Search Results (6,278)

In recent years, pseudo point clouds generated from depth completion of RGB images and LiDAR data have provided a robust foundation for multimodal 3D object detection. However, the generation process often introduces noise, reducing data quality and detection accuracy. Moreover, existing methods fail to effectively capture channel correlations and global contextual information during the 2D feature extraction stage after the 3D backbone network, limiting detection performance. To address these challenges, this paper proposes NRAP-RCNN, a pseudo point cloud-based 3D object detection method with two key innovations: (1) A noise-reduction sparse convolution network (NRConvNet), comprising NRConv (noise-resistant submanifold sparse convolution), SRB (sparse convolution residual block), and MHSA (multi-head self-attention). NRConv suppresses pseudo point cloud noise by jointly encoding 2D and 3D features, SRB enhances feature extraction depth and robustness, and MHSA optimizes global feature representation. (2) An attention fusion module (ECA_GCA) is introduced to enhance the feature representation of the 2D backbone network by combining channel and global contextual information. The experimental results demonstrate that NRAP-RCNN achieves 88.4% car AP (R40) on the KITTI validation set and 85.1% on the test set, significantly outperforming advanced 3D detection methods, showcasing its effectiveness in improving detection performance. Full article

Existing image fusion algorithms involve extensive models and high computational demands when processing source images that require non-rigid registration, which may not align with the practical needs of engineering applications. To tackle this challenge, this study proposes a comprehensive framework for convolutional sparse fusion in the context of non-rigid registration of visible–infrared images. Our approach begins with an attention-based convolutional sparse encoder to extract cross-modal feature encodings from source images. To enhance feature extraction, we introduce a feature-guided loss and an information entropy loss to guide the extraction of homogeneous and isolated features, resulting in a feature decomposition network. Next, we create a registration module that estimates the registration parameters based on homogeneous feature pairs. Finally, we develop an image fusion module by applying homogeneous and isolated feature filtering to the feature groups, resulting in high-quality fused images with maximized information retention. Experimental results on multiple datasets indicate that, compared with similar studies, the proposed algorithm achieves an average improvement of 8.3% in image registration and 30.6% in fusion performance in mutual information. In addition, in downstream target recognition tasks, the fusion images generated by the proposed algorithm show a maximum improvement of 20.1% in average relative accuracy compared with the original images. Importantly, our algorithm maintains a relatively lightweight computational and parameter load. Full article

Efficiently extracting effective information from the massive experimental data from physical mechanics and accurately identifying the premonitory failure information from coal rock are key and difficult points of intelligent research on rock mechanics. In order to reveal the deterioration characteristics and the forewarning law of fractured coal rock, the digital image correlation method and the acoustic emission technology were adopted in this study to non-destructively detect the strain field, displacement field, and acoustic emission response in time and frequency domains. Additionally, by introducing the derivative functions of the multi-source information function for quantitative analysis, a comprehensive evaluation method was proposed based on the multi-source information fusion monitoring to forewarn red sandstone failure by levels during loading. The results show that obvious premonitory failure information, such as strain concentration areas, appears on red sandstone’s surface before macro-cracks can be observed. With an increase in the inclination angle of the prefabricated crack, the macroscopic failure mode gradually transforms from tensile splitting failure to tensile-shear mixed failure. Moreover, the dominant frequency signals of high frequency–low amplitude (HF–LA), intermediate frequency–low amplitude (IF–LA) and low frequency–low amplitude (LF–LA) are denser near the stress peak. The initial crack expansion time and failure limit time measured by multi-source information fusion are 20.72% and 26.71% earlier, respectively, than those measured by direct observation, suggesting that the forewarning of red sandstone failure by levels is realized with multi-source information fusion. Full article

To address the urgent demand for autonomous rapid initial alignment of vehicular inertial navigation systems in complex battlefield environments, this study overcomes the technical limitations of traditional stationary base alignment methods by proposing a robust moving-base autonomous alignment approach based on multi-source information fusion. First, a federal Kalman filter-based multi-sensor fusion architecture is established to effectively integrate odometer, laser Doppler velocimeter, and SINS data, resolving the challenge of autonomous navigation parameter calculation under GNSS-denied conditions. Second, a dual-mode fault diagnosis and isolation mechanism is developed to enable rapid identification of sensor failures and system reconfiguration. Finally, an environmentally adaptive dynamic alignment strategy is proposed, which intelligently selects optimal alignment modes by real-time evaluation of motion characteristics and environmental disturbances, significantly enhancing system adaptability in complex operational scenarios. The experimental results show that the method proposed in this paper can effectively improve the accuracy of vehicle-mounted alignment in motion, achieve accurate identification, effective isolation, and reconstruction of random incidental faults, and improve the adaptability and robustness of the system. This research provides an innovative solution for the rapid deployment of special-purpose vehicles in GNSS-denied environments, while its fault-tolerant mechanisms and adaptive strategies offer critical insights for engineering applications of next-generation intelligent navigation systems. Full article

Surface soil moisture (SSM) plays a pivotal role various fields, including agriculture, hydrology, water environment, and meteorology. To investigate the impact of land use types and fractional vegetation cover (FVC) on the accuracy of SSM estimation, this study conducted a comprehensive analysis of SSM estimation performance across diverse land use scenarios (e.g., multiple land use combinations and cropland) and varying FVC conditions. Sentinel-2 NDVI and MOD09A1 NDVI were fused by the Enhanced Spatial and Temporal Adaptive Reflection Fusion Model (ESTARFM) to obtain NDVI with a temporal resolution better than 8 d and a spatial resolution of 20 m, which improved the matching degree between NDVI and the Sentinel-1 backscattering coefficient (σ⁰). Based on the σ⁰, NDVI, and in situ SSM, combined with the water cloud model (WCM), the SSM estimation model is established, and the model of each land use and FVC is validated. The model has been applied in Handan. The results are as follows: (1) Compared with vertical–horizontal (VH) polarization, vertical–vertical (VV) polarization is more sensitive to soil backscattering (${\mathsf{\sigma }}_{\mathrm{soil}}^0$). In the model for multiple land use combinations (Multiple-Model) and the model for the cropland (Cropland-Model), the R² increases by 0.084 and 0.041, respectively. (2) The estimation accuracy of SSM for the Multiple-Model and Cropland-Model is satisfactory (Multiple-Model, RMSE = 0.024 cm³/cm³, MAE = 0.019 cm³/cm³, R² = 0.891; Cropland-Model, RMSE = 0.023 cm³/cm³, MAE = 0.018 cm³/cm³, R² = 0.886). (3) When the FVC > 0.75, the accuracy of SSM by the WCM is low. It is suggested the model should be applied to the cropland where the FVC ≤ 0.75. This study clarified the applicability of SSM estimation by microwave remote sensing (RS) in different land uses and FVCs, which can provide scientific reference for regional agricultural irrigation and agricultural water resources management. The findings highlight that the VV polarization-based model significantly improves SSM estimation accuracy, particularly in croplands with FVC ≤ 0.75, offering a reliable tool for optimizing irrigation schedules and enhancing water use efficiency in agriculture. These results can aid in better water resource management, especially in regions with limited water availability, by providing precise soil moisture data for informed decision-making. Full article

Deep learning-based semantic segmentation algorithms have proven effective in landslide detection. For the past decade, convolutional neural networks (CNNs) have been the prevailing approach for semantic segmentation. Nevertheless, the intrinsic limitations of convolutional operations hinder the acquisition of global contextual information. Recently, Transformers have garnered attention for their exceptional global modeling capabilities. This study proposes a dual-branch semantic aggregation network (DBSANet) by integrating ResNet and a Swin Transformer. A Feature Fusion Module (FFM) is designed to effectively integrate semantic information extracted from the ResNet and Swin Transformer branches. Considering the significant semantic gap between the encoder and decoder, a Spatial Gate Attention Module (SGAM) is used to suppress the noise from the decoder feature maps during decoding and guides the encoder feature maps based on its output, thereby reducing the semantic gap during the fusion of low-level and high-level semantic information. The DBSANet model demonstrated superior performance compared to existing models such as UNet, Deeplabv3+, ResUNet, SwinUNet, TransUNet, TransFuse, and UNetFormer on the Bijie and Luding datasets, achieving IoU values of 77.12% and 75.23%, respectively, with average improvements of 4.91% and 2.96%. This study introduces a novel perspective for landslide detection based on remote sensing images, focusing on how to effectively integrate the strengths of CNNs and Transformers for their application in landslide detection. Furthermore, it offers technical support for the application of hybrid models in landslide detection. Full article

The detection of mirrors and glass, which possess unique optical surface properties, has garnered significant attention in recent years. Due to their reflective and transparent nature, these surfaces are often difficult to distinguish from their surrounding environments, posing substantial challenges even for advanced deep learning models tasked with performing such detection. Current research primarily relies on complex network models that learn and fuse different modalities of images, such as RGB, depth, and thermal, to achieve mirror and glass detection. However, these approaches often overlook the inherent limitations in the raw data caused by sensor deficiencies when facing mirrors and glass surfaces. To address this issue, we applied mathematical and physical methods, such as three-point plane determination and steady-state heat conduction in two-dimensional planes, along with an RGB enhancement module, to reconstruct RGB, depth, and thermal data for mirrors and glass in two publicly available datasets: an RGB-D mirror detection dataset and an RGB-T glass detection dataset. Additionally, we synthesized four enhanced and ideal datasets. Furthermore, we propose a double weight Mamba fusion network (DWMFNet) that strengthens the model’s global perception of image information by extracting low-level clue weights and high-level contextual weights from the input data using the prior fusion feature extraction module (PFFE) and the deep fusion feature guidance module (DFFG). This is complemented by the Mamba module, which efficiently captures long-range dependencies, facilitating information complementarity between multi-modal features. Extensive experiments demonstrate that our data enhancement method significantly improves the model’s capability in detecting mirrors and glass surfaces. Full article

Point cloud registration is a fundamental problem in computer vision and 3D computing, aiming to align point cloud data from different sensors or viewpoints into a unified coordinate system. In recent years, the rapid development of RGB-D sensor technology has greatly facilitated the acquisition of RGB-D data. In previous unsupervised point cloud registration methods based on RGB-D data, there has often been an overemphasis on matching local features, while the potential value of global information has been overlooked, thus limiting the improvement in registration performance. To address this issue, this paper proposes a self-attention-based global information attention module, which learns the global context of fused RGB-D features and effectively integrates global information into each individual feature. Furthermore, this paper introduces alternating self-attention and cross-attention layers, enabling the final feature fusion to achieve a broader global receptive field, thereby facilitating more precise matching relationships. We conduct extensive experiments on the ScanNet and 3DMatch datasets, and the results show that, compared to the previous state-of-the-art methods, our approach reduces the average rotation error by 26.9% and 32% on the ScanNet and 3DMatch datasets, respectively. Our method also achieves state-of-the-art performance on other key metrics. Full article

Currently, there is no publicly available dataset for the classification of potato pest and disease-related queries. Moreover, traditional query classification models generally adopt a single maximum-pooling strategy when performing down-sampling operations. This mechanism only extracts the extreme value responses within the local receptive field, which leads to the degradation of fine-grained feature representation and significantly amplifies text noise. To address these issues, a dataset construction method based on prompt engineering is proposed, along with a question classification method utilizing a gated fusion–convolutional neural network (GF-CNN). By interacting with large language models, prompt words are used to generate potato disease and pest question templates and efficiently construct the Potato Pest and Disease Question Classification Dataset (PDPQCD) by batch importing named entities. The GF-CNN combines outputs from convolutional kernels of varying sizes, and after processing with max-pooling and average-pooling, a gating mechanism is employed to regulate the flow of information, thereby optimizing the text feature extraction process. Experiments using GF-CNN on the PDPQCD, Subj, and THUCNews datasets show F1 scores of 100.00%, 96.70%, and 93.55%, respectively, outperforming other models. The prompt engineering-based method provides a new paradigm for constructing question classification datasets, and the GF-CNN can also be extended for application in other domains. Full article

Cervical cancer screening through computer-aided diagnosis often faces challenges like inaccurate segmentation and incomplete boundary detection in colposcopic images. This study proposes a lightweight segmentation model to improve accuracy and computational efficiency. The architecture integrates dual encoder backbones (ResNet50 and MobileNetV2) for high-level and efficient feature extraction. While a lightweight atrous spatial pyramid pooling (ASPP) module records multi-scale contextual information, a novel attention module improves feature details by concentrating on relevant locations. The decoder employs advanced upsampling and feature fusion for refined segmentation boundaries. The experimental results show exceptional performance: training accuracy of 97.56%, validation accuracy of 96.04%, 97.00% specificity, 96.78% sensitivity, 98.71% Dice coefficient, and 97.56% IoU, outperforming the existing methods. In collaboration with the MNJ Institute of Oncology Regional Center, Hyderabad, this work demonstrates potential for real-world clinical applications, delivering precise and reliable colposcopic image segmentation. This research advances efficient, accurate tools for cervical cancer diagnosis, improving diagnostic workflows and patient outcomes. Full article

For fine-grained recognition, capturing distinguishable features and effectively utilizing local information play a key role, since the objects of recognition exhibit subtle differences in different subcategories. Finding subtle differences between subclasses is not straightforward. To address this problem, we propose a weakly supervised fine-grained classification network model with Local Diversity Guidance (LDGNet). We designed a Multi-Attention Semantic Fusion Module (MASF) to build multi-layer attention maps and channel–spatial interaction, which can effectively enhance the semantic representation of the attention maps. We also introduce a random selection strategy (RSS) that forces the network to learn more comprehensive and detailed information and more local features from the attention map by designing three feature extraction operations. Finally, both the attention map obtained by RSS and the feature map are employed for prediction through a fully connected layer. At the same time, a dataset of ancient towers is established, and our method is applied to ancient building recognition for practical applications of fine-grained image classification tasks in natural scenes. Extensive experiments conducted on four fine-grained datasets and explainable visualization demonstrate that the LDGNet can effectively enhance discriminative region localization and detailed feature acquisition for fine-grained objects, achieving competitive performance over other state-of-the-art algorithms. Full article

Time series remote sensing images (TSIs) offer essential data for time series remote sensing image change detection with remote sensing technology advances. However, most existing methods focus on bi-temporal images, lacking the exploration of temporal information between images. This presents a significant challenge in effectively utilizing the rich spatio-temporal and object information inherent to TSIs. In this work, we propose a History-Integrated Spatial–Temporal Information Extraction Network (HiSTENet), which comprehensively utilize the spatio-temporal information of TSIs to achieve change detection of continuous image pairs. A Spatial-Temporal Relationship Extraction Module is utilized to model the spatio-temporal relationship. Simultaneously, a Historical Integration Module is introduced to fuse the objects’ characteristics across historical temporal images, while leveraging the features of historical images. Furthermore, the Feature Alignment Fusion Module mitigates pseudo changes by computing feature offsets and aligning images in the feature space. Experiments on SpaceNet7 and DynamicEarthNet demonstrate that HiSTENet outperforms other representative methods, achieving a better balance between precision and recall. Full article

Bearing fault diagnosis is critical for the reliability and safety of rotating machinery in industrial applications. Traditional fault diagnosis methods and single-view deep learning models often fail to capture the complex, multi-dimensional nature of vibration signals, limiting their effectiveness in accurately identifying faults under varying conditions. To address this, we propose a novel multi-view framework that leverages complementary views—time, frequency, and wavelet domains—of vibration signals for robust fault diagnosis. Our framework integrates a cross-attention mechanism (CAM) that combines self-attention and cross-attention to capture both intra-view and inter-view dependencies, enabling the effective fusion of multi-domain information. By modeling the interactions between different views, the proposed approach enhances feature representation, leading to improved diagnostic accuracy even under noisy industrial conditions. Experimental results on public bearing datasets demonstrate the superior performance of our method compared to state-of-the-art approaches, with significant improvements in robustness and accuracy. This framework provides a promising solution for intelligent fault diagnosis in complex industrial environments. Full article

The spectral information extracted from hyperspectral images is characterized by redundancy and complexity, while the spectral morphological features extracted from the spectral information help to simplify the data and provide rich information about the material composition. This study is based on using spectral morphological features to quantitatively detect the water content of winter jujubes, and it extends the research scope to the composite effect of spectral morphological features on the basis of previous research. Firstly, a multiple linear regression analysis was carried out on different characteristic bands. Secondly, the multiple regression terms with high significance levels were used as the characteristic variables to be fused with the extracted characteristic wavelength variables for the data fusion. Finally, a partial least squares model was established for the water content of the winter jujubes. The results of the study show that a quantitative relationship can be established between the spectral morphology characteristics and the water content of winter jujubes. The coefficients of determination of the regression equations under the characteristic bands with center wavelengths of 1024 nm, 1146 nm, 1348 nm, and 1405 nm were 0.8449, 0.7944, 0.7479, and 0.9477, respectively. After fusing the spectral morphological features, the partial least squares modeling effects were all improved. The optimal model was the fusion model at a center wavelength of 1146 nm with a correlation coefficient of 0.9942 for the calibration set and 0.8698 for the prediction set. The overall results showed that the wave valley is more reflective of the fruit quality, and the morphological characteristics of the wave valley are more suitable than those of the wave peak for the quantitative detection of the moisture content of winter jujubes. Full article

In response to the issues of high complexity and low efficiency associated with the current reliance on manual sampling and instrumental measurement for obtaining maize leaf area index (LAI), this study constructed a maize image dataset comprising 624 images from three growth stages of summer maize in the Henan region, namely the jointing stage, small trumpet stage, and large trumpet stage. Furthermore, a maize LAI estimation model named LAINet, based on an improved convolutional neural network (CNN), was proposed. LAI estimation was carried out at these three key growth stages. In this study, the output structure was improved based on the ResNet architecture to adapt to regression tasks. The Triplet module was introduced to achieve feature fusion and self-attention mechanisms, thereby enhancing the accuracy of maize LAI estimation. The model structure was adjusted to enable the integration of growth-stage information, and the loss function was improved to accelerate the convergence speed of the network model. The model was validated on the self-constructed dataset. The results showed that the incorporation of attention mechanisms, integration of growth-stage information, and improvement of the loss function increased the model’s R² by 0.04, 0.15, and 0.05, respectively. Among these, the integration of growth-stage information led to the greatest improvement, with the R² increasing directly from 0.54 to 0.69. The improved model, LAINet, achieved an R² of 0.81, which indicates that it can effectively estimate the LAI of maize. This model can provide information technology support for the phenotypic monitoring of field crops. Full article