machine learning and hyperspectral imaging (HSI) have been widely demonstrated as efficient detection techniques that can be applied to rapidly evaluate sensory characteristics and quality attributes of food products nondestructively and efficiently. This paper first briefly described the basic concepts of hyperspectral imaging and machine learning, including the imaging process of HSI, the type of algorithms contained in machine learning, and the data processing flow. Secondly, this paper provided an objective and comprehensive overview of the current applications of machine learning and HSI in the food supply chain for sorting, packaging, transportation, storage, and sales, based on the state-of-art literature from 2017 to 2022. Finally, the potential of the technology is further discussed to provide optimized ideas for practical application.