Pushbroom Hyperspectral Imaging System Research Articles

A Vicarious Technique for Understanding and Diagnosing Hyperspectral Spatial Misregistration.

Pushbroom hyperspectral imaging (HSI) systems intrinsically measure our surroundings by leveraging 1D spatial imaging, where each pixel contains a unique spectrum of the observed materials. Spatial misregistration is an important property of HSI systems because it defines the spectral integrity of spatial pixels and requires characterization. The IEEE P4001 Standards Association committee has defined laboratory-based methods to test the ultimate limit of HSI systems but negates any impacts from mounting and flying the instruments on airborne platforms such as unmanned aerial vehicles (UAV's) or drones. Our study was designed to demonstrate a novel vicarious technique using convex mirrors to bridge the gap between laboratory and field-based HSI performance testing with a focus on extracting hyperspectral spatial misregistration. A fast and simple extraction technique is proposed for estimating the sampled Point Spread Function's width, along with keystone, as a function of wavelength for understanding the key contributors to hyperspectral spatial misregistration. With the ease of deploying convex mirrors, off-axis spatial misregistration is assessed and compared with on-axis behavior, where the best performance is often observed. In addition, convex mirrors provide an easy methodology to exploit ortho-rectification errors related to fixed pushbroom HSI systems, which we will show. The techniques discussed in this study are not limited to drone-based systems but can be easily applied to other airborne or satellite-based systems.

Linear Spatial Misregistration Detection and Correction Based on Spectral Unmixing for FAHI Hyperspectral Imagery.

In push-broom hyperspectral imaging systems, the sensor rotation to the optical plane leads to linear spatial misregistration (LSM) in hyperspectral images (HSIs). To compensate for hardware defects through software, this paper develops four methods to detect LSM in HSIs. Different from traditional methods for grayscale images, the method of fitting the sum of abundance (FSAM) and the method of searching for equal abundance (SEAM) are achieved by hyperspectral unmixing for a selected rectangular transition areas containing an edge, which makes good use of spatial and spectral information. The method based on line detection for band-interleaved-by-line (BIL) images (LDBM) and the method based on the Fourier transform of BIL images (FTBM) aim to characterize the slope of line structure in BIL images and get rid of the dependence on scene and wavelength. A full strategy is detailed from aspects of data selection, LSM detection, and image correction. The full spectrum airborne hyperspectral imager (FAHI) is China's new generation push-broom scanner. The HSIs obtained by FAHI are tested and analyzed. Experiments on simulation data compare the four proposed methods with traditional methods and prove that FSAM outperforms other methods in terms of accuracy and stability. In experiments on real data, the application of the full strategy on FAHI verifies its effectiveness. This work not only provides reference for other push-broom imagers with similar problems, but also helps to reduce the requirement for hardware calibration.

HyScreen: A Ground-Based Imaging System for High-Resolution Red and Far-Red Solar-Induced Chlorophyll Fluorescence

Solar-induced chlorophyll fluorescence (SIF) is used as a proxy of photosynthetic efficiency. However, interpreting top-of-canopy (TOC) SIF in relation to photosynthesis remains challenging due to the distortion introduced by the canopy's structural effects (i.e., fluorescence re-absorption, sunlit-shaded leaves, etc.) and sun-canopy-sensor geometry (i.e., direct radiation infilling). Therefore, ground-based, high-spatial-resolution data sets are needed to characterize the described effects and to be able to downscale TOC SIF to the leafs where the photosynthetic processes are taking place. We herein introduce HyScreen, a ground-based push-broom hyperspectral imaging system designed to measure red (F687) and far-red (F760) SIF and vegetation indices from TOC with single-leaf spatial resolution. This paper presents measurement protocols, the data processing chain and a case study of SIF retrieval. Raw data from two imaging sensors were processed to top-of-canopy radiance by dark-current correction, radiometric calibration, and empirical line correction. In the next step, the improved Fraunhofer line descrimination (iFLD) and spectral-fitting method (SFM) were used for SIF retrieval, and vegetation indices were calculated. With the developed protocol and data processing chain, we estimated a signal-to-noise ratio (SNR) between 50 and 200 from reference panels with reflectance from 5% to 95% and noise equivalent radiance (NER) of 0.04 (5%) to 0.18 (95%) mW m-2 sr-1 nm-1. The results from the case study showed that non-vegetation targets had SIF values close to 0 mW m-2 sr-1 nm-1, whereas vegetation targets had a mean F687 of 1.13 and F760 of 1.96 mW m-2 sr-1 nm-1 from the SFM method. HyScreen showed good performance for SIF retrievals at both F687 and F760; nevertheless, we recommend further adaptations to correct for the effects of noise, varying illumination and sensor optics. In conclusion, due to its high spatial resolution, Hyscreen is a promising tool for investigating the relationship between leafs and TOC SIF as well as their relationship with plants' photosynthetic capacity.

Burn characterization using object-oriented hyperspectral image classification.

This paper presents a new approach based on hyperspectral imaging combined with an object-oriented classification method that allows the generation of burn depth classification maps facilitating easier characterization of burns. Hyperspectral images of 14 patients diagnosed with burns on the upper and lower limbs were acquired using a pushbroom hyperspectral imaging system. The images were analyzed using an object-oriented classification approach that uses objects with specific spectral, textural and spatial attributes as the minimum unit for classifying information. The method performance was evaluated in terms of overall accuracy, sensitivity, precision and specificity computed from the confusion matrix. The results revealed that the approach proposed in this study performed well in differentiating burn classes with a high level of overall accuracy (95.99% ± 0.60%), precision (97.30% ± 2.46%), sensitivity (97.23% ± 3.02%) and specificity (98.02% ± 1.98%). In conclusion, the object-based approach for burns hyperspectral images classification can provide maps that can help surgeons identify with better precision different depths of burn wounds.

Design and Experiment of Push-Broom Hyperspectral Microscopic Imaging System

Design and Experiment of Push-Broom Hyperspectral Microscopic Imaging System

Assessing the impact of illumination on UAV pushbroom hyperspectral imagery collected under various cloud cover conditions

The recent development of small form-factor (<6 kg), full range (400–2500 nm) pushbroom hyperspectral imaging systems (HSI) for unmanned aerial vehicles (UAV) poses a new range of opportunities for passive remote sensing applications. The flexible deployment of these UAV-HSI systems have the potential to expand the data acquisition window to acceptable (though non-ideal) atmospheric conditions. This is an important consideration for time-sensitive applications (e.g. phenology) in areas with persistent cloud cover. Since the majority of UAV studies have focused on applications with ideal illumination conditions (e.g. minimal or non-cloud cover), little is known to what extent UAV-HSI data are affected by changes in illumination conditions due to variable cloud cover. In this study, we acquired UAV pushbroom HSI (400–2500 nm) over three consecutive days with various illumination conditions (i.e. cloud cover), which were complemented with downwelling irradiance data to characterize illumination conditions and in-situ and laboratory reference panel measurements across a range of reflectivity (i.e. 2%, 10%, 18% and 50%) used to evaluate reflectance products. Using these data we address four fundamental aspects for UAV-HSI acquired under various conditions ranging from high (624.6 ± 16.63 W·m2) to low (2.5 ± 0.9 W·m2) direct irradiance: atmospheric compensation, signal-to-noise ratio (SNR), spectral vegetation indices and endmembers extraction. For instance, two atmospheric compensation methods were applied, a radiative transfer model suitable for high direct irradiance, and an Empirical Line Model (ELM) for diffuse irradiance conditions. SNR results for two distinctive vegetation classes (i.e. tree canopy vs herbaceous vegetation) reveal wavelength dependent attenuation by cloud cover, with higher SNR under high direct irradiance for canopy vegetation. Spectral vegetation index (SVIs) results revealed high variability and index dependent effects. For example, NDVI had significant differences (p < 0.05) across illumination conditions, while NDWI appeared insensitive at the canopy level. Finally, often neglected diffuse illumination conditions may be beneficial for revealing spectral features in vegetation that are obscured by the predominantly non-Lambertian reflectance encountered under high direct illumination. To our knowledge, our study is the first to use a full range pushbroom UAV sensor (400–2500 nm) for assessing illumination effects on the aforementioned variables. Our findings pave the way for understanding the advantages and limitations of ultra-high spatial resolution full range high fidelity UAV-HSI for ecological and other applications.

Design and research of fully automatic push-broom hyperspectral microscopic imaging system

The predecessor of China Optics was China Optics and Applied Optics Digest, which was founded in 1985. At that time, it was the only retrieval journal in the field of optics in China. At the end of 2008, China Optics and Applied Optics Digest was renamed

Spectral imaging application to discriminate different diets of live rainbow trout (Oncorhynchus mykiss)

The main aim of this study was to evaluate the feasibility of hyperspectral imagery for determining the correlation between fish skin changes and different diets. Rainbow trout (Oncorhynchus mykiss) were fed either a commercial based diet (N = 80) or a 100% plant-based diet (N = 80). Hyperspectral images were made using a push-broom hyperspectral imaging system in the spectral region of 394–1009 nm. All images were calibrated using dark and white reference, and the average spectral data from the region of interest were extracted. Seven spectral pre-treatment methods were used, including Savitzky-Golay (SG), First Derivative (FD), Second Derivative (SD), Standard Normal Variate (SNV), Multiplicative Scatter Correction(MSC) and Continuum removal (CR) then a support vector machine (SVM) with linear kernel was applied to establish the classification models. Overall classification models developed from full wavelengths with different preprocessing methods showed good performance (Correct Classification Rate (CCR) = 0.83, Kappa = 0.66) when coupled with SG and SD or SG and MSC. The overall results indicate that the integration of Vis/NIR hyperspectral imaging system and machine learning algorithms have promise for discriminating different diets based on the live fish skin. These procedures can be used to not only identify the diet used for fish feeding in the case where we are not sure but also monitor different diets impacts on live fish skin for more precise monitoring of fish status during cultivation and ultimately for better implementation of precision fish farming.

Differentiation of Maize Ear Rot Pathogens, on Growth Media, with Near Infrared Hyperspectral Imaging

In this study, the major maize ear rot pathogens were differentiated from one another, on growth media, with near-infrared (NIR) hyperspectral imaging. Fungal isolates of four pathogens commonly associated with maize grain (Fusarium verticillioides, F. graminearum s.s., F. boothii and Stenocarpella maydis) were plated on potato dextrose agar, in triplicate, and incubated at 25°C for 5 days. Images were collected with a SisuChema short-wave infrared (SWIR) pushbroom hyperspectral imaging system ranging from 1000 to 2500 nm. Pixel and object wise classification algorithms were compared to determine the best approach. These were evaluated with principal component analysis and partial least squares discriminant analysis. The pathogens were distinguished using three-way classification models that were validated with independent images. The object wise approach proved to be more effective in distinguishing ear rot pathogens with a higher average overall classification accuracy (93.75%). The object wise models achieved two 100% classification accuracies from the four pathogen models, while none of the pixel wise classification models were error free. The specificity and sensitivity further highlighted the superiority of the object wise approach. The high accuracy of object wise classification (> 86%) can be attributed to the representation of the mean spectra per object. NIR hyperspectral imaging can thus accurately distinguish between the major maize ear rot pathogens, with object wise classification proving to be the optimal approach.

Spatial resolution enhancement for pushbroom-based microscopic hyperspectral imaging.

Hyperspectral imaging (HSI) is a promising tool for microscopic histopathology studies. Pushbroom microscopic hyperspectral imaging systems are widely used because of their low cost and easy implementation. However, the spatial resolution of pushbroom HSI systems is limited by the width of the optical entrance slit. A narrower slit leads to longer exposure time and slower imaging speed. In this paper, we explored several spatial resolution enhancement algorithms, originally designed for remote-sensing hyperspectral imaging, for pushbroom microscopic HSI systems. Our results demonstrate that those algorithms could effectively achieve a higher spatial resolution without sacrificing imaging speed.

Non-destructive and rapid discrimination of hard-to-cook beans using hyperspectral imaging

Dry beans stored under sub-optimal conditions tend to develop hard-to-cook (HTC) defect, which extends the cooking time making them less palatable while reducing their nutritional value. The current methods of identifying HTC beans are time-consuming, destructive, and unreliable. A rapid non-destructive inspection technique for pre-screening beans could help identify and discard HTC beans prior to processing. To this end, the potential of hyperspectral imaging technique covering the entire visible to near infrared (NIR) spectral range (400‒2500 nm) was evaluated for rapid and non-destructive identification of HTC beans. The HTC phenomenon was artificially induced in healthy white beans using two different combinations of suboptimal storage conditions of temperature and relative humidity (35℃, 75% RH for 45 days and 60℃, 75% RH for 10 days). Subsequently, the beans were cooked for specified durations and their hardness measured using a texture analyzer. The HTC and control (i.e. easy-to-cook (ETC)) beans were scanned with push-broom hyperspectral imaging systems. Results indicate that both sets of storage conditions rendered the beans HTC but the phenomenon induced by the two different methods was detected in different spectral ranges using hyperspectral imaging. Wavelengths across the entire visible and NIR ranges of electromagnetic spectrum were found useful in detecting HTC as beans stored at 35℃ and 75% RH for 45 days were identified mainly in the 1000‒2500 nm range and those stored at 60℃ and 75% RH for 10 days were identified in the 400‒1000 nm region. The degree of HTC defect could not be ascertained using this technique and requires further investigation.

Sex determination of silkworm pupae using VIS-NIR hyperspectral imaging combined with chemometrics.

To explore an accurate and non-destructive method to discriminate the sex of silkworm pupae, the visible and near-infrared (VIS-NIR) hyperspectral imaging (HSI) technique was employed in this paper. First, a total of 520 hyperspectral images of silkworm pupae of four species were captured using a push-broom HSI system in the spectral region of 363 nm to 1026 nm and then calibrated for reflectance. The mean spectral data were extracted from the region of interest (ROI). Second, five optimal wavelengths (403, 440, 505, 533, 721 nm) were selected by successive projection algorithm (SPA). Then gray-level co-occurrence matrix (GLCM) analysis was implemented on the 500 nm image. Finally, support vector machine (SVM) and radial basis function and neutral network (RBF-NN) models were established based on full spectra, textural data, spectral data and fusion data, respectively. The SVM and RBF-NN models using fusion data reached the most satisfactory performance with a high correct classification rate of 98.75%. Furthermore, the built SVM model based on fusion data could be promoted to identify the sex of another two species of silkworm pupae with accuracy of 97% and 96%, indicating that HSI technology can be served as a new method to differentiate the sex of silkworm pupae.

Geometric super-resolution on push-broom hyperspectral imaging for plasma optical emission spectroscopy

Sub-pixel shifting allows geometric super-resolution on a push-broom hyperspectral imaging system used for plasma OES.

Hyperspectral Imaging for Predicting the Internal Quality of Kiwifruits Based on Variable Selection Algorithms and Chemometric Models

We investigated the feasibility and potentiality of determining firmness, soluble solids content (SSC), and pH in kiwifruits using hyperspectral imaging, combined with variable selection methods and calibration models. The images were acquired by a push-broom hyperspectral reflectance imaging system covering two spectral ranges. Weighted regression coefficients (BW), successive projections algorithm (SPA) and genetic algorithm–partial least square (GAPLS) were compared and evaluated for the selection of effective wavelengths. Moreover, multiple linear regression (MLR), partial least squares regression and least squares support vector machine (LS-SVM) were developed to predict quality attributes quantitatively using effective wavelengths. The established models, particularly SPA-MLR, SPA-LS-SVM and GAPLS-LS-SVM, performed well. The SPA-MLR models for firmness (Rpre = 0.9812, RPD = 5.17) and SSC (Rpre = 0.9523, RPD = 3.26) at 380–1023 nm showed excellent performance, whereas GAPLS-LS-SVM was the optimal model at 874–1734 nm for predicting pH (Rpre = 0.9070, RPD = 2.60). Image processing algorithms were developed to transfer the predictive model in every pixel to generate prediction maps that visualize the spatial distribution of firmness and SSC. Hence, the results clearly demonstrated that hyperspectral imaging has the potential as a fast and non-invasive method to predict the quality attributes of kiwifruits.

Hyperspectral Imaging for Presymptomatic Detection of Tobacco Disease with Successive Projections Algorithm and Machine-learning Classifiers

We investigated the feasibility and potentiality of presymptomatic detection of tobacco disease using hyperspectral imaging, combined with the variable selection method and machine-learning classifiers. Images from healthy and TMV-infected leaves with 2, 4, and 6 days post infection were acquired by a pushbroom hyperspectral reflectance imaging system covering the spectral range of 380–1023 nm. Successive projections algorithm was evaluated for effective wavelengths (EWs) selection. Four texture features, including contrast, correlation, entropy, and homogeneity were extracted according to grey-level co-occurrence matrix (GLCM). Additionally, different machine-learning algorithms were developed and compared to detect and classify disease stages with EWs, texture features and data fusion respectively. The performance of chemometric models with data fusion manifested better results with classification accuracies of calibration and prediction all above 80% than those only using EWs or texture features; the accuracies were up to 95% employing back propagation neural network (BPNN), extreme learning machine (ELM), and least squares support vector machine (LS-SVM) models. Hence, hyperspectral imaging has the potential as a fast and non-invasive method to identify infected leaves in a short period of time (i.e. 48 h) in comparison to the reference images (5 days for visible symptoms of infection, 11 days for typical symptoms).

Early Detection of Aphid (Myzus persicae) Infestation on Chinese Cabbage by Hyperspectral Imaging and Feature Extraction

Abstract. Hyperspectral imaging (HSI) in the visible/near-infrared region coupled with textural feature extraction were used to detect aphid infestation on Chinese cabbage plants. Hyperspectral images of Chinese cabbage plants, including 31 healthy plants and 32 aphid-infested plants, were captured in the region of 380 to 1030 nm with a pushbroom HSI system. A total of 160 spectra were selected from the hyperspectral images of plants. Six important wavelengths (IWs) at 550, 671, 694, 742, 911, and 968 nm were identified by x-loadings of PC-1 and PC-2 in principle component analysis (PCA). Textural features (TFs) including contrast, homogeneity, energy, and correlation based on gray-level co-occurrence matrix (GLCM) were extracted from the six selected IWs. Finally, least squares support vector machine (LS-SVM) models were developed to differentiate between healthy and aphid-infested plants based on the IWs and TFs. The results showed that the reflectance and textural features at 694 and 742 nm offered good performance in distinguishing aphid-infested samples from healthy samples. The overall results revealed the feasibility of HSI for early detection of aphid infestation on Chinese cabbage plants and its promising potential in practice. Keywords: Aphid infestation, Chinese cabbage, Hyperspectral imaging, Least squares support vector machine, Principal component analysis, Textural feature extraction.

Spatial-scanning hyperspectral imaging probe for bio-imaging applications.

The three common methods to perform hyperspectral imaging are the spatial-scanning, spectral-scanning, and snapshot methods. However, only the spectral-scanning and snapshot methods have been configured to a hyperspectral imaging probe as of today. This paper presents a spatial-scanning (pushbroom) hyperspectral imaging probe, which is realized by integrating a pushbroom hyperspectral imager with an imaging probe. The proposed hyperspectral imaging probe can also function as an endoscopic probe by integrating a custom fabricated image fiber bundle unit. The imaging probe is configured by incorporating a gradient-index lens at the end face of an image fiber bundle that consists of about 50,000 individual fiberlets. The necessary simulations, methodology, and detailed instrumentation aspects that are carried out are explained followed by assessing the developed probe's performance. Resolution test targets such as United States Air Force chart as well as bio-samples such as chicken breast tissue with blood clot are used as test samples for resolution analysis and for performance validation. This system is built on a pushbroom hyperspectral imaging system with a video camera and has the advantage of acquiring information from a large number of spectral bands with selectable region of interest. The advantages of this spatial-scanning hyperspectral imaging probe can be extended to test samples or tissues residing in regions that are difficult to access with potential diagnostic bio-imaging applications.

Estimation and Correction of Geometric Distortion in Pushbroom Hyperspectral System for Imaging Art Paintings

Spectral imaging has been proved as a promising technology to perform scientific documentation and analysis of cultural heritage objects. Imaging systems used for this purpose varies in different forms and complexity. Most of the spectral devices are designed not specifically for cultural heritage imaging, but for other application domains like remote sensing, satellite imaging etc. While these imaging systems used for cultural heritage scanning, the same processing workflow may not be adequate to meet the required image quality. In this paper, we investigate one of the several quality parameters, geometrical distortions in a spectral image caused by the translation stage of the camera. For this study, we used a hyperspectral image dataset derived from a pushbroom hyperspectral imaging system attached in a rotational translator. Using geometrical model, we have estimated the distortion, which is function of the scanning angle and distance. Image correction has been proposed and tested over a number of images acquired at the laboratory and in a museum environment.

Near-infrared hyperspectral imaging system coupled with multivariate methods to predict viability and vigor in muskmelon seeds

A near-infrared (NIR) hyperspectral imaging (HSI) system was used to predict viability and vigor (in term of germination periods) in muskmelon seeds. Hyperspectral images of muskmelon seeds were acquired using a NIR push-broom HSI system covering the spectral range of 948–2494nm. After NIR spectra collection, all seeds underwent a germination test to confirm their viability and vigor. The spectra from seeds with 3 and 5 germination days and nongerminated seeds were further used for development of a classification model of partial least-squares discriminant analysis (PLSDA). Most effective wavelengths were selected using three model-based variable selection methods, i.e., variable important in projection (VIP), selectivity ratio (SR), and significance multivariate correlation (sMC), which selected 23, 18, and 19 optimal variables, respectively, from full set of 208 variables. The selected variables from different waveband selection methods were found genuine and significant for interpreting the prediction results of seed viability and vigor. Subsequently, the PLS-DA model was constructed using individual VIP-, SR-, or sMC-selected variables. The results demonstrated that the PLSDA model developed with the selected optimal variables from the different methods provided comparable results for the calibration set; however, the PLSDA-SR method afforded the highest classification accuracy (94.6%) for a validation set used to determine the viability and vigor of muskmelon seeds. The wavelengths selected by the different methods represents chemical components of the seed and the attribute of germination ability was chosen most often.

Pushbroom hyperspectral imaging system with selectable region of interest for medical imaging.

A spatial-scanning pushbroom hyperspectral imaging (HSI) system incorporating a video camera (VC) which is not only used for direct video imaging but also for the selection of the region of interest within the VC’s full field-of-view is presented. Using a VC for these two applications brings many benefits to a pushbroom HSI system, such as a minimized data acquisition time and smaller data storage requirement. A detailed description of the system followed by the methods and formulas used for calibration and electronic hardware interfacing were discussed and analyzed using United States Air Force resolution chart, chicken breast tissue, and fluorescent targets as test samples. The proposed concepts and developed system can find potential biomedical imaging applications and can be extended to endoscopic imaging applications as well.