Search Results (1,188)

3D reconstruction and classification of aircraft are two active research areas in optical remote sensing image processing which are of great significance for applications such as airport monitoring and intelligence analysis. The traditional approaches usually focus only on one of these two tasks, and all these methods suffer from inherent limitations. In the field of 3D reconstruction, most current methods require multiple-view images as input, which is rarely feasible in remote sensing. However, single-view 3D reconstruction is an inherently ill-posed problem. Existing methods, including voxel generation and mesh template deformation, still suffer from limited accuracy and poor shape fidelity. In the field of image classification, the existing methods are mainly based on deep learning. These methods require a large amount of labeled data, and they may also be misled by the color and texture features of the target in the dataset. In this paper, we propose a unified framework for simultaneous 3D reconstruction and classification, specifically tailored for aircraft targets in optical remote sensing imagery. The key innovations are threefold: First, we introduce the Signed Distance Field (SDF) implicit representation to build a prior-guided 3D reconstruction framework pre-trained on 3D model datasets. Second, to achieve the reconstruction process with a single image as input, we design a new joint optimization pipeline. We propose a novel dual-kernel differentiable rendering method, which is fused behind the SDF generation network for iterative optimization of the implicit code and pose parameters. Third, a gated feature fusion module is developed to combine the optimal latent vector from reconstruction with the classification backbone. This integration enables the joint output of 3D meshes and category labels within a unified loop. The resulting optimal latent code plays a dual role as a generative seed for high-fidelity 3D reconstruction and as a low-dimensional feature representation for target classification. Quantitative evaluations validate the superiority of our joint framework. Compared with the strong mesh-based competitor AtlasNet, the proposed method yields a 12.2% boost in mean F-score. In object classification, leveraging the 3D implicit geometric features boosts the performance to a peak accuracy of 97.88%, outperforming advanced remote sensing backbones such as RSMamba and EAM by 2.03% and 2.54%. Additionally, ablation studies confirm the indispensability of our key designs, revealing that our dual-task feature fusion strategy brings an absolute gain of 1.18% in classification accuracy, while omitting the clustering prior stages and the dual-kernel rendering method leads to a 30.4% and 10.1% degradation in Chamfer distance. Full article

In recent years, Neural Radiance Fields (NeRFs) and 3D Gaussian Splatting (3DGS) have emerged as promising new approaches for 3D reconstruction. NeRFs rely on neural fields that generate a three-dimensional representation of a scene from photographs, estimating reflectance properties and reconstructing the underlying geometry. Since their introduction in 2020, NeRFs have attracted significant attention due to their wide range of potential applications. Conversely, 3D Gaussian Splatting (3DGS), introduced in 2023, employs Gaussian primitives to efficiently model objects and structures, offering a flexible and adaptive representation of 3D scenes. Starting from an established pipeline for the use of reality-based models for structural analysis, this paper investigates the performance of 3DGS in handling complex geometries and surfaces characterised by challenging acquisition conditions. Full article

Cloud removal is an indispensable preprocessing step in optical remote sensing. Reconstructing cloud-free imagery by combining multi-temporal optical observations with cloud-penetrating synthetic aperture radar (SAR) has become a mainstream approach. However, the existing studies mostly adopt simple composites, such as per-pixel least-cloudy selection or the temporal median, as baselines, and average accuracy metrics over entire scenes; together, these two practices may overstate the true gains of deep-learning methods. This paper proposes a temporal cross-modal cloud removal method (TCM-CR). In a multi-temporal sequence, the acquisition with the lowest cloud fraction retains true surface reflectance at its cloud-free pixels and is itself a high-accuracy baseline. TCM-CR exploits this baseline in two ways. First, on clear and light inputs, cloud-free pixels are taken unchanged from the reference image, so the true reflectance is preserved without loss, independent of training. Second, only cloud-covered pixels receive a bounded correction, in which SAR supplies the surface structure beneath clouds and multi-temporal observations are integrated along time while suppressing heavily clouded acquisitions. Experiments on the SEN12MS-CR-TS dataset show that TCM-CR maintains accuracy on par with the reference image on clear and light samples and improves the peak signal-to-noise ratio on heavy samples by 7.93 dB. In a cross-region experiment where one region is excluded from training entirely and used only for testing, heavy samples still improve by 7.27 dB. Full article

Traditional carvings, such as stone and wooden carvings, are important material carriers of intangible cultural heritage craftsmanship. High-quality three-dimensional (3D) digital replicas of these carvings provide essential support for their preservation, inheritance, interpretation, and dissemination. Owing to their intricate geometries, fine surface details, diverse materials, and complex acquisition backgrounds, high-fidelity 3D reconstruction of traditional carvings remains a challenging issue. In this study, we propose an object-centric 3D Gaussian Splatting (3DGS) framework for traditional craft carving reconstruction. Built upon the baseline 3DGS model, the proposed framework leverages the advanced segmentation capability of Segment Anything Model 2 (SAM-2) to extract foreground masks and generate alpha-channel inputs, enabling the reconstruction process to focus on the target carving. In addition, depth priors are used to guide local densification in regions with insufficient Gaussian coverage, providing auxiliary support for weakly textured or locally blurred carving details. Experiments were conducted on a self-built image dataset of stone and wooden carvings collected from Hui’an County, Quanzhou, Fujian Province, China. The experimental results show that the proposed object-centric strategy effectively preserves the original visual textures and local geometric features of traditional carvings while improving rendering efficiency. Furthermore, the optimized 3D Gaussian models are exported as lightweight digital assets and integrated into Unreal Engine 5, enabling multi-perspective visualization and interactive virtual exhibition in a contextualized digital environment. These results suggest that the proposed workflow is more suitable for producing compact object-level Gaussian assets for carving exhibition, while the depth-guided module mainly improves local details in weakly textured or shallow-relief regions. Full article

The flatness of bearing pads directly affects structural load transfer safety. However, conventional total station-based inspection methods suffer from limited spatial sampling, low inspection efficiency, and high safety risks associated with working at height. To address these limitations, this paper proposes UAV-FIBP (Unmanned Aerial Vehicle-based Flatness Inspection for Bridge Pads), an automated and intelligent method for pad flatness assessment utilizing UAV-based 3D reconstruction. By designing a close-range, multi-orbit circumnavigational UAV flight path and acquiring high-overlap imagery (85% forward and 80% side overlap), a millimeter-accurate 3D model is generated via photogrammetry, achieving a high-density point cloud of ≥200 points/cm² on the pad surface. Following point cloud denoising and region-of-interest segmentation using the Random Sample Consensus (RANSAC) algorithm, Principal Component Analysis (PCA) is employed to fit a reference plane. A dual-parameter evaluation framework is subsequently introduced: the root mean square (RMS) deviation quantifies local surface roughness, while the maximum elevation difference is derived from the angle between the normal vectors of the fitted plane and the horizontal plane, thereby enabling a comprehensive assessment of global inclination. Validation experiments conducted on laboratory-scale setups and real construction sites (involving four bridge pads) demonstrate that the proposed method achieves deviations ≤ 2 mm compared to total station measurements, satisfying the requirements stipulated in the Standards for Quality Inspection and Verification of Highways (JTG F80/1-2017). Results indicate that UAV-FIBP enables non-contact, full-coverage, and automated flatness inspection, significantly improving inspection efficiency and construction safety. This work establishes a scalable technical pathway for intelligent bridge construction. Full article

Portable digital microscopy (PDM) has become standard equipment in archaeological and restoration campaigns. In recent years, several studies have proposed combining PDM and micro-photogrammetry for preliminary documentation and in situ three-dimensional (3D) digitization, testing its applicability on a variety of materials and artifacts, including stone, bone, mural paintings, and small jewelry. However, limited data is available on its use with some of the most common types of artworks, such as paintings on canvas and wooden panels. In this study, the performance of micro-photogrammetry using a Dino-Lite digital microscope (Dino-Lite Europe, Almere, The Netherlands) was evaluated on five representative artistic materials: metal, stone, and paintings, including mural, canvas, and panel works. Textured mesh models were generated from the acquired datasets, and the quality of the 3D reconstruction was assessed through analysis of dense cloud confidence and Digital Elevation Model (DEM) accuracy. The metrical reliability of the micro-photogrammetric results was validated by comparing the 3D data obtained with the digital microscope against height measurements acquired using an optical microprofilometer. The results indicate that the performance of the method is strongly influenced by the physical properties of the substrates examined. Accurate and metrically consistent reconstructions were achieved for matte and textured surfaces (e.g., fresco and stone), whereas significant limitations emerged on dark or glossy painted surfaces, such as canvas and panel paintings. Full article

Spatiotemporal change detection of surface cracks in concrete structures is of great importance for evaluating and maintaining their structural health. The development of robotics and 3D computer vision technologies provides new solutions for key subtasks in this process, including automated data acquisition, spatial localization and quantification of cracks, and multi-temporal crack registration. This study proposes an automated UAV- and point cloud-based framework for detecting spatiotemporal changes in cracks in concrete structures. First, the proposed autonomous UAV path-planning algorithm is used to achieve data acquisition that conforms to complex structural geometries. Then, an improved SfM algorithm is employed to realize spatial crack localization and local point cloud densification. Finally, accurate registration of crack point clouds from different periods is achieved based on a two-step registration strategy. Experimental results on a real large-scale concrete structure show that the proposed path-planning algorithm can achieve complete envelope coverage conforming to the structural geometry, with an effective coverage ratio above 99.7%. The dimensional error of structural reconstruction is controlled within 20 mm. The average crack localization time is 4.00 s, and mean absolute error of crack width quantification is 0.44 mm. The average crack registration error is 0.97 mm, thereby enabling accurate tracking of crack evolution. Full article

Isolated drowned carbonate platforms in the Xisha Area represent critical research archives for deciphering the evolutionary mechanisms of carbonate systems and paleoenvironmental changes within the marginal sea of the northern South China Sea. Nevertheless, existing academic research lacks comprehensive investigations into the sedimentary processes and episodic demise signatures of typical drowned platforms in this region. This study targets the Ganquan Isolated Drowned Carbonate Platform—the largest and best-preserved drowned carbonate platform in the Xisha area. Based on high-resolution 2D multi-channel seismic data and multibeam bathymetric data, this research systematically characterizes the stratigraphic architecture and sedimentary differentiation of the Ganquan Carbonate Platform, and reconstructs its complete Cenozoic evolutionary history via interval velocity inversion, seismic sequence stratigraphic interpretation, sedimentary facies identification, and geomorphological feature analysis. The results reveal that the carbonate succession of the Ganquan Carbonate Platform exhibits prominent vertically zoned interval velocity variations, with two sharp velocity discontinuities identified at the seabed surface and the contact boundary between carbonate rocks and metamorphic basement. Four stratigraphic units are distinguished from top to bottom: seawater column, loose Upper Miocene carbonate deposits, compact Lower–Middle Miocene carbonate strata, and metamorphic basement. Five key Cenozoic seismic boundaries are recognized within the study area, which divide the evolutionary history of the Ganquan Carbonate Platform into four successive episodes: incipient initiation during the Early Miocene, vigorous expansion in the Early–Middle Miocene, transitional decline in the late Middle Miocene, and drowning reworking since the Late Miocene. This study establishes a comprehensive evolutionary model for isolated drowned carbonate platforms in the Xisha region, and advances the theoretical framework governing carbonate platform evolution in marginal seas. Full article

Mars exploration is an advanced field of global deep space exploration. Accurate three-dimensional reconstruction of the Martian surface topography is very important for autonomous navigation, scientific target recognition, and operation planning. In order to meet the analysis requirements of the Martian surface scene, this paper proposes an explicit surface-geometry-constrained Gaussian splatting (ESGS) method. Firstly, this method includes a normal and depth prior estimation network (NDN) that generates normal and depth priors from Martian surface image data, thereby promoting the fusion of semantic and multi-view contextual information to enhance the geometric accuracy of 3D reconstruction of the Martian surface. Secondly, we designed the Gaussian parameter-based deformable fusion network (GPDFN) to fuse multi-receptive-field feature information. Finally, we collected Martian surface remote sensing images from NASA, constructed a Martian surface 3D reconstruction dataset named Mars_3D using the COLMAP method, annotated depth and normal labels for its seven real-world scenes and two Blender-generated scenes, and conducted comparative experiments with eight excellent algorithms on this dataset to validate the effectiveness of our method in 3D reconstruction of the Martian surface using remote sensing images. Experiments show that the average SSIM of the ESGS method in this article is 0.6946, PSNR is 23.40 dB, and LPIPS is 0.253 on the Mars_3D dataset, demonstrating superior overall performance compared to all other models and enhancing the quality of 3D reconstruction of the Martian surface. Full article

Developing digital twins for legacy hydropower units is difficult when turbine documentation, calibrated performance data, and integrated measurements are incomplete. This study presents a Computational Fluid Dynamics (CFD)-assisted reconstruction workflow for a Kaplan turbine at the Wały Śląskie Hydropower Plant and evaluates its use as a physics-informed foundation for a digital twin. The flow passage was reconstructed from archival documentation, direct measurements, and optical 3D scanning of the runner. A steady-state Reynolds-averaged Navier–Stokes model was then prepared in OpenFOAM v2506 for selected head levels, guide-vane openings, and runner-blade angles. The simulations determined hydraulic performance, flow-field structures, and combinatory characteristics of the double-regulated turbine. The computed hydraulic efficiency reached approximately 85% in the nominal-head range, and the highest-efficiency region formed a broad plateau rather than a sharp optimum. CFD-derived and measurement-derived combinatory trends were consistent, although absolute values remain limited by relative field measurements and uncalibrated Winter–Kennedy flow estimation, a differential-pressure-based method. The CFD results were reduced to compact response surfaces and integrated with reconstructed geometry into an advisory digital twin for operating-point assessment, visualization, documentation, and training. This study establishes a robust workflow for this specific Kaplan turbine case where reverse engineering, integrated with CFD analysis, generates high-fidelity surrogate models for hydropower digital twins, effectively addressing the challenge of incomplete legacy documentation. Full article

Reliable three-dimensional (3D) reconstruction from endoscopic video is essential for endoscopic digital twins, scene review, and minimally invasive visual analysis. However, endoscopic images are not clean observations of intrinsic tissue appearance. Depth-dependent blur, shallow mucosal color diffusion, wet-surface specular reflection, and frame-wise color variation are often coupled with the captured signal. When such observation-dependent effects are directly optimized as Gaussian colors, conventional 3D Gaussian Splatting may encode transient imaging artifacts as persistent tissue appearance, leading to blurred textures, color drift, specular residues, and unstable novel-view synthesis. This paper presents EndoDGS (Endoscopic Degradation-Decoupled Gaussian Splatting), a degradation-decoupled Gaussian Splatting framework for endoscopic novel-view reconstruction. The core idea is to keep stable geometry and base tissue appearance in the Gaussian representation, while modeling endoscope-induced degradations separately in a bounded render-space compensation pipeline. EndoDGS combines lightweight appearance modulation for frame-wise color stabilization with sequential degradation compensation for optical blur, mucosal color transport, and wet-surface specular response. This design reduces the entanglement between persistent tissue appearance and transient imaging degradations without changing the underlying Gaussian geometry and visibility ordering. Experiments on synthetic colonoscopy and real endoscopic/laparoscopic datasets covering 38 scenes show that EndoDGS consistently improves reconstruction quality over representative implicit and explicit reconstruction baselines. The results demonstrate that separating stable tissue representation from observation-dependent endoscopic degradations provides a more faithful, stable, and interpretable foundation for endoscopic 3D reconstruction. Full article

Visible light communication (VLC) can reduce radio-frequency (RF) congestion in indoor industrial monitoring, but a short transmitted payload does not by itself prove that visual or task-relevant information has been preserved. This study therefore frames the proposed method as an autoencoder-based compact latent-payload transmission scheme and explicitly distinguishes it from channel-aware joint source-channel coding (JSCC). Raw red–green–blue (RGB), lossless Huffman, and autoencoder latent payloads are first compared under the same VLC model using bit error rate (BER), calculated/model-derived VLC transmission energy, reconstruction quality, and task utility. The 128-component latent representation contains 4096 bits, corresponding to a 294-fold payload-size reduction relative to an uncompressed $224\times 224$, 24-bit RGB image; this ratio is used only as a raw-payload reference and not as a general codec-compression claim. An independent industrial-domain audit is conducted on 30 Northeastern University (NEU) steel-surface images using four-fold out-of-fold evaluation, peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and a fixed defect-class proxy. An NEU architecture-and-resolution comparison shows that changing from Compact-NEU to the full $224\times 224$ model increases PSNR from 14.12 dB to 15.61 dB at the same 4096-bit bottleneck, while a regularized full model gives the highest SSIM of 0.471. Because input resolution and network capacity both change in this comparison, the result is interpreted as evidence that the architecture/resolution setting contributes to the industrial-domain gap, not as a strict isolation of capacity alone. Finally, an end-to-end JSCC-VLC baseline with 4096 nonnegative optical channel uses is trained through a differentiable intensity channel. It obtains 23.82 dB/0.568 SSIM in the clean case and 22.18 dB/0.519 SSIM at 5 dB SNR, showing more channel-aware behavior and more graceful degradation than the separated serialized-latent pipeline. Overall, the results support the modeled energy and active-time benefits of compact latent payloads while showing that robust industrial visual transmission requires architecture/resolution controls, practical codec baselines, and channel-aware JSCC comparisons. Full article

High-precision 3D reconstruction of objects with complex surfaces, such as ancient architecture and detailed artworks, requires close-range image acquisition, which remains challenging for Unmanned Aerial Vehicle (UAV) systems. The operational proximity of current UAV workflows is often insufficient to capture fine geometric and textural details, limiting high-fidelity digitization. This paper presents a georeferenced NeRF-based UAV acquisition framework for automated waypoint planning and supervised close-proximity execution. The core of the framework is a path-planning module that operates on a metric geometric prior established through Geographic Neural Radiance Fields (Geo-NeRF), which denotes a georeferenced NeRF modeling pipeline rather than a new NeRF architecture or loss function. By generating waypoints directly on this neural representation and optimizing the flight path via a nearest-neighbor strategy, the proposed framework supports close-proximity image acquisition for static targets under controlled conditions. Empirical validation demonstrates improved close-range flight proximity, photographic accuracy, and 3D reconstruction fidelity compared with the evaluated baselines. Full article

Surface deformation monitoring is essential for preventing geological hazards, safeguarding major engineering projects, and extracting mineral resources. Due to its all-weather, continuous, wide-coverage observation capabilities and millimeter-level accuracy, InSAR technology has emerged as an indispensable tool for monitoring surface deformation in mining regions. This paper presents a comprehensive review of recent advancements in InSAR technology for monitoring surface subsidence and horizontal displacement. It evaluates both fundamental D-InSAR techniques and advanced time-series analysis frameworks, such as PS-InSAR, SBAS-InSAR, and DS-InSAR. The review details the methodologies for extracting LOS deformation, multi-track combined horizontal displacement retrieval, and 3D deformation reconstruction. Furthermore, it provides a comprehensive overview of the causal mechanisms, correction strategies, impacts on deformation retrieval results, and developmental trends regarding major error sources such as decorrelation noise, residual orbital errors, and atmospheric delay. Drawing on typical case studies of large-gradient subsidence and landslide monitoring, this paper evaluates the efficacy and limitations of time-series InSAR in complex mining environments. Our analysis indicates that the InSAR monitoring system for mining areas is transitioning from single-dimensional deformation detection to multi-dimensional deformation analysis and intelligent dynamic monitoring. However, significant challenges remain regarding phase decorrelation in rapid deformation areas, the complete reconstruction of 3D displacements, the suppression of atmospheric errors under complex meteorological conditions, and real-time processing capabilities. Concurrently, this paper identifies multi-source data fusion, artificial intelligence-assisted retrieval, and cloud-based real-time processing as crucial future directions. These advancements will provide robust technical support for safe production and the early warning of geological hazards in mining regions. Full article