Search Results (1,336)

In forest ecosystems, the application of hyperspectral (HS) imagery offers unprecedented opportunities for refined identification and classification. The diversity and complexity of forest cover make it challenging for traditional remote-sensing techniques to capture subtle spectral differences. Hyperspectral imagery, however, can reveal the nuanced changes in different tree species, vegetation health status, and soil composition through its nearly continuous spectral information. This detailed spectral information is crucial for the monitoring, management, and conservation of forest resources. While Convolutional Neural Networks (CNNs) have demonstrated excellent local context modeling capabilities in HS image classification, their inherent network architecture limits the exploration and representation of spectral feature sequence properties. To address this issue, we have rethought HS image classification from a sequential perspective and proposed a hybrid model, the Spectral Weaver, which combines CNNs and Transformers. The Spectral Weaver replaces the traditional Multi-Head Attention mechanism with a Channel Attention mechanism (MCA) and introduces Centre-Differential Convolutional Layers (Conv2d-cd) to enhance spatial feature extraction capabilities. Additionally, we designed a cross-layer skip connection that adaptively learns to fuse “soft” residuals, transferring memory-like components from shallow to deep layers. Notably, the proposed model is a highly flexible backbone network, adaptable to both hyperspectral and multispectral image inputs. In comparison to traditional Visual Transformers (ViT), the Spectral Weaver innovates in several ways: (1) It introduces the MCA mechanism to enhance the mining of spectral feature sequence properties; (2) It employs Centre-Differential Convolutional Layers to strengthen spatial feature extraction; (3) It designs cross-layer skip connections to reduce information loss; (4) It supports both multispectral and hyperspectral inputs, increasing the model’s flexibility and applicability. By integrating global and local features, our model significantly improves the performance of HS image classification. We have conducted extensive experiments on the Gaofen dataset, multispectral data, and multiple hyperspectral datasets, validating the superiority of the Spectral Weaver model in forest hyperspectral image classification. The experimental results show that our model achieves 98.59% accuracy on multispectral data, surpassing ViT’s 96.30%. On the Jilin-1 dataset, our proposed algorithm achieved an accuracy of 98.95%, which is 2.17% higher than ViT. The model significantly outperforms classic ViT and other state-of-the-art backbone networks in classification performance. Not only does it effectively capture the spectral features of forest vegetation, but it also significantly improves the accuracy and robustness of classification, providing strong technical support for the refined management and conservation of forest resources. Full article

Recently, there has been a surge in the adoption of deep learning (DL) techniques, especially convolutional neural networks (CNNs), to perform hyperspectral image (HSI) classification. Although deep learners have been shown to achieve impressive performance in HSI classification, they are known to be extremely sensitive to even slight perturbations to their inputs and models. When applied in safety-critical applications, it is crucial to know how robust they really are against perturbations. However, there is still limited tool support for DL testing in terms of their robustness, nor are the existing RGB testing approaches able to address the HSI-specific challenges. In this paper, we propose a mutation analysis framework specialized for DL models trained to classify HSIs, which facilitates a critical evaluation of the robustness of DL-based HSI classifiers. First, we introduce a set of mutation operators to inject faults into the inputs and models to simulate distortions of remote sensing HSI classifiers. By utilizing the mutation testing technique, we implement a novel framework which supports the multidimensional evaluation of individual DL-based classifiers. Finally, a comparative study of the robustness of seven popular CNN-based HSI classifiers (i.e., 3D-CNN, FDSSC, HybridSN, MCNN, FC3DCNN, DWTDENSE, and Tri-CNN) on six HSI datasets is provided. Results show that FDSSC and Tri-CNN achieve higher robustness in the presence of distortions, and FDSSC maintains a relatively stable level of robustness even with few training samples. These empirical findings can be partly explained by the characteristics of the classifiers’ architectures. The results substantiate the efficacy of our evaluation framework in assessing the robustness of HSI classifiers and thus confirm its contribution to the field of remote sensing image classification. Full article

Accurately understanding the stand composition of shelter forests is essential for the construction and benefit evaluation of shelter forest projects. This study explores classification methods for dominant tree species in shelter forests using UAV-derived RGB, hyperspectral, and LiDAR data. It also investigates the impact of individual tree crown (ITC) delineation accuracy, crown morphological parameters, and various data sources and classifiers. First, as a result of the overlap and complex structure of tree crowns in shelterbelt forests, existing ITC delineation methods often lead to over-segmentation or segmentation errors. To address this challenge, we propose a watershed and multi-feature-controlled spectral clustering (WMF-SCS) algorithm for ITC delineation based on UAV RGB and LiDAR data, which offers clearer and more reliable classification objects, features, and training data for tree species classification. Second, spectral, texture, structural, and crown morphological parameters were extracted using UAV hyperspectral and LiDAR data combined with ITC delineation results. Twenty-one classification images were constructed using RF, SVM, MLP, and SAMME for tree species classification. The results show that (1) the proposed WMF-SCS algorithm demonstrates significant performance in ITC delineation in complex mixed forest scenarios (Precision = 0.88, Recall = 0.87, F1-Score = 0.87), resulting in a 1.85% increase in overall classification accuracy; (2) the inclusion of crown morphological parameters derived from LiDAR data improves the overall accuracy of the random forest classifier by 5.82%; (3) compared to using LiDAR or hyperspectral data alone, the classification accuracy using multi-source data improves by an average of 7.94% and 7.52%, respectively; (4) the random forest classifier combined with multi-source data achieves the highest classification accuracy and consistency (OA = 90.70%, Kappa = 0.8747). Full article

How to effectively extract spectral and spatial information and apply it to hyperspectral image classification (HSIC) has been a hot research topic. In recent years, the transformer-based HSIC models have attracted much interest due to their advantages in long-distance modeling of spatial and spectral features in hyperspectral images (HSIs). However, the transformer-based method suffers from high computational complexity, especially in HSIC tasks that require processing large amounts of data. In addition, the spatial variability inherent in HSIs limits the performance improvement of HSIC. To handle these challenges, a novel Spectral–Spatial Unified Mamba (SSUM) model is proposed, which introduces the State Space Model (SSM) into HSIC tasks to reduce computational complexity and improve model performance. The SSUM model is composed of two branches, i.e., the Spectral Mamba branch and the Spatial Mamba branch, designed to extract the features of HSIs from both spectral and spatial perspectives. Specifically, in the Spectral Mamba branch, a nearest-neighbor spectrum fusion (NSF) strategy is proposed to alleviate the interference caused by the spatial variability (i.e., same object having different spectra). In addition, a novel sub-spectrum scanning (SS) mechanism is proposed, which scans along the sub-spectrum dimension to enhance the model’s perception of subtle spectral details. In the Spatial Mamba branch, a Spatial Mamba (SM) module is designed by combining a 2D Selective Scan Module (SS2D) and Spatial Attention (SA) into a unified network to sufficiently extract the spatial features of HSIs. Finally, the classification results are derived by uniting the output feature of the Spectral Mamba and Spatial Mamba branch, thus improving the comprehensive performance of HSIC. The ablation studies verify the effectiveness of the proposed NSF, SS, and SM. Comparison experiments on four public HSI datasets show the superior of the proposed SSUM. Full article

Seafood quality preservation remains a critical focus in the food industry, particularly as the freeze–thaw process significantly impacts the freshness and safety of aquatic products. This study investigated quality changes in frozen mackerel subjected to two thawing methods, room temperature (RT) and running water (WT), and assessed the potential of hyperspectral imaging (HSI) for classifying these methods. After thawing, mackerel samples were stored at 5 °C for 21 days, with physicochemical, textural, and spectroscopic analyses tracking quality changes and supporting the development of a spectroscopic classification model. Compared with the WT method, the RT method delayed changes in key quality indicators, including pH, total volatile basic nitrogen (TVB-N), and total viable count (TVC), by 1–2 days, suggesting it may better preserve initial quality. Texture profile analysis showed similar trends, supporting the benefit of RT in maintaining quality. A major focus was on using HSI to assess quality and classify thawing methods. HSI achieved high classification accuracy (R_c² = 0.9547) in distinguishing thawing methods up to three days post-thaw, with 1100, 1200, and 1400 nm wavelengths identified as key spectral markers. The HIS’s ability to detect differences between thawing methods, even when conventional analyses showed minimal variation, highlights its potential as a powerful tool for quality assessment and process control in the seafood industry, enabling detection of subtle quality changes that traditional methods may miss. Full article

Environmental pollution from plastic debris is raising concerns not only for the vulnerability of marine species to ingestion but also for potential human health hazards posed by small particles, known as microplastics. In this context, marine areas suffer from a lack of constant shoreline cleanups to remove accumulated debris, preventing their degradation and fragmentation. To establish optimal strategies for streamlining plastic recovery and recycling operations, it is important to have a system for recognizing plastic debris on the beach and, more specifically, for identifying the type of polymer and mapping (e.g., topologically assessing) the distribution of plastic debris on shoreline sands. This study aims to provide an operative tool finalized to perform an in situ detection, analysis, and characterization of plastic debris present in the coastal environment (i.e., beaches), adopting a near-infrared (NIR)-based hyperspectral imaging (HSI) approach. In more detail, the possibility of identifying and classifying polymers of plastic debris by NIR-HSI in three different areas along the Pontine coastline of the Lazio region (Latina, Italy) was investigated. The study focused on three distinct beaches (i.e., Foce Verde, Capo Portiere, and Sabaudia), each characterized by a different type of sand. For each location, the adopted approach allowed for the systematic classification of the various types of plastic waste found. Three Partial Least Squares Discriminant Analysis (PLS-DA) classification models were developed using a cascade detection strategy. The first model was designed to distinguish plastics from other materials in sand samples, the second to detect plastic particles in the sand, and the third to classify the type of polymer composing each identified plastic particle. Obtained results showed that, on the one hand, plastics were correctly detected from sand and other materials (i.e., sensitivity = 0.892–1.000 and specificity = 0.909–0.996), and on the other, the recognition of polymer type was satisfactory, according to the performance statistical parameters (i.e., sensitivity = 1.000 and specificity = 0.991–1.000). This research highlights the potential of the NIR-HSI approach as a reliable, non-invasive method for plastic debris monitoring and polymer classification. Its scalability and adaptability suggest possible future integration into mobile systems, enabling large-scale monitoring and efficient debris management. Full article

Over the last few years, neural architecture search (NAS) technology has achieved good results in hyperspectral image classification. Nevertheless, existing NAS-based classification methods have not specifically focused on the complex connection between spectral and spatial data. Strengthening the integration of spatial and spectral features is crucial to boosting the overall classification efficacy of hyperspectral images. In this paper, a triple-unit hyperspectral NAS network (TUH-NAS) aimed at hyperspectral image classification is introduced, where the fusion unit emphasizes the enhancement of the intrinsic relationship between spatial and spectral information. We designed a new hyperspectral image attention mechanism module to increase the focus on critical regions and enhance sensitivity to priority areas. We also adopted a composite loss function to enhance the model’s focus on hard-to-classify samples. Experimental evaluations on three publicly accessible hyperspectral datasets demonstrated that, despite utilizing a limited number of samples, TUH-NAS outperforms existing NAS classification methods in recognizing object boundaries. Full article

Ultraviolet (UV) hyperspectral imaging shows significant promise for the classification and quality assessment of raw cotton, a key material in the textile industry. This study evaluates the efficacy of UV hyperspectral imaging (225–408 nm) using two different light sources: xenon arc (XBO) and deuterium lamps, in comparison to NIR hyperspectral imaging. The aim is to determine which light source provides better differentiation between cotton types in UV hyperspectral imaging, as each interacts differently with the materials, potentially affecting imaging quality and classification accuracy. Principal component analysis (PCA) and Quadratic Discriminant Analysis (QDA) were employed to differentiate between various cotton types and hemp plant. PCA for the XBO illumination revealed that the first three principal components (PCs) accounted for 94.8% of the total variance: PC1 (78.4%) and PC2 (11.6%) clustered the samples into four main groups—hemp (HP), recycled cotton (RcC), and organic cotton (OC) from the other cotton samples—while PC3 (6%) further separated RcC. When using the deuterium light source, the first three PCs explained 89.4% of the variance, effectively distinguishing sample types such as HP, RcC, and OC from the remaining samples, with PC3 clearly separating RcC. When combining the PCA scores with QDA, the classification accuracy reached 76.1% for the XBO light source and 85.1% for the deuterium light source. Furthermore, a deep learning technique called a fully connected neural network for classification was applied. The classification accuracy for the XBO and deuterium light sources reached 83.6% and 90.1%, respectively. The results highlight the ability of this method to differentiate conventional and organic cotton, as well as hemp, and to identify distinct types of recycled cotton, suggesting varying recycling processes and possible common origins with raw cotton. These findings underscore the potential of UV hyperspectral imaging, coupled with chemometric models, as a powerful tool for enhancing cotton classification accuracy in the textile industry. Full article

Apple proliferation is among the most important diseases in European fruit production. Early and reliable detection enables farmers to respond appropriately and to prevent further spreading of the disease. Traditional phenotyping approaches by human observers consider multiple symptoms, but these are difficult to measure automatically in the field. Therefore, the potential of hyperspectral imaging in combination with data analysis by machine learning algorithms was investigated to detect the symptoms solely based on the spectral signature of collected leaf samples. In the growing seasons 2019 and 2020, a total of 1160 leaf samples were collected. Hyperspectral imaging with a dual camera setup in spectral bands from 400 nm to 2500 nm was accompanied with subsequent PCR analysis of the samples to provide reference data for the machine learning approaches. Data processing consists of preprocessing for segmentation of the leaf area, feature extraction, classification and subsequent analysis of relevance of spectral bands. The results show that imaging multiple leaves of a tree enhances detection results, that spectral indices are a robust means to detect the diseased trees, and that the potentials of the full spectral range can be exploited using machine learning approaches. Classification models like rRBF achieved an accuracy of 0.971 in a controlled environment with stratified data for a single variety. Combined models for multiple varieties from field test samples achieved classification accuracies of 0.731. Including spatial distribution of spectral data further improves the results to 0.751. Prediction of qPCR results by regression based on spectral data achieved RMSE of 14.491 phytoplasma per plant cell. Full article

Natural secondary forests play a crucial role in global ecological security, climate change mitigation, and biodiversity conservation. However, accurately delineating individual tree crowns and identifying tree species in dense natural secondary forests remains a challenge. This study combines deep learning with traditional image segmentation methods to improve individual tree crown detection and species classification. The approach utilizes hyperspectral, unmanned aerial vehicle laser scanning data, and ground survey data from Maoershan Forest Farm in Heilongjiang Province, China. The study consists of two main processes: (1) combining semantic segmentation algorithms (U-Net and Deeplab V3 Plus) with watershed transform (WTS) for tree crown detection (U-WTS and D-WTS algorithms); (2) resampling the original images to different pixel densities (16 × 16, 32 × 32, and 64 × 64 pixels) and inputting them into five 3D-CNN models (ResNet10, ResNet18, ResNet34, ResNet50, VGG16). For tree species classification, the MSFB combined with the CNN models were used. The results show that the U-WTS algorithm achieved a recall of 0.809, precision of 0.885, and an F-score of 0.845. ResNet18 with a pixel density of 64 × 64 pixels achieved the highest overall accuracy (OA) of 0.916, an improvement of 0.049 over the original images. After incorporating MSFB, the OA improved by approximately 0.04 across all models, with only a 6% increase in model parameters. Notably, the floating-point operations (FLOPs) of ResNet18 + MSFB were only one-eighth of those of ResNet18 with 64 × 64 pixels, while achieving similar accuracy (OA: 0.912 vs. 0.916). This framework offers a scalable solution for large-scale tree species distribution mapping and forest resource inventories. Full article

Subtropical tree species identification is a crucial aspect of forest resource monitoring, and the advancement of deep learning has introduced new opportunities for subtropical tree species identification. But, its performance often relies heavily on the availability of sufficient training samples. In this study, we propose a method for tree species identification via domain generalization with hyperspectral images. The network comprises a generator and a discriminator; the former produces similar samples, and the latter outputs predicted probabilities and classification loss to guide model optimization. The results demonstrate its superiority over traditional CNN-based algorithms. Full article

In traditional neural network designs, a multilayer perceptron (MLP) is typically employed as a classification block following the feature extraction stage. However, the Kolmogorov–Arnold Network (KAN) presents a promising alternative to MLP, offering the potential to enhance prediction accuracy. In this paper, we studied KAN-based networks for pixel-wise classification of hyperspectral images. Initially, we compared baseline MLP and KAN networks with varying numbers of neurons in their hidden layers. Subsequently, we replaced the linear, convolutional, and attention layers of traditional neural networks with their KAN-based counterparts. Specifically, six cutting-edge neural networks were modified, including 1D (1DCNN), 2D (2DCNN), and 3D convolutional networks (two different 3DCNNs, NM3DCNN), as well as transformer (SSFTT). Experiments conducted using seven publicly available hyperspectral datasets demonstrated a substantial improvement in classification accuracy across all the networks. The best classification quality was achieved using a KAN-based transformer architecture. Full article

In Hyperspectral Image (HSI) classification, acquiring large quantities of high-quality labeled samples is typically costly and impractical. Traditional deep learning methods are limited in such scenarios due to their dependence on sample quantities. To address this challenge, researchers have turned to Few-Shot Learning (FSL). Although existing FSL methods improve classification performance by enhancing domain invariance through domain adaptation, they often overlook the critical issue of high inter-class similarity and large intra-class variability. Moreover, during domain alignment, features of different categories within the same domain may become confused. To address these issues, this paper proposes a novel Domain-Invariant Few-Shot Contrastive Learning (DIFSCL) method, which combines domain adaptation and contrastive learning strategies to not only learn domain-invariant features but also significantly enhance inter-class discriminability. Based on this, we further design a multi-scale adaptive attention mechanism for a hyperspectral feature extraction network to more effectively extract and optimize generalized features within the DIFSCL framework, significantly improving intra-class consistency and inter-class discriminative capability of features. Experimental results on three widely used HSI datasets demonstrate that our method significantly outperforms existing techniques in few-shot classification tasks. Full article

Background: Rodents severely damage the ecological environment of grasslands, and rodent mounds of different ages require distinct management strategies. Understanding the age of these mounds aids in formulating targeted restoration measures, which can enhance grassland productivity and biodiversity. Current surveys of rodent mounds rely on ground exposure and mound height to determine their age, which is time-consuming and labor-intensive. Remote sensing methods can quickly and easily identify the distribution of rodent mounds. Existing remote sensing images use ground exposure and mound height for identification but do not distinguish between mounds of different ages, such as one-year-old and two-year-old mounds. According to the existing literature, rodent mounds of different ages exhibit significant differences in vegetation structure, soil background, and plant diversity. Utilizing a combination of vegetation indices and hyperspectral data to determine the age of rodent mounds aims to provide a better method for extracting rodent hazard information. This experiment investigates and analyzes the age, distribution, and vegetation characteristics of rodent mounds, including total coverage, height, biomass, and diversity indices such as Patrick, Shannon–Wiener, and Pielou. Spectral data of rodent mounds of different ages were collected using an Analytical Spectral Devices field spectrometer. Correlation analysis was conducted between vegetation characteristics and spectral vegetation indices to select key indices, including NDVI₆₇₀, NDVI₇₀₅, EVI, TCARI, Ant, and SR. Multiple stepwise regression and Random Forest (RF) inversion models were established using vegetation indices, and the most suitable model was selected through comparison. Random Forest modeling was conducted to classify plateau zokor rat mounds of different ages, using both vegetation characteristic indicators and vegetation indices for comparison. The rodent mound classification models established using vegetation characteristic indicators and vegetation indices through Random Forest could distinguish rodent mounds of different ages, with out-of-bag error rates of 36.96% and 21.74%, respectively. The model using vegetation indices performed better. Conclusions: (1) Rodent mounds play a crucial ecological role in alpine meadow ecosystems by enhancing plant diversity, biomass, and the stability and vitality of the ecosystem. (2) The vegetation indices SR and TCARI are the most influential in classifying rodent mounds. (3) Incorporating vegetation indices into Random Forest modeling facilitates a precise and robust remote sensing interpretation of rodent mound ages, which is instrumental for devising targeted restoration strategies. Full article

An unescapable trend of hyperspectral image (HSI) has been toward classification with high accuracy and splendid performance. In recent years, Transformers have made remarkable progress in the HSI classification task. However, Transformer-based methods still encounter two main challenges. First, they concentrate on extracting spectral information and are incapable of using spatial information to a great extent. Second, they lack the utilization of multiscale features and do not sufficiently combine the advantages of the Transformer’s global feature extraction and multiscale feature extraction. To tackle these challenges, this article proposes a new solution named the hybrid-scale dual-domain Transformer with crisscrossed interaction (HSD²Former) for HSI classification. HSD²Former consists of three functional modules: dual-dimension multiscale convolutional embedding (D²MSCE), mixed domainFormer (MDFormer), and pyramid scale fusion block (PSFB). D²MSCE supersedes conventional patch embedding to generate spectral and spatial tokens at different scales, effectively enriching the diversity of spectral-spatial features. MDFormer is designed to facilitate self-enhancement and information interaction between the spectral domain and spatial domain, alleviating the heterogeneity of the spatial domain and spectral domain. PSFB introduces a straightforward fusion manner to achieve advanced semantic information for classification. Extensive experiments conducted on four datasets demonstrate the robustness and significance of HSD²Former. The classification evaluation indicators of OA, AA, and Kappa on four datasets almost exceed 98%, reaching state-of-the-art performance. Full article