Reflectance Imaging Research Articles

Interpretable multimodal classification for age-related macular degeneration diagnosis.

Explainable Artificial Intelligence (XAI) is an emerging machine learning field that has been successful in medical image analysis. Interpretable approaches are able to "unbox" the black-box decisions made by AI systems, aiding medical doctors to justify their diagnostics better. In this paper, we analyze the performance of three different XAI strategies for medical image analysis in ophthalmology. We consider a multimodal deep learning model that combines optical coherence tomography (OCT) and infrared reflectance (IR) imaging for the diagnosis of age-related macular degeneration (AMD). The classification model is able to achieve an accuracy of 0.94, performing better than other unimodal alternatives. We analyze the XAI methods in terms of their ability to identify retinal damage and ease of interpretation, concluding that grad-CAM and guided grad-CAM can be combined to have both a coarse visual justification and a fine-grained analysis of the retinal layers. We provide important insights and recommendations for practitioners on how to design automated and explainable screening tests based on the combination of two image sources.

Natural Cover suppression patterns used in PRODES Caatinga mapping

Abstract. Since 2022, we have mapped native vegetation removal for all Brazilian biomes within the scope of the PRODES Project. In this study, our main objective was to present interpretation criteria for feature detection used to identify natural vegetation removal throughout the Caatinga. To this end, we created a catalog of typical patterns that can be used for supervised and unsupervised deforestation mapping. We used PRODES data from 2023 year's PRODES Caatinga based on the composition of top of atmosphere (TOA) reflectance images from MSI/Sentinel-2 satellite. The original reflectance data was used to generate false color images in the following combination: band 8 (Red), band 11 (Green) and band 4 (Blue). This image was then visually classified into different natural vegetation removal classes. These are good examples of what PRODES interpreters have encountered throughout this diverse region and can help any future mapping efforts of the Caatinga.

Blue light reflectance imaging in non-perfusion areas detection: insights from multimodal analysis

DesignA retrospective, cross-sectional image analysis using a convenience sample.SubjectsFive cases selected based on the availability of comprehensive imaging data.MethodsThis study involved a retrospective review of images from five cases, focusing on the use of retinal monochromatic blue light reflectance (BLR) imaging to detect non-perfusion areas. Two cases of sickle-cell retinopathy demonstrated peripheral retinal non-perfusion identified through widefield fluorescein angiography. Three other cases—one with branch retinal vein occlusion, one with branch retinal artery occlusion, and one presenting paracentral acute middle maculopathy showed focal macular non-perfusion detected by structural OCT and OCTA. The areas of nonperfused retinal tissue, confirmed by fluorescein angiography, OCT, and OCTA, were then correlated with findings from the BLR image. This correlation aimed to identify any potential associations between these imaging modalities.Main outcome measuresEnhance understanding of the utilization of retinal monochromatic BLR images as a non-perfusion biomarker.ResultsThe perfusion defects identified through fluorescein angiography were qualitatively correlated with hypo-reflective regions observed in the BLR images. A notable correlation was also observed between the OCTA deep capillary plexus findings and the BLR images. Additionally, areas of retinal thinning identified on structural OCT thickness maps corresponded with the hypo-reflective regions in the BLR images. This indicates the potential of BLR in identifying non-perfused retinal areas.ConclusionsThis study reinforces the evidence, through OCT, OCTA, and angiographic correlation, that the BLR can effectively identify areas of retinal non-perfusion in a non-invasive manner. Further research is warranted to assess the method’s sensitivity, specificity, and limitations. While the interaction of blue light with the retina, leading to specular reflections and scattering, is established, this research represents a pioneering effort in suggesting which specific retinal structures may be implicated in this phenomenon. This novel insight opens avenues for deeper exploration into the underlying mechanisms and potential clinical applications of utilizing the BLR imaging technique for assessing retinal vascular abnormalities.

Detecting Water Stress in Winter Wheat Based on Multifeature Fusion from UAV Remote Sensing and Stacking Ensemble Learning Method

The integration of remote sensing technology and machine learning algorithms represents a new research direction for the rapid and large-scale detection of water stress in modern agricultural crops. However, in solving practical agricultural problems, single machine learning algorithms cannot fully explore the potential information within the data, lacking stability and accuracy. Stacking ensemble learning (SEL) can combine the advantages of multiple single machine learning algorithms to construct more stable predictive models. In this study, threshold values of stomatal conductance (gs) under different soil water stress indices (SWSIs) were proposed to assist managers in irrigation scheduling. In the present study, six irrigation treatments were established for winter wheat to simulate various soil moisture supply conditions. During the critical growth stages, gs was measured and the SWSI was calculated. A spectral camera mounted on an unmanned aerial vehicle (UAV) captured reflectance images in five bands, from which vegetation indices and texture information were extracted. The results indicated that gs at different growth stages of winter wheat was sensitive to soil moisture supply conditions. The correlation between the gs value and SWSI value was high (R2 > 0.79). Therefore, the gs value threshold can reflect the current soil water stress level. Compared with individual machine learning models, the SEL model exhibited higher prediction accuracy, with R2 increasing by 6.67–17.14%. Using a reserved test set, the SEL model demonstrated excellent performance in various evaluation metrics across different growth stages (R2: 0.69–0.87, RMSE: 0.04–0.08 mol m−2 s−1; NRMSE: 12.3–23.6%, MAE: 0.03–0.06 mol m−2 s−1) and exhibited excellent stability and accuracy. This research can play a significant role in achieving large-scale monitoring of crop growth status through UAV, enabling the real-time capture of winter wheat water deficit changes, and providing technical support for precision irrigation.

Diagnosing Fungal Infection in Wheat Kernels by Integrating Spectroscopic Technology and Digital Color Imaging System: Artificial Neural Network, Principal Component Analysis and Correlation Feature Selection Techniques

ABSTRACTContamination of cereal grain, especially wheat, with fungal infections can cause significant economic impacts and it endangers the health of humans and livestock. This study aims to appraise the UV/VIS–NIR and digital color (RGB) imaging systems and spectroscopic methodology to detect wheat kernels infected by fungi such as Penicillium expansum and Fusarium graminearum. NIR spectra of 190–1100 nm at 10 nm intervals, visible color reflectance images and non‐visible reflectance images of wheat kernels in the ultraviolet and near‐infrared ranges were applied to develop the multi‐layer perceptron (MLP) artificial neural network model. The optimum wavelengths were selected by application of the principal component analysis (PCA) after preprocessing the raw spectra. A confusion matrix was used in the correlation feature selection method (CFS) for the decision tree classifier of selected features. The results showed that the four UV wavelengths of 310, 330, 400, and 410 nm were the best wavelengths using PCA to distinguish healthy and unhealthy wheat kernels. Considering the intensity of the wavelengths as the neural network inputs, samples were classified into healthy and unhealthy categories with an accuracy of 90.9%. Also, 18 features of color images in RGB, LAB, HSV, HSI, YCbCr, and YIQ spaces provided the highest average accuracy of 44.4% in classifying healthy and infected wheat kernels by using a CCD Proline camera in the ultraviolet range. In contrast, other cameras in the visible and invisible range showed low accuracy. Furthermore, the best classification accuracy of the healthy and infected samples by the use of the CFS method was obtained at 88.1%. Based on the findings, spectroscopic methodology proved to be highly effective for detecting, classifying and automatic cleaning of various agricultural seeds, with a particular emphasis on wheat kernals.

A novel management challenge in age-related macular degeneration: Artificial intelligence and expert prediction of geographic atrophy

PurposeThe progression of geographic atrophy (GA) secondary to age-related macular degeneration (AMD) is highly variable among individuals. Prediction of the progression is critical to identify patients who will benefit most from the first treatments currently approved. The aim of this study was the investigation of the value and difference in predictive power between ophthalmologists and artificial intelligence (AI) to reliably assess individual speed of GA progression. DesignProspective, expert and AI comparison study. ParticipantsEyes with natural progression of GA from a prospective study (NCT02503332). MethodsOphthalmologists predicted yearly growth speed of GA, as well as selecting the potentially faster growing lesions from two eyes based on fundus autofluorescence (FAF), near-infrared reflectance (NIR), and optical coherence tomography (OCT). A deep learning algorithm predicted progression solely on the baseline OCT (Spectralis, Heidelberg Engineering, Germany). Main Outcome MeasuresAccuracy, weighted kappa (κ), and concordance index (c-index) between the prediction made by ophthalmology specialists, ophthalmology residents and the AI. ResultsA total of 134 eyes of 134 patients from a phase II clinical trial were included, among those 53 were from the sham arm and 81 from untreated fellow eyes. 2880 gradings were performed by four ophthalmologists. Human experts reached an accuracy of 0.37, 0.43, 0.41 and a κ of 0.06, 0.16, 0.18 on FAF, NIR+OCT and FAF+NIR+OCT, respectively. On a pairwise comparison task, human experts achieved a c-index of 0.62, 0.59 and 0.60. Automated AI-based analysis reached an accuracy of 0.48 and a κ of 0.23 on the first task, and a c-index of 0.69 on the second task solely utilizing OCT imaging. ConclusionsPrediction of individual progression will become an important task for patient counseling, most importantly with treatments becoming available. Human gradings improved with the availability of OCT. However, automated AI performed better than ophthalmologists in several comparisons. AI-supported decisions improve clinical precision, access to timely care for the community, and socioeconomic feasibility in the management of the leading cause for irreversible vision loss.

Non-destructive estimation for Kyoho grape shelf-life using Vis/NIR hyperspectral imaging and deep learning algorithm

Grape shelf-life estimation is a substantial challenge for the grape industry. The objective of this study is to investigate the potential of grape shelf-life estimation using HSI technique and a deep learning algorithm. The visible and near-infrared (400.68–1001.61 nm) hyperspectral reflectance images data of grape samples was acquired and preprocessed with different spectral preprocessing methods. Additionally, a stacked denoising autoencoder (SDAE)-based deep learning algorithm was developed to extract deep features from pixel-level hyperspectral data of grapes, and then these features were used as inputs to establish support vector machine (SVM) models for estimating grape shelf-life. Furthermore, SVM, one-dimensional convolutional neural network (1D CNN) and long short-term memory (LSTM) models were used as traditional machine learning and end to end models for comparison. The results demonstrated that the SDAE-SVM model achieved reasonable recognition accuracy of 100 % and 98.125 % for the shelf-life of grapes in the training and test sets, respectively. The overall results suggested that SDAE-based deep learning method can be used as a powerful tool to deal with large-scale hyperspectral data as well as this research confirms the feasibility of non-destructive estimation for grapes shelf-life by the combination of HSI technique and deep learning method, which would provide a valuable guidance for shelf-life estimation of other postharvest fruit.

Targeted isolation of extracellular vesicles from cell culture supernatant using immuno-affinity chromatography

Extracellular vesicles (EVs) have emerged as promising therapeutics with broad clinical applications as diagnostic biomarkers and therapeutic drug delivery systems. Yet, these biopharmaceuticals pose a challenge in terms of manufacturing due to their complexity and heterogeneity. Despite advancements in the field, current purification technologies lack scalability and/or selectivity. Affinity chromatography (AC) − coupling unmatched specificity and scalability − could be used to simplify purification processing and generate clinical-grade EVs with higher titers and purity.In the present work, we report the implementation of an immuno-AC resin to capture and purify EVs directly from clarified cellular feedstocks. Firstly, to guide and support marker selection, vesicle phenotype characterization was conducted using single particle interferometric reflectance image sensing (SP-IRIS) coupled with immunofluorescence. CD81 was the marker which shown to be more present and more likely to have the other markers (CD63 and CD9). Thus, anti-CD81 VHH ligand was generated and evaluated towards recombinant CD81 protein and CD81 bearing EV particles using surface plasmon resonance (SPR). Different chromatographic studies with Anti-CD81 ligand immobilized onto agarose beads resin were conducted to optimize the process parameters (residence time, dynamic binding capacity and impurity clearance). At residence time of 2 min, on average 40 % of pure triple tetraspanin-positive EV fraction was recovered. The enrichment in EV particles herein obtained, based on scale-up calculations, it would be possible to produce 1 × 1013 EVs from a 1L cell culture, while meeting impurity requirements in a single-step purification process (impurity removal over 2 log reduction value). A single-step purification process is possible, enabling the successful isolation of homogeneous EVs population, counting with a final HCP titer of 60 ng/mL and 9 ng/mL of dsDNA impurities. EV’s morphological integrity and internalization ability were also demonstrated, showcasing elution’s efficiency under mild conditions.Overall, this work contributes to the development of a novel, highly specific, AC technology using a camelid-derived affinity ligand which, bridging the scalability requirements demanded of large-scale production, could potentiate the advent of EV-based therapies.

Reference-based image super-resolution of hyperspectral and red-green-blue image for determination of wheat kernel quality using deep learning networks

In the process of cultivation and harvest, wheat kernel quality is highly susceptible to various factors, such as disease, mildew, atrophy and impurities, and detection of kernel quality is essential to avoid hazard proliferation, facilitate product grading, and ensure food safety. Possessing abundant image and spectral characteristics, hyperspectral imaging (HSI) has gained impressive achievements in kernel quality analysis, but its low spatial resolution limits its detection accuracy. In this study, reference-based image super-resolution (RefSR) of HSI and Red-Green-Blue image was adopted to improve resolution to determine wheat kernel quality using deep learning networks. Firstly, RefSR was conducted by the improved transformer network with dual-branch feature extraction and weighted fusion operation and achieved excellent RefSR with significant resolution improvement, peak signal to noise ratio of 35.521 and structural similarity index of 0.97, outweighing the existing state-of-the-art networks. Then, the reflectance images (RIs) of effective wavelengths (EWs) from generated HSI images were combined with the residual network with a spatial, channel attention and multi-scale residual to determine wheat kernel quality. Precise analysis was achieved with the accuracy in calibration, validation and prediction sets of 100.00%, 95.26% and 92.78%. RefSR provides a novel and efficient approach for obtaining HSI images of high spatial resolution and facilitates the application of HSI in analysis of crop kernels. RIs of several sporadic EWs can be easily acquired and processed, achieving field and rapid kernel detection. Therefore, the proposed method furnishes the efficient, accurate and applicable determination of wheat kernel quality.

Classification of Citrus Leaf Diseases Using Hyperspectral Reflectance and Fluorescence Imaging and Machine Learning Techniques

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.

Biofabricated osteoblast-derived nanovesicles as extracellular vesicle mimics for bone repair

Osteoblast-derived extracellular vesicles (EVs) have demonstrated therapeutic utility for bone repair as transporters of key biomolecules capable of accelerating biomineralisation and tissue repair. The clinical translation of these biologically derived nanoparticles, however remains limited due to scalability, heterogeneity and standardisation issues. Herein we investigate the generation of nanovesicles (NVs) from mineralising osteoblasts by extrusion directly compared against natural EV counterparts from the same parental cells.Mineralising osteoblast-derived EVs (MO-EVs) were isolated via ultracentrifugation from cell culture media. The parental osteoblasts were then processed via serial extrusion to <200 nm. EVs and NVs were characterised by comparing their size, concentration and morphology. The presence of tetraspanin markers was detected by Single Particle Interferometric Reflectance Imaging Sensor (SP-IRIS). Osteoblasts viability and metabolic activity was assessed after both EV and NV-treatment before comparing their mineralising potency via alizarin red staining.EVs and NVs exhibited similar diameters of approximately 100 nm with a vesicular morphology. EVs were found to be richer in proteins and exhibited a significantly more negative ζ-potential compared to NVs. SP-IRIS analysis confirmed the presence of CD9, CD63 and CD81 in EVs. At both 1 and 10 μg/mL, EVs and NVs reduced the metabolic activity of osteoblasts, however, this was not associated with any cytotoxic effects. The biomineralisation study performed using osteoblasts showed that only EVs significantly increased mineral deposition (p < 0.05) compared to untreated control.In this study, we have established for the first time the biofabrication of cell-derived nanovesicles as a promising alternative to extracellular vesicles derived from mineralising osteoblasts.

A Review of the Application of the Laser-Light Backscattering Imaging Technique to Agricultural Products

Growing concerns about food safety and waste have increased consumer demand for high-quality agricultural products, particularly at the postharvest stage. This demand has prompted the development of non-destructive methods to assess or inspect the internal quality of fruits and vegetables. The backscattering imaging technique, also known as diffuse reflectance imaging, is considered a highly promising approach. Numerous studies have focused on practical applications, using laser light at selected wavelengths to develop quick multispectral methods. Due to the rapid interaction of photons with biological tissue, together with the highly computational performance of machine vision, backscattering imaging can offer a valuable alternative to traditional methods. Its primary benefits include quick measurements without chemical sample preparation, easy integration with high-throughput automatic quality control, and reduced waste, since this non-destructive technique does not damage samples. This review presents a comprehensive overview of backscattering imaging, including the measurement geometry, data analysis, and design considerations for vision systems. Additionally, it explores this technique’s advantages, challenges, and accuracy, as demonstrated using various case studies.

Constructing 3D SPAD distribution using hyperspectral LiDAR point cloud by PROSPECT model inversion

ABSTRACT Accurately estimating leaf chlorophyll concentrations in a three-dimensional (3D) manner is crucial for precisely monitoring crop growth and development. To reduce destructive detection and achieve real-time estimation, researchers often utilize crop leaves’ SPAD (Soil and Plant Analyzer Development) to represent relative chlorophyll content. However, SPAD measurement is time-consuming and laborious work. Many inversion methods have been proposed to estimate SPAD based on one-dimensional reflectance or two-dimensional spectral images, which rely on external light sources and also cannot obtain 3D distribution. Hyperspectral LiDAR (HSL), as an active remote sensing technology, can acquire target spatial and spectral data simultaneously, making 3D SPAD inversion and reconstruction possible. This research aims to invert SPAD values with HSL spectral information aided by the PROSPECT model and reconstruct 3D SPAD distribution with HSL spatial information. Firstly, we constructed new ratio chlorophyll indices (RCI) based on the multi-feature analysis method with the PROSPECT model and HSL spectral data. Next, we employed RCIs to modify the Look-Up Table (LUT) cost function (SPAD-LUT) of the PROSPECT model, which was validated with the ANGERS dataset. Then, we conducted SPAD inversion with HSL spectral information from vegetable crop samples and integrated the inverted SPAD values into their corresponding 3D coordinates, achieving 3D SPAD distribution reconstruction. The predicted values of the model inversion with the HSL dataset are consistent with the measured SPAD (R2 = 0.421, RMSE = 10.615). The results indicate that our method can invert sample SPAD values and reconstruct their distribution with HSL point cloud despite their complex spatial structure.

Assessment of land dynamics transformation into tea plantations using support vector machine

Precisely monitoring land use dynamics and spatial distributions is essential for sustainable development and long-term land management. Tea is one of the leading beverage crops cultivated in Bangladesh, expanding rapidly in northern districts and forcing land use change. This study aims to decipher the expansion of tea cultivation and land dynamics transformation to tea land areas in the northern region of Bangladesh by using Landsat 5 TM and Landsat 8 OLI/TIRS surface reflectance images. The supervised support vector machine (SVM) method was used for classification purposes, resulting in three classified maps for the years 2004, 2013 and 2022, having overall accuracies of 91.43 %, 98.67 %, and 98.48 % and kappa coefficients of 89.51, 98.37, and 98.13, respectively. The images were classified into six land use classes: agriculture, tea cultivation area, settlement, waterbody, bare land, and forest. Land transformation results reveal that overall tea land increased by 41.08 % from 2004 to 2022, experiencing a downward trend during 2004–2013, while from 2013 to 2022, a dramatic rise of 70.01 % (equal to 4683.60 ha) tea cultivation area was found in Panchagarh district. The bare land was among the most highly transformed land classes into tea plantations, followed by the forest area. This study provides evidence of transforming underutilized land class into a profitable land use practice, i.e., tea plantation. Thus, cutting-edge technologies would be imperative in land transformation detection for sustainable land management and policy implications for the sustainable development of small landholding livelihoods and the tea industry.

Detection and characterization of aeolian sand strips on the beach using permanent laser scanning

Aeolian sand strips are bedforms that often form on the beach during high wind events. The sand strips can move across the beach resulting in sediment transport, but these sediment fluxes have not yet been quantified. Permanent laser scanning (PLS) provides an opportunity to gather long-term and detailed information about sand strips. Here, we present a method that detects sand strips in PLS reflectance images and extracts their wavelength, orientation, and height. The detection method is used on a month-long hourly dataset of Noordwijk beach, the Netherlands. The average sand strip wavelength (13 m), height (3 cm), and orientation (onshore-oblique) found in this dataset are consistent with previous research. Migration rates are obtained by examining successive images of sand strip patterns. Using the concurrent sand strip height, the sediment transport rate associated with sand strips is estimated. During a single 8-hour event, a total sediment flux of 0.16 m3/m is observed of which 0.07 m3/m in onshore direction. Furthermore, the detection method is shown to be applicable to other beaches by calibrating its thresholds. This opens a new possibility for future research to use PLS to investigate the behavior and sediment transport rates of aeolian sand strips.

Construction of a stable YOLOv8 classification model for apple bruising detection based on physicochemical property analysis and structured-illumination reflectance imaging

Effective and accurate detection of bruises at all stages has always been a challenge in non-destructive grading of apples. In this study, the visible structured-illumination reflectance imaging (SIRI) combing with deep learning method was proposed to identify bruised ‘Fuji’ apples at four different time stages (0, 6, 12 and 24 h). The macroscopic/microscopic structures and physicochemical properties of bruised tissue were measured and analyzed to determine the relationship between bruising time and these properties, as well as how they affect the accuracy of bruising detection. Results indicated that classification accuracy increased with the decrease of water and total phenolic content of the bruised tissue, as well as with the increase of color browning and bruised area. The YOLOv8 model achieved the highest detection accuracy (99.5 %) and stability. This research enhances understanding of apple bruise optics and aids in developing advanced nondestructive testing techniques.

Hyperspectral imaging with deep learning for quantification of tissue hemoglobin, melanin, and scattering.

Hyperspectral cameras capture spectral information at each pixel in an image. Acquired spectra can be analyzed to estimate quantities of absorbing and scattering components, but the use of traditional fitting algorithms over megapixel images can be computationally intensive. Deep learning algorithms can be trained to rapidly analyze spectral data and can potentially process hyperspectral camera data in real time. A hyperspectral camera was used to capture wide-field reflectance images of in vivo human tissue at 205 wavelength bands from 420 to 830nm. The optical properties of oxyhemoglobin, deoxyhemoglobin, melanin, and scattering were used with multi-layer Monte Carlo models to generate simulated diffuse reflectance spectra for 24,000 random combinations of physiologically relevant tissue components. These spectra were then used to train an artificial neural network (ANN) to predict tissue component concentrations from an input reflectance spectrum. The ANN achieved low root mean square errors in a test set of 6000 independent simulated diffuse reflectance spectra while calculating concentration values more than 4000× faster than a conventional iterative least squares approach. In vivo finger occlusion and gingival abrasion studies demonstrate the ability of this approach to rapidly generate high-resolution images of tissue component concentrations from a hyperspectral dataset acquired from human subjects.

52746 Using artificial intelligence hierarchical ensemble deep learning model to diagnose basal-cell carcinoma in reflective confocal microscopy images

52746 Using artificial intelligence hierarchical ensemble deep learning model to diagnose basal-cell carcinoma in reflective confocal microscopy images

Hyperspectral reflectance imaging for visualizing reducing sugar content, moisture, and hollow rate in red ginseng

Red ginseng (RG) has been traditionally valued in Northeast Asia for its health-enhancing properties. Recent advancements in hyperspectral imaging (HSI) offer a non-destructive, efficient, and reliable method to assess critical quality indicators of RG, such as reducing sugar content (RSC), water content (WC), and hollow rate (HR). This study developed predictive models using HSI technology to monitor these quality indicators over the spectral range of 400–1700 nm. Image features were enhanced using Principal Component Analysis (PCA) and Minimum Noise Fraction (MNF), followed by classification through Spectral Angle Mapping (SAM). The best-performing model for RSC achieved an R2 value of 0.6198 and a root mean square error (RMSE) of 0.013. For WC, the optimal model obtained an R2 value of 0.6555 and an RMSE of 0.014. The spatial distribution of RSC, WC, and HR was effectively visualized, demonstrating the potential of HSI for on-site quality control of RG. This study provides a foundation for real-time, non-invasive monitoring of RG quality, addressing industry needs for rapid and reliable assessment methods.

Polarized reflection removal with dual-stream attention guidance

Addressing the challenge of removing reflections from images captured through glass surfaces holds significant importance across various practical applications. Existing methods often fall short due to reliance on intermediate predictions or special constraints, leading to undesirable artifacts. This paper addresses this issue by recognizing the inherent correlation between background and reflection. Leveraging the distinct optical properties of polarized images, we propose a novel dual-stream network with feature attention guidance for reflection removal. The model, taking multi-channel polarized images as input, exploits the optical characteristics of transmitted and reflected light to tackle the ill-posed problem. Our dual-stream structure predicts both reflectance and transmission images simultaneously, capitalizing on the unique relationship between these two elements. Additionally, we introduce the innovative differential feature fusion (DFF) block to enhance communication between the two streams. Our model shows superior performance to the state-of-the-art methods when tested on real-world polarization datasets.