Estimation of pigment concentration in LDPE via in-line hyperspectral imaging and machine learning

Research on predicting photosynthetic pigments in tomato seedling leaves based on near-infrared hyperspectral imaging and machine learning

Citrus diseases are significant threats to citrus groves, causing financial losses through reduced fruit size, blemishes, premature fruit drop, and tree death. The detection of citrus diseases via leaf inspection can improve grove management and mitigation efforts. This study explores the potential of a portable reflectance and fluorescence hyperspectral imaging (HSI) system for detecting and classifying a control group and citrus leaf diseases, including canker, Huanglongbing (HLB), greasy spot, melanose, scab, and zinc deficiency. The HSI system was used to simultaneously collect reflectance and fluorescence images from the front and back sides of the leaves. Nine machine learning classifiers were trained using full spectra and spectral bands selected through principal component analysis (PCA) from the HSI with pixel-based and leaf-based spectra. A support vector machine (SVM) classifier achieved the highest overall classification accuracy of 90.7% when employing the full spectra of combined reflectance and fluorescence data and pixel-based analysis from the back side of the leaves, whereas a discriminant analysis classifier yielded the best accuracy of 94.5% with the full spectra of combined reflectance and fluorescence data and leaf-based analysis. Among the diseases, control, scab, and melanose were classified most accurately, each with over 90% accuracy. Therefore, the integration of the reflectance and fluorescence HSI with advanced machine learning techniques demonstrated the capability to accurately detect and classify these citrus leaf diseases with high precision.

Due to developments in European legislation, several halogenated flame retardants are banned due to their toxicity, and the use of phosphor-based flame retardants in plastics is increasing. A revision of ammonium polyphosphate (APP) flame retardant revealed that it is an eye irritant and toxic, thus posing a health issue. Hence APP identification is needed for enabling safe recycling of plastic waste streams. Herein an industrial in-line method for quantitative estimation of APP in low density polyethylene (LDPE) and polypropylene (PP) is demonstrated, by using an industrial hyperspectral imaging system (955 to 1700 nm) and principal component analysis (PCA). Spectra of plastic samples with varying concentrations of APP were applied to build and calibrate a quantitative determination method. PCA and band area ratios (of selected bands) were made and fitted with continuous functions for concentration determination. The plastic samples were characterised by elemental analysis, attenuated total reflection, differential scanning calorimetry, and thermogravimetric analysis. The PCA model outperforms the band area ratio model and predicts APP concentrations between 24.3 and 1.5 wt% in LDPE (R2 = 0.98) and 20.0 and 1.7 wt% in PP (R2 = 0.97). Unknown samples with APP ranging from 23.7 to 2.7 wt% in LDPE and from 18.6 to 2.3 wt% in PP were predicted and correlated to the actual concentrations. The proposed approach is valuable for the plastic recyclers and waste management industries where inline concentration determination of flame retardants is key.

The nutritional and commercial value of lettuce (Lactuca sativa L.) is determined by its foliar pigment and phenolic composition, which varies among cultivars. This study aimed to assess the capacity of hyperspectral and applied multispectral imaging, combined with machine learning algorithms, to predict and map key biochemical traits, such as chloroplastidic pigments (chlorophylls and carotenoids) and extrachloroplastidic pigments (anthocyanins, flavonoids, and phenolic compounds). Eleven cultivars exhibiting contrasting pigmentation profiles were grown under controlled greenhouse conditions, and their chlorophyll a and b, carotenoid, anthocyanin, flavonoid, and total phenolic contents were evaluated. Spectral reflectance data were acquired via a Headwall hyperspectral sensor and a MicaSense multispectral sensor, and the pigment contents were quantified via solvent extraction and a UV microplate reader. We developed predictive models via seven machine learning approaches, with partial least squares regression (PLSR) and random forest (RF) emerging as the most robust algorithms for pigment estimation. Chlorophyll a and b are highly and positively correlated (r > 0.9), which is consistent with their hyperspectral reflectance imaging results. The hyperspectral data consistently outperformed the multispectral data in terms of predictive accuracy (e.g., R2 = 0.91 and 0.76 for anthocyanins and flavonoids via RF) and phenolic compounds with R2 = 0.79, capturing subtle spectral features linked to biochemical variation. Spatial maps revealed strong genotype-dependent heterogeneity in pigment and phenolic distributions, supporting the potential of this approach for cultivar discrimination and pigment phenotyping. These findings demonstrate that hyperspectral imaging integrated with data-driven modelling offers a powerful, nondestructive framework for the biochemical monitoring of leafy vegetables, supporting breeding, precision agriculture, and food quality assessment.

Detection of common defects on oranges using hyperspectral reflectance imaging

From leaf to canopy: Inversion of lettuce pigment distribution using hyperspectral imaging technology combined with deep learning algorithms.

In order to quickly distinguish infertile eggs from fertile eggs, the hyperspectral imaging technology consisting of imaging and spectral information was used for detecting the fertile information of eggs. Before hatching eggs were incubated, a hyperspectral imaging system (wavelength between 400 to 1 000 nm) was used to acquire the images one-by-one manually. The characteristic information of ratios of length to short axis, elongation, roundness and the ratios of the yolk area to the whole area was extracted based on the images. The normalization method was used as the spectral data preprocessing, and then 155 spectral characteristic variables were extracted from 520 nm waveband through the correlation coefficient method. Principal component analysis (PCA) method was adopted to reduce the dimensions of image-spectrum fusion information; the top six principal components were extracted. Support vector machine (SVM) method was used to establish classification of fertile and infertile eggs models, which are based on image, spectrum and image-spectrum fusion information respectively. The accuracy rates of the SVM models were 84.00%, 90.00% and 93.00% respectively. The experimental results show that the model based on image-spectrum fusion information technology is superior to the single information model. Hyperspectral transmission imaging technology is effective and feasible to detect the fertile hatching eggs before incubation. Keywords: hyperspectral image, hatching eggs, information fusion, nondestructive detection DOI: 10.3965/j.ijabe.20150804.1672 Citation: Zhu Z H, Liu T, Xie D J, Wang Q H, Ma M H. Nondestructive detection of infertile hatching eggs based on spectral and imaging information. Int J Agric & Biol Eng, 2015; 8(4): 69－76.

Plastic and paper foreign materials in tobacco shreds mainly originate from tobacco processing and packaging. These materials are highly similar to tobacco shreds in color and size, making them difficult for traditional machine vision systems to detect, thus posing a significant risk to product quality and safety. To address this issue, this study proposes a method combining hyperspectral imaging (HSI) and machine learning. After segmenting and denoising hyperspectral images to extract pixel-level spectral data, principal component analysis (PCA) is used to construct classification models of support vector machine (SVM), linear discriminant analysis (LDA), and random forest (RF). The PCA dimensionality reduction results show that SVM accuracy decreased by 0.08%, LDA by 2.67%, while RF maintained a 100% accuracy rate. The LDA model trained with full spectral bands achieved 100% accuracy, demonstrating better robustness and generalization ability. The final classification results are mapped back to the original hyperspectral images for visual classification and detection of foreign materials. This method can reliably detect partially obscured foreign materials. This study overcomes the limitations of detection based on color and size, providing a solution for real-time quality control in the food and pharmaceutical industries.

Near infrared (NIR) hyperspectral imaging and hyperspectral image analysis were evaluated for their potential to distinguish between Fusarium verticillioides infected and sound whole maize ( Zea mays L.) kernels. Hyperspectral images of infected and sound kernels were acquired using a Spectral Dimensions MatrixNIR camera with a spectral range of 960–1662nm and a sisuChema hyperspectral pushbroom imaging system with a spectral range of 1000–2498 nm. Background, bad pixels and shading were removed using exploratory principal component analysis (PCA) on absorbance images. PCA could be used effectively on the cleaned images to identify classes including infected and non-infected regions on individual kernels. A distinct difference between infected and sound kernels along principal component (PC) one with two distinguishable clusters was found. The loading line plot of the first PC of the sisuChema hypercube showed important absorbance peaks for the two classes, i.e. 1960nm and 2100nm for the infected class and U50 nm, 2300nm and 2350nm for the non-infected class. Partial least squares discriminant analysis (PLS-DA) was applied. The coefficient of determination was 0.73 for the MatrixNIR image and 0.86 for the sisuChema image, both after three PLS components. These PLS-DA models could be used to calculate predictions on a test set image. The predictions could be shown as prediction images and an acceptable root mean square error of prediction was obtained (0.23). NIR hyperspectral imaging has the potential to be used as a rapid, objective means of indentifying fungal infected maize kernels and infected regions.

Hyperspectral imaging technique in simultaneous reflectance and transmittance modes was investigated for detection of hollow or bloater damage on whole pickles that was caused by mechanical injury during harvesting and handling or developed during the brining process. Normal and bloated pickle samples were collected from a commercial pickle processing plant. Hyperspectral images were acquired from the pickle samples using a hyperspectral reflectance and transmittance imaging system covering the spectral range of 400-1000 nm. Principal component analysis was applied to the hyperspectral images of the pickle samples, and the second principal component score images were used for defect detection by means of image segmentation method. An overall classification accuracy of 86% was achieved using this method. Transmittance images at 675-1000 nm were much more effective for internal defect detection compared to reflectance images for the visible region of 500-675 nm. With further improvement, the hyperspectral imaging system could meet the need of bloated pickles detection in a commercial plant setting.

The striking variation in colour and maculation of bird eggs has fascinated biologists since centuries, and many hypotheses based on mechanical, physiological or signalling functions have been proposed for its evolution by natural and sexual selection. Protoporphyrin is the main eggshell pigment found in brown maculated eggs, and is assumed to function as a mediator of these selection processes. It is a precursor of heme with pro‐oxidant properties, and hence a link between brown maculation and female condition has been proposed and tested in a number of studies, albeit with contrasting results. A variety of different visual methods have been used to quantify outer eggshell pigmentation, which has been assumed to correspond to overall quantity of protoporphyrin in the shell. Yet, this relationship has rarely been tested. The aim of this study was to apply four commonly used methods to assess pigmentation in great tit eggs with protoporphyrin as the predominant eggshell pigment, and to compare the results of these methods. Specifically, we 1) ranked eggshell pigmentation by human naked eye, 2) applied a granularity approach and 3) measured spectrophotometric reflectance of eggshell pigments. Second, we estimated the relationship between outer eggshell pigmentation (i.e. estimated by three different methods above) and true protoporphyrin concentration deposited in the entire shell measured by HPLC. Among‐method estimates were significantly correlated for the traits describing pigment ‘darkness’ only. While the model including scores based on human naked eyes explained 16% of the variance of pigment concentration in the entire shell, spectrometry explained 27%, and the granularity approach explained 40%. Thus, the estimation of true pigment concentration in the entire shell from the visible outer side of the shell is most reliable with the granularity approach. It is relevant for studies where the maintenance of the integrity of the eggs is essential.

Olfactory Atlases with an Array of Porphyrinoids Coated ZnO Nanoparticle

Pigments are the biochemical material basis for energy and material exchange between vegetation and the external environment, therefore quantitative determination of pigment content is crucial. Unmanned Aerial Vehicle (UAV)-borne remote sensing data coupled with radiative transfer models (RTM) provide marked strengths for three-dimensional (3D) visualization, as well as accurate determination of the distributions of pigment content in forest canopies. In this study, Light Detection and Ranging (LiDAR) and hyperspectral images acquired by a multi-rotor UAV were assessed with the PROSAIL model (i.e., PROSPECT model coupled with 4SAIL model) and were synthetically implemented to estimate the horizontal and vertical distribution of pigments in canopies of Ginkgo plantations in a study site within coastal southeast China. Firstly, the fusion of LiDAR point cloud and hyperspectral images was carried out in the frame of voxels to obtain fused hyperspectral point clouds. Secondly, the PROSAIL model was calibrated using specific model parameters of Ginkgo trees and the corresponding look-up tables (LUTs) of leaf pigment content were constructed and optimally selected. Finally, based on the optimal LUTs and combined with the hyperspectral point clouds, the horizontal and vertical distributions of pigments in different ages of ginkgo trees were mapped to explore their distribution characteristics. The results showed that 22-year-old ginkgo trees had higher biochemical pigment content (increase 3.37–55.67%) than 13-year-old ginkgo trees. Pigment content decreased with the increase of height, whereas pigment content from the outer part of tree canopies showed a rising tendency as compared to the inner part of canopies. Compared with the traditional vegetation index models (R2 = 0.25–0.46, rRMSE = 16.25–19.37%), the new approach developed in this study exhibited significant higher accuracies (R2 = 0.36–0.60, rRMSE = 13.53–16.86%). The results of this study confirmed the effectiveness of coupling the UAV-borne LiDAR and hyperspectral image with the PROSAIL model for accurately assessing pigment content in ginkgo canopies, and the developed estimation methods can also be adopted to other regions under different conditions, providing technical support for sustainable forest management and precision silvicuture for plantations.

Knowledge of the constituents of the Martian surface and their distributions over the planet informs us about Mars’ geomorphological formation and evolutionary history. In this research, an unsupervised end-to-end unmixing model using autoencoders is proposed, that can produce physically plausible abundance maps of the surface mineralology from hyperspectral imaging data. 1 Introduction Mars’ formationary and evolutionary history can be inferred by studying the mineralogical constituents of the Martian surface and their distributions. Knowledge of the specific minerals present and their abundances constrain the possible histories of the planet’s surface (Liu et al., 2016). Hyperspectral imaging of Mars from the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM; Murchie et al., 2007) provides unprecedented insight into the distribution of surface mineralogy. Usually, minerals are identified from spectral absorption features and comparison to spectral libraries (see Viviano-Beck et al., 2014). This can be incredibly time consuming, and relies upon detailed a priori knowledge of the surface chemistry and geology. In order to obtain constituent mineralogical spectra (endmembers) and their fractional abundances one must perform spectral unmixing. Machine learning methods have been used extensively for Earth based hyperspectral imaging analysis, but only a few works to our knowledge have attempted to produce end-to-end unmixing models: Guo et al. (2015); Palsson et al. (2018, 2017); Zhang et al. (2018). Data obtained from hyperspectral imaging of Mars is essentially of the same format as that obtained from Earth, so similar kinds of analysis can be performed. 2 Method Signal decomposition techniques, namely principal component analysis (PCA) and independent component analysis (ICA) are used in order to investigate the presence - and the approximate number - of distinct, clearly detectable signals (component spectra) within CRISM imaging data. Clustering algorithms are then used to obtain an alternative estimate of the number of distinct identifiable mineralogical components present within the spectra. The intuition is that the optimal number of clusters identified should match the number of components determined through the use of signal decomposition techniques. Autoencoders (a type of neural network consisting of an encoder and decoder function that are trained to reproduce their input to their output) are then trained on the spectra from the image. They learn an approximation of the LMM whereby the weights of the decoder correspond to the endmember spectra and the activations of the encoder are the fractional abundances of the endmembers for a given spectrum (Palsson et al., 2018, 2017). The number of endmembers in the learned approximation to the LMM are then set as the estimates of the number of component signals from PCA and clustering. Figure 1 shows a schematic of this. Learned endmembers are extracted from the networks and abundance maps are produced by projecting the activation of each neuron in the decoder to the pixel location of the input spectrum for all of the spectra in the image. These techniques all get around the requirement of detailed a priori knowledge by making no assumptions about the state of the surface. To our knowledge, this is the first work to address the use of unsupervised neural networks for an end-to-end spectral unmixing of Martian hyperspectral images. Figure 1: Schematic view of an autoencoder used for spectral unmixing. 3 Results and Conclusion The signal decomposition techniques provide strong evidence for the presence of detectable mineralogical signals within the data, and the number of signal components estimated from PCA are consistent with the estimates from clustering. PCA, ICA, and the abundance maps produced by the trained autoencoders all appear to convincingly replicate features of the surface topography. Figure 2 shows an example of this, comparing the quicklook image for phyllosilicates (a false colour RGB image constructed from wavelengths using spectral parameters; Viviano-Beck et al., 2014) to one of the produced abundance maps. This provides strong evidence that these methods have learnt mineralogically significant spectra, and that the autoencoders have learnt a physically plausible approximation to the LMM. This is despite the fact that at no spatial information is provided at any point to any of these methods – their results are purely based on spectral information. Hence, when combined with methods to estimate the number of endmembers present, one can construct an unsupervised end-to-end linear unmixing model using autoencoders, that learns to identify physically plausible mineralogical endmembers. Once trained, this pipeline will allow the large number of CRISM data sets to be characterised and abundances mapped over the surface of the planet. This method is also extendable to any kind of planetary hyperspectral imaging, allowing simple mineralogical mapping on an unprecedented scale.   Figure 2: Quicklook image for phyllosilicates (a false colour RGB image made using spectral parameters; Viviano-Beck et al., 2014) alongside one of the produced abundance maps.

Hyperspectral reflectance imaging to classify lettuce varieties by optimum selected wavelengths and linear discriminant analysis