Multispectral Research Articles

Sun Glint-Aware Restoration (SUGAR): a robust sun glint correction algorithm for UAV imagery to enhance monitoring of turbid coastal environments.

Sun glint contamination on unmanned aerial vehicles (UAV) imagery is a ubiquitous problem and poses a significant impediment in the retrieval of water quality parameters for coastal monitoring applications. Previous studies using near-infrared (NIR) and regression-based sun glint corrections have shown overcorrection at turbid regions as water-leaving NIR radiance is non-negligible. A spatial shift in the band channels would also result in suboptimal correction in the visible spectrum. Recent total variation (TV) methods show promise in reducing spectral variation associated with glint-affected regions and achieve effective correction of sun glint while leaving non-glint regions largely unaltered. To that end, this study proposes an open-source Sun Glint-Aware Restoration (SUGAR) algorithm that bridges principles in NIR and TV methods for the effective correction of sun glint in multispectral and hyperspectral UAV imagery. The present study shows that SUGAR achieves the best sun glint correction performance among existing regression and pixel-based sun glint correction methods when applied on UAV imagery of turbid and shallow regions. Around 40-80% of the total variation at glint-affected regions have been reduced while preserving features in non-glint regions. Validation of SUGAR with in situ UAV flight surveys and turbidity measurements in the coastal region of Singapore demonstrated significant improvement in turbidity retrieval, with root-mean-squared error (RMSE) reducing from 0.464 to 0.183 FNU and 0.551 to 0.285 FNU for multispectral and hyperspectral imagery, respectively.

A 20 m spatial resolution peatland extent map of Alaska

Peatlands are prevalent across northern regions, including bogs, fens, marshes, meadows, and select tundra wetlands that all vary in size (e.g., 0.01 s to 10 s km2) and shape (e.g., circular to elongated). However, our best remotely sensed products describing the regional-scale distribution of peatland extents are constrained to 1 km2 pixels, often representing notable sub-pixel heterogeneity and local-scale uncertainties. Here we develop a new 20 m spatial resolution wall-to-wall ~1.5 million km2 peatland map of Alaska, using peat cores, ground observations, and sub-meter resolution image interpretation. Ground-data were used to train machine learning classifiers to detect peatlands using a fusion of Sentinel-1 (Dual-polarized Synthetic Aperture Radar), Sentinel-2 (Multi-Spectral Imager), and derivatives from the Arctic Digital Elevation Model (ArcticDEM), that were spatially constrained by a peatland suitability model. Statewide peatland mapping (overall agreement:85%) identified peatlands to cover 4.6, 10.4, and 5.3% of polar, boreal, and maritime ecoregions, respectively, and 7.3% of the total terrestrial land area. This new dataset will improve the representation of peatland carbon, nutrient, and fire dynamics across Alaska.

Irrigation Water Treated with Oxygen Nanobubbles Decreases Irrigation Volume While Maintaining Turfgrass Quality in Central Chile

The efficient use of water resources is a growing priority in multiple sectors, including the turfgrass industry. Nanobubbles (NB) represent an innovative technology that, by enriching solutions with various gases, offers significant benefits in several industrial areas. In crop irrigation, they have been shown to increase dissolved oxygen in the root zone and thereby boost yields. The objective of this study was to evaluate the impact of the use of oxygen NB in irrigation water on turfgrass quality, considering different levels of water restriction (0%, 30%, and 50% of daily crop evapotranspiration), compared to conventional irrigation. During the summer of 2024, trials were conducted using turf quality indices based on multispectral reflectance and RGB digital image analysis. The results showed that the use of NB allowed for a reduction in irrigation by 50% without compromising turf quality, reaching values similar to treatments without water restriction. In contrast, treatment with the same restriction but without NB (WNB50%) showed a deterioration in quality. This study shows NB as an innovative tool to optimize water use, with great potential for applications in landscape green areas, promote water use efficiency, and reduce the costs associated with irrigation.

Integration of UAV Multispectral Remote Sensing and Random Forest for Full-Growth Stage Monitoring of Wheat Dynamics

Wheat is a key staple crop globally, essential for food security and sustainable agricultural development. The results of this study highlight how innovative monitoring techniques, such as UAV-based multispectral imaging, can significantly improve agricultural practices by providing precise, real-time data on crop growth. This study utilized unmanned aerial vehicle (UAV)-based remote sensing technology at the wheat experimental field of the Hebei Academy of Agriculture and Forestry Sciences to capture the dynamic growth characteristics of wheat using multispectral data, aiming to explore efficient and precise monitoring and management strategies for wheat. A UAV equipped with multispectral sensors was employed to collect high-resolution imagery at five critical growth stages of wheat: tillering, jointing, booting, flowering, and ripening. The data covered four key spectral bands: green (560 nm), red (650 nm), red-edge (730 nm), and near-infrared (840 nm). Combined with ground-truth measurements, such as chlorophyll content and plant height, 21 vegetation indices were analyzed for their nonlinear relationships with wheat growth parameters. Statistical analyses, including Pearson’s correlation and stepwise regression, were used to identify the most effective indices for monitoring wheat growth. The Normalized Difference Red-Edge Index (NDRE) and the Triangular Vegetation Index (TVI) were selected based on their superior performance in predicting wheat growth parameters, as demonstrated by their high correlation coefficients and predictive accuracy. A random forest model was developed to comprehensively evaluate the application potential of multispectral data in wheat growth monitoring. The results demonstrated that the NDRE and TVI indices were the most effective indices for monitoring wheat growth. The random forest model exhibited superior predictive accuracy, with a mean squared error (MSE) significantly lower than that of traditional regression models, particularly during the flowering and ripening stages, where the prediction error for plant height was less than 1.01 cm. Furthermore, dynamic analyses of UAV imagery effectively identified abnormal field areas, such as regions experiencing water stress or disease, providing a scientific basis for precision agricultural interventions. This study highlights the potential of UAV-based remote sensing technology in monitoring wheat growth, addressing the research gap in systematic full-cycle analysis of wheat. It also offers a novel technological pathway for optimizing agricultural resource management and improving crop yields. These findings are expected to advance intelligent agricultural production and accelerate the implementation of precision agriculture.

Multispectral Interferometric Scattering Microscopy Imaging of Gold and Silver Nanoparticles

Multispectral Interferometric Scattering Microscopy Imaging of Gold and Silver Nanoparticles

Multispectral polarization image demosaicing using redundant Stokes representation

This paper proposes a deep-learning-based demosaicing algorithm, multispectral polarization demosaicing with redundant Stokes (MPD-RS), designed for multispectral polarization filter arrays. The proposed MPD-RS effectively learns the correlation across spatial, spectral, and polarization domains, utilizing a newly constructed dataset of multispectral polarization images (MSPIs). Initially, MPD-RS performs interpolation using a position-variant convolutional kernel to generate a preliminary MSPI. This is followed by conversion to a new Stokes representation, to our knowledge, where the data is decomposed into four components, including a term to capture polarization redundancy. The intensity component is processed with a multi-stage three-dimensional convolutional network, while the remaining components are handled by a lightweight, attention-based network. Experimental results validate the effectiveness of MPD-RS, demonstrating superior peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM) for MSPI reconstruction, with an average PSNR improvement of 3.873 dB over the Global Cross-Attention Network, as well as reduced mean squared error in Stokes parameters. The method maintains high accuracy across images with a diverse range of polarization levels, highlighting its adaptability.

Immune microenvironment spatial landscapes of tertiary lymphoid structures in gastric cancer

BackgroundTertiary lymphoid structures (TLS) correlate with tumour prognosis and immunotherapy responses in gastric cancer (GC) studies. However, understanding the complex and diverse immune microenvironment within TLS requires comprehensive analysis.MethodsWe examined the prognostic impact of TLS within the tumour core (TC) of 59 GC patients undergoing immunotherapy. Multispectral fluorescence imaging was employed to evaluate variations in immune cell infiltration across different TLS sites among 110 GC patients, by quantifying immune cell density and spatial characteristics. We also generated a single-cell transcriptomic atlas of TLS-positive (n = 4) and TLS-negative (n = 8) microenvironments and performed spatial transcriptomics (ST) analysis on two samples.ResultsTLS presence in the TC significantly correlated with improved immune-related overall survival (P = 0.049). CD8+LAG-3−PD-1+TIM-3−, CD4+PD-L1+, and CD4+FoxP3− T cell densities were significantly higher in the TLS within TC compared to tumour and stromal regions. Immune cells within TLS exhibited closer intercellular proximity than those outside TLS. Five key density and spatial characteristics of immune cells within TLS in the TC were selected to develop the Density and Spatial Score risk model. Single-cell RNA sequencing revealed strong intercellular interactions in the presence of TLS within the microenvironment. However, TLS-absent environment facilitated tumour cell interactions with immune cells through MIF- and galectin-dependent pathways, recruiting immunosuppressive cells. ST analysis confirmed that T and B cells co-localise within TLS, enhancing immune response activation compared to cancer nests and exerting a strong anti-tumour effect.ConclusionsTLS presence facilitates frequent cell-to-cell communication, forming an active immune microenvironment, highlighting the prognostic value of TLS.

Self-supervised constrained super-resolution fast coded spectral imaging system

Current coded aperture spectral imaging techniques suffer from slow imaging speeds and poor image quality at low sampling rates. To address these issues, we propose a self-supervised constrained super-resolution fast coded spectral imaging system by combining a physical process of spectral imaging with image fusion super-resolution techniques. We constructed a discrete cosine transform spectrometer (DCTS) and successfully acquired low-resolution hyperspectral images (LR-HSI) and high-resolution multispectral images (HR-MSI). In order to enhance the resolution of hyperspectral images, we designed a self-supervised spectral image super-resolution network based on the physical process of spectral imaging: SSAM-Unet. This network successfully reconstructs the HR-HSI by fusing the LR-HSI with the HR-MSI. We verified the performance and generalization ability of this method through extensive experiments. Experimental results show that our proposed method can achieve good imaging results at very low sampling rates.

A Comprehensive Analysis of Soil Erosion in Coastal Areas Based on an Unmanned Aerial Vehicle and Deep Learning Approach

Soil is an important nonrenewable resource. Soil erosion is increasingly severe, and the accurate identification of soil erosion is crucial for ecological sustainability. In recent years, advancements in artificial intelligence have significantly contributed to the development of precise modeling technologies. This study utilizes high-resolution multispectral images captured by unmanned aerial vehicles and applies five machine learning models, namely convolutional neural network (CNN), support vector classification, random forest, extreme gradient boosting, and fully connected neural network, to identify regional soil erosion. The performance of each model is evaluated using F1-score, precision, and recall measurements. The results show that all models exhibit strong recognition capabilities, with CNN outperforming the others in both training and testing phases. Specifically, CNN achieved a recall rate of 0.99 on the training set and an F1-score of 0.98. Given the black-box nature of machine learning models, the shapley additive explanations method is further used for interpreting model outputs. The analysis reveals that the normalized difference salinity index and soil erodibility factor are the primary factors influencing soil erosion in the study area.

Training State-of-the-Art Deep Learning Algorithms with Visible and Extended Near-Infrared Multispectral Images of Skin Lesions for the Improvement of Skin Cancer Diagnosis

An estimated 60,000 people die annually from skin cancer, predominantly melanoma. The diagnosis of skin lesions primarily relies on visual inspection, but around half of lesions pose diagnostic challenges, often necessitating a biopsy. Non-invasive detection methods like Computer-Aided Diagnosis (CAD) using Deep Learning (DL) are becoming more prominent. This study focuses on the use of multispectral (MS) imaging to improve skin lesion classification of DL models. We trained two convolutional neural networks (CNNs)—a simple CNN with six two-dimensional (2D) convolutional layers and a custom VGG-16 model with three-dimensional (3D) convolutional layers—using a dataset of MS images. The dataset included spectral cubes from 327 nevi, 112 melanomas, and 70 basal cell carcinomas (BCCs). We compared the performance of the CNNs trained with full spectral cubes versus using only three spectral bands closest to RGB wavelengths. The custom VGG-16 model achieved a classification accuracy of 71% with full spectral cubes and 45% with RGB-simulated images. The simple CNN achieved an accuracy of 83% with full spectral cubes and 36% with RGB-simulated images, demonstrating the added value of spectral information. These results confirm that MS imaging provides complementary information beyond traditional RGB images, contributing to improved classification performance. Although the dataset size remains a limitation, the findings indicate that MS imaging has significant potential for enhancing skin lesion diagnosis, paving the way for further advancements as larger datasets become available.

Inversion of Soil Moisture Content in Silage Corn Root Zones Based on UAV Remote Sensing

Accurately monitoring soil moisture content (SMC) in the field is crucial for achieving precision irrigation management. Currently, the development of UAV platforms provides a cost-effective method for large-scale SMC monitoring. This study investigates silage corn by employing UAV remote sensing technology to obtain multispectral imagery during the seedling, jointing, and tasseling stages. Field experimental data were integrated, and supervised classification was used to remove soil background and image shadows. Canopy reflectance was extracted using masking techniques, while Pearson correlation analysis was conducted to assess the linear relationship strength between spectral indices and SMC. Subsequently, convolutional neural networks (CNNs), back-propagation neural networks (BPNNs), and partial least squares regression (PLSR) models were constructed to evaluate the applicability of these models in monitoring SMC before and after removing the soil background and image shadows. The results indicated that: (1) After removing the soil background and image shadows, the inversion accuracy of SMC for CNN, BPNN, and PLSR models improved at all growth stages. (2) Among the different inversion models, the accuracy from high to low was CNN, PLSR, BPNN. (3) From the perspective of different growth stages, the inversion accuracy from high to low was seedling stage, tasseling stage, jointing stage. The findings provide theoretical and technical support for UAV multispectral remote sensing inversion of SMC in silage corn root zones and offer validation for large-scale soil moisture monitoring using remote sensing.

Microbial Identification Through Multispectral Infrared Imaging of Colonies: A New Type of Morpho-Spectral Fingerprinting.

We describe a proof of concept for a new microbial identification technique using Direct Frequency Infrared (DFIR) multispectral imaging. This approach combines Quantum Cascade Laser (QCL) light sources with a microbolometer array in a lensless configuration to capture detailed multispectral images of microbial colonies. These optical fingerprints blend both morphological and spectral information, without the need for staining or colony picking. A proof-of-concept database was acquired, comprising 10 strains from 8 species across 4 distinct genera. In total, 2253 microbial colonies were imaged at 9 different mid-infrared wavelengths. Machine learning classification correctly identified up to 94.4% ± 1.6 of colonies fingerprints, efficiently discriminating even closely related strains. Reducing the number of wavelengths to 4 maintained high classification performance, demonstrating the method's robustness. The resulting system is faster and simpler than existing FTIR imaging systems, making it a promising tool for microbial identification.

Dataset of aerial photographs acquired with UAV using a multispectral (green, red and near-infrared) camera for cherry tomato (Solanum lycopersicum var. cerasiforme) monitoring.

A dataset of aerial photographs acquired with an Unmanned Aerial Vehicle (UAV) DJI Phantom 4 Pro is presented for monitoring a cherry tomato (Solanum lycopersicum var. cerasiforme) crop in Navolato, Mexico. Seven photogrammetric flights were carried out to assess the plant growth using a Mapir Survey 3W multispectral camera. Multispectral images with an approximate spatial resolution of 1.83 cm/px were obtained in each photogrammetric flight. These images were acquired every 15 days starting on October 15, 2021, and ending on January 23, 2022. The dataset contains the radiometrically calibrated images of the tomato crop divided into 2 open field parcels. The dataset also includes the processed photogrammetric products (ortho-mosaics) using a binary mask to exclude the soil from the plant area. The dataset was originally acquired to assess plant growth, stress levels, and overall crop health. However, this multispectral imagery dataset can also have various uses, such as creating training datasets with accurate labels or classes which can then be used to develop, train, and/or validate machine learning algorithms for image classification, object detection tasks, or change detection analysis.

Leveraging airborne imaging spectroscopy and multispectral satellite imagery to map glacial sediment plumes in Kachemak Bay, Alaska.

Study Region Kachemak Bay is a fjord-type estuary in the northern Gulf of Alaska. Water quality and habitat characteristics are strongly influenced by freshwater and sediment input from multiple glacierized catchments. Study Focus We present a new method combining imaging spectroscopy from an airborne survey with Landsat and Sentinel-2 imagery to map water surface turbidity originating from glacial runoff based on spectral abundance. We compare the spectral characteristics of turbid glacial water to clear water and generate a high resolution reference map of glacial turbidity in Kachemak Bay. This informs the subsequent analysis of a homogenized, Rayleigh corrected time series of Landsat and Sentinel-2 images and seasonal patterns of turbidity. New Hydrological Insights for the Region Our results provide the most comprehensive data set on water surface turbidity in Kachemak Bay to date and improve understanding of spatial and seasonal variability of glacial turbidity in a data sparse region. July and August have the largest plumes with median sizes around 150 km , or around a quarter of Kachemak Bay. Plume sizes typically decrease with decreasing glacier runoff in September and October. We show that imaging spectroscopy aids assessments of turbid water in glacial marine catchments across scales. Leveraging high resolution spectral information allows for water color analyses that are customized to local conditions and catchment characteristics as well as scalable to wider regions.

Towards an integrated imaging for melanoma diagnosis: A review of multispectral, hyperspectral, and thermal technologies with preliminary system development.

The diagnosis of melanoma traditionally relies on visual inspection or on the use of the dermoscope, which do not have capabilities for early and precise detection. In this review, we aimed to explore other imaging technologies that can provide non-invasive and detailed information on skin lesions, such as multispectral, hyperspectral and thermal imaging. In this regard, the systems were evaluated in terms of hardware, performance and clinical applications. Since there is currently a very big interest in developing artificial intelligence (AI) applications in dermatology, the review also focuses on analysing studies that integrated this technology with newer imaging systems. To obtain clinical validation for such systems, there is an extensive need for publicly available datasets, as the current ones are limited. Expanding and obtaining new datasets is crucial in advancing research for a more accurate melanoma diagnosis. Taking into consideration the benefits that these imaging modalities can provide if they are combined with AI, we propose a prototype that can distinguish between melanoma and its precursor, the nevus, for which the set-up, components, imaging processing pipeline and classification techniques are described. The final system benefits of the advantages provided by near infrared, thermal and visible cameras, that allow a more in-depth characterizations of melanoma for a better understanding of its behaviour, an early detection improvement and diagnostic precision.

New robotic tools for multimodal non-destructive analysis and characterization of 2D and 3D objects

The forensic field has until now been missing more versatile equipment for non-destructive characterization, analysis and inspection of 2D and 3D objects. Also, the need for increasingly frequent analysis of art forgeries, where non-destructive analysis is required at least in the first step, is calling for a multimodal solution. A prototype device for robotic analysis, imaging and mapping of 3D objects is being developed and tested to be used in these areas. The system is based on the principle of integrating imaging and analytical technologies onto six-axis robotic arms, which allow substantial flexibility in the sample size or shape. The system enables non-destructive examination of a wide spectrum of samples with complicated curvatures. The new generation of X-ray imaging detectors provide a high picture quality with a spatial resolution level in the micrometre range in 2D or 3D imaging. The basic version of the robotic scanner allows transmission X-ray imaging and mapping of the individual photons with high-sensitivity and high-resolution detectors. The broad capabilities of XRD imaging are now being complemented by X-ray fluorescence point analysis and mapping, multispectral macro imaging, and multispectral X-ray diffraction analysis.

Learning a more compact representation for low-rank tensor completion

Transform-based tensor nuclear norm (TNN) methods have gained considerable attention for their effectiveness in addressing tensor recovery challenges. The integration of deep neural networks as nonlinear transforms has been shown to significantly enhance their performance. Minimizing transform-based TNN is equivalent to minimizing the ℓ1 norm of singular values in the transformed domain, which can be interpreted as finding a sparse representation with respect to the bases supported by singular vectors. This work aims to advance deep transform-based TNN methods by identifying a more compact representation through learnable bases, ultimately improving recovery accuracy. We specifically employ convolutional kernels as these learnable bases, demonstrating their capability to generate more compact representation, i.e., sparser coefficients of real-world tensor data compared to singular vectors. Our proposed model consists of two key components: a transform component, implemented through fully connected neural networks (FCNs), and a convolutional component that replaces traditional singular matrices. Then, this model is optimized using the ADAM algorithm directly on the incomplete tensor in a zero-shot manner, meaning all learnable parameters within the FCNs and convolution kernels are inferred solely from the observed data. Experimental results indicate that our method, with this straightforward yet effective modification, outperforms state-of-the-art approaches on video and multispectral image recovery tasks.

Road feature extraction from LANDSAT-8 operational land imager images using simplified U-Net model

Automatic road feature extraction from the remote sensing (RS) imagery has a significant role in various applications such as urban planning, transportation management, and environmental monitoring. In this paper, propose a method based on the U-Net model to extract the road features from the LANDSAT-8 operational land imager (OLI) images. This method aims to extract road features in OLI images that appear as curvilinear features and roads with widths greater than 25 meters, which are mostly covered within a single pixel of the OLI resolution of multi-spectral images. The U-Net architecture is well-known for its effectiveness in image segmentation tasks. However, to optimize the complexity in the U-Net model, simplified the architecture while retaining its key components and principles. The proposed model by decreasing the convolution layers and the parameters which are involved to optimize the model called as simplified U-Net model. To train this model, the label images were generated for LANDSAT-8 OLI images, by using the saturation based adaptive thresholding and morphology (SATM) method. This reduces the efforts to draw the labels in the vector format labels and convert to raster images. The model is able to effectively generate weights, which are able to extract the road features. This model weights applied on the OLI images which covers the urban and rural areas of India, producing the satisfactory results. The experimental results with the quantitative analysis presented in the paper.

Enhancing visibility in hazy conditions: A multimodal multispectral image dehazing approach

Enhancing visibility in hazy conditions: A multimodal multispectral image dehazing approach

Remote sensing monitoring system for ship exhaust using infrared multispectral imaging

Remote sensing monitoring system for ship exhaust using infrared multispectral imaging