Image Quality Assessment (IQA) is a valuable and vital research topic in image processing and computer vision, aiming to construct image quality assessment models in accordance with human visual perception. With the rapid advancement of Artificial Intelligence-Generated Content (AIGC), remarkable progress has been made in visual generation, particularly in image synthesis. However, due to the inherent limitations of image generation mechanisms, Artificial Intelligence-Generated Images (AIGI) often exhibit various quality defects, including semantic drift, unreasonable structures, and visual artifacts. Traditional IQA methods fail to adequately capture these unique defects, thus falling short in meeting the evaluation needs for AIGI. Consequently, AIGI Quality Assessment (AIGIQA) has emerged as a prominent research topic in computer vision. Existing technical frameworks and research methodologies of AIGIQA were comprehensively reviewed, systematically investigated, and thoroughly analyzed. First, three representative image generation paradigms, namely generative adversarial networks, autoregressive models, and diffusion models, were reviewed, and the typical image quality defects associated with different generation mechanisms were analyzed. Next, the quality evaluation dimensions of AIGI were summarized from five perspectives, including technical issues, AI artifacts, unnaturalness, discrepancy, and aesthetics. Furthermore, commonly used performance evaluation metrics for AIGIQA models were reviewed, and publicly available datasets were introduced and compared. In addition, the development of traditional IQA methods was revisited, and their limitations in assessing AIGI were thoroughly discussed. Existing AIGIQA approaches were further categorized, with particular emphasis on the technical frameworks, core ideas, and limitations of representative methods based on cross-modal alignment and Multimodal Large Language Models (MLLMs). Finally, the opportunities and challenges faced by AIGIQA in quality representation, dataset support, model construction, and interpretability were discussed, and potential directions for future research were identified, providing valuable references for subsequent studies in this field.

Dynamic Three-Dimensional (3D) human reconstruction represents a pivotal research direction at the intersection of computer vision and computer graphics, harboring substantial application potential in virtual reality and digital entertainment. Traditional reconstruction methodologies, primarily relying on explicit representations such as polygonal meshes, voxel grids, and point clouds, suffer from the limited representation capabilities inherent in underlying parametric models, making it difficult to capture complex non-rigid geometry, fast joint movements, and fine-grained texture details. Conversely, implicit reconstruction techniques, exemplified by Neural Radiance Fields (NeRF), are severely bottlenecked by prohibitive volume rendering computational overhead, which hinders real-time interaction and direct physical editing. In this context, 3D Gaussian Splatting (3DGS) has emerged as a revolutionary and highly promising paradigm. It leverages discrete, anisotropic 3D Gaussian primitives-characterized by attributes such as position, covariance, color, and opacity-, alongside an exceptionally efficient differentiable rasterization mechanism to establish a new frontier for high-fidelity, real-time dynamic human reconstruction. To elucidate the evolution of this technology, a universal baseline reconstruction pipeline was systematically abstracted to dissect the fundamental workflow into three core components: canonical space representation, pose-driven deformation fields, and tile-based differentiable rasterization. On this basis, mainstream improvements addressing three major research bottlenecks were analyzed and compared: First, regarding detail reconstruction enhancement, methods for enhancing the reconstruction of human details and complex garments; their integration of expressive parametric templates, multi-shell volumetric mappings, and semantic graph constraints to dramatically improve the geometric fidelity and texture details of high-degree-of-freedom areas like faces, hands, and multi-layered loose clothing. Second, in terms of computational resource management, methods for accelerating training and inference and reducing memory consumption; including hash grid encoding, keyframe-based initialization, adaptive Gaussian pruning, and neural attribute quantization-demonstrate collective efforts to reduce training and inference times to seconds and compress storage footprints to the sub-megabyte scale for mobile deployment. Third, concerning system robustness enhancement, meticulous analyses clarify how strategies employing generative diffusion priors, feedforward regression networks, material-geometry-lighting decoupling, and physical simulator bindings resolve complex challenges associated with severe self-occlusions, sparse inputs, models for improving robustness to occlusion, cross-identity generalization, relighting, and physical applications. Furthermore, the loss-function system, including reconstruction losses and geometric, deformation, and attribute regularization constraints, was comprehensively reviewed, and major datasets based on monocular video, multi-view video, and 3D scanning, together with key rendering evaluation metrics. Finally, promising future research avenues focus on the co-optimization of quality, efficiency, and robustness; multi-instance complex scene reconstruction; and advanced multimodal physical-semantic interaction driven by large foundation models. Ultimately, future directions were discussed to provide a systematic reference for subsequent studies.

Plankton are essential components of aquatic ecosystems, and variations in their community structure, abundance, and dominant species succession can reflect changes in water trophic status, ecological disturbances, and environmental quality. Therefore, plankton serve as important bioindicators in aquatic ecological monitoring, harmful algal bloom early warning, and water quality assessment. Traditional microscopic identification methods rely heavily on expert experience and are characterized by low efficiency, strong subjectivity, and limited capability for continuous monitoring, making it difficult to meet the it difficult real-time, fine-scale, and automated aquatic environmental management. As imaging technologies and image recognition algorithms advance, plankton monitoring has gradually shifted from manual sampling and microscopic identification toward automated imaging and intelligent recognition. Based on this development trend, this review systematically summarized the current progress of plankton imaging devices, publicly available datasets, and image recognition technologies from the perspectives of image acquisition, data support, methodological evolution, and application scenarios. The transition of recognition approaches from handcrafted feature extraction and traditional machine learning to deep learning-based methods was analyzed, along with recent applications in community monitoring and ecological response analysis, harmful algal bloom prediction, and water quality assessment. Overall, microscopic imaging, flow imaging, and in situ imaging technologies were shown to provide diverse approaches for plankton image acquisition, with in situ imaging being better suited to continuous monitoring in natural aquatic environments. Public datasets established through expert annotation or manual verification were found to provide reliable data foundations for model training and performance evaluation. Deep learning has become a major technical approach to plankton image recognition, demonstrating significant advantages in improving recognition accuracy and automation capability. However, challenges remain, including insufficient dataset standardization, limited model generalization ability, high field-deployment costs, and inadequate integration of multi-source information. In the future, emerging technologies such as self-supervised learning, Vision Transformer (ViT), and edge intelligence are expected to reduce dependence on manual annotation, enhance feature representation for complex images, and improve field deployment capabilities. Integrating in situ imaging monitoring, automated recognition methods, and water quality parameter analysis will further enhance the application potential of plankton image recognition technologies in aquatic environmental monitoring, early warning, and assessment.

With the widespread adoption of document images in critical fields such as government administration and finance, tampering with these images not only incurs economic losses but also undermines public trust, making information security essential. Meanwhile, rapid advancement in tampering techniques has escalated the challenge of authenticity verification, thereby fueling growing research interest in document image tampering detection. To enhance the localization accuracy of tampered regions in document images, a multi-feature fusion-based segmentation method for document image tampered regions was proposed, namely the Multi-Feature Fusion Network (MF-Net). Firstly, a multi-feature encoder was designed, comprising three parallel branches. The spatial branch extracted rich spatial-domain information, the frequency branch integrates multi-level wavelet transform and convolution operations to fully capture tampering-related features in the frequency domain; and the noise branch employed the Spatial Rich Model (SRM) filter bank to analyze noise characteristics, providing the model with an additional perspective. These three branches were used jointly enable comprehensive extraction of multi-view features. Secondly, a hybrid attention feature-fusion module was developed, which serially combined channel attention, spatial attention, and parameter-free attention. Through multi-level refinement involving dual-feature-guided channel weighting, spatial focusing, and parameter-free suppression, deep fusion of multi-source features was achieved. Finally, inspired by the human visual system’s mechanism of prioritizing attention to abnormal tampered regions, an attention module was incorporated into a multi-view iterative decoder. This allowed the decoder to adaptively focus on anomalous regions, thereby realizing precise localization of tampered areas. Experimental validation of the proposed method was conducted on the public TextTamper dataset. Compared with state-of-the-art methods, the proposed MF-Net achieved improvements of 2.8% and 5.2% in F1-score and IoU, respectively. The experimental results demonstrated that the proposed method effectively enhanced the overall performance of tampered region localization in document images.

The condition of coal mine power equipment condition the power supply safety of mine shafts. UAV-borne infrared thermal inspection has become an important monitoring method. However, existing lightweight detection algorithms lack satisfactory performance in identifying early tiny defects under the conditions of low infrared contrast, large target scale differences and complex thermal backgrounds, making it difficult to balance real-time performance and detection accuracy for edge devices. To address these problems, MobileNetV3 lightweight network was adopted as the backbone, and a multi-kernel dynamic convolution mechanism was embedded into inverted bottleneck residual units. Multiple convolution kernels ware fused through input-aware attention weights to adaptively adjust the convolution operators. Parameter-sensitivity experiments confirmed that setting the kernel number to 4 achieved the optimal balance between detection performance and computational cost. Meanwhile, the Efficient Channel Attention (ECA) module was introduced into shallow layers of the backbone network to highlight thermal edge details and suppress background thermal noise. The Squeeze-and-Excitation (SE) module was retained in deep layers to strengthen high-level semantic discrimination capacity. The Ghost module was integrated in the channel expansion stage to cheaply expand feature maps via linear transformation. The three modules formed a collaborative link of “denoising and retaining details in shallow layers, enhancing semantics in deep layers, and supplementing textures during channel expansion”, enriching feature representation with minimal extra computational overhead. Experiments were conducted on a self-built infrared image dataset of coal mine power equipment. Containing 5 000 images of five typical equipment types and their defects: disconnecting switches, insulator strings, transformer bushings, cable joints and surge arresters. Experimental results showed that the improved model achieved an accuracy of 88.5%, an mAP of 86.7% at IoU threshold 0.5 and an F1-score of 86.6%. The computational complexity of the model was around 3.18 GFLOPs, and the inference latency on Central Processing Unit (CPU) was approximately 39 ms. Ablation experiments confirmed the collaborative gain of dynamic convolution and hierarchical feature enhancement, with optimal performance obtained at the kernel number K=4. Compared with recent mainstream lightweight detection algorithms, including YOLOv8n, YOLOv9-Tiny, RT-DETR-R18 and YOLOv10-N, the proposed method achieves the optimal comprehensive balance between accuracy and efficiency. Visual detection results also demonstrated that the model maintained high detection confidence for all types of power equipment under low resolution, heavy noise and occlusion. The proposed method greatly improved infrared defect detection performance while meeting the real-time reasoning requirements for UAV edge terminals, demonstrating engineering practicality in complex underground coal mine scenarios.

To address the challenges of identity confusion and trajectory fragmentation caused by similar appearances and rapid motion in multi-football-player tracking scenarios, a Multi-Object Tracking (MOT) model based on an improved Deep-EIoU framework was proposed. Building on the original detection-matching pipeline of Deep-EIoU, the proposed method introduced a trajectory-level global optimization mechanism combining appearance consistency and motion coherence, and the initial association results at the tracklet level were further refined. First, a tracklet splitter module was used to identify and separates mixed-identity segments, eliminating the propagation of erroneous identities. Second, a tracklet connector module performed global fusion of fragmented trajectories under spatio-temporal constraints to recover long-term continuous tracking of targets, thereby improving identity preservation and trajectory continuity in complex dynamic scenarios. Experimental results showed that the improved Deep-EIoU achieved superior performance on multiple metrics, with HOTA, IDF1, AssA, MOTA, and DetA reaching 89.1%, 82.1%, 79.4%, 96.7%, and 99.1% on the SoccerNet-Tracking dataset, and 78.1%, 80.1%, 68.4%, 97.1%, and 88.6% on the SportsMOT dataset, respectively, outperforming mainstream multi-object tracking methods in both identity stability and continuity. The proposed model demonstrated high tracking accuracy in football and other complex sports scenarios, providing a technical foundation for intelligent sports analysis and tactical research.

Scene Text Image Super-Resolution (STISR) aims to recover clear, complete, and recognizable character structures from Low-Resolution (LR) text images, thereby providing high-quality visual inputs for scene text recognition, intelligent document processing, and visual retrieval in natural environments. Owing to factors such as imaging distance, focal variation, motion blur, and complex background interference, LR scene text images often contain broken strokes, blurred character boundaries, distorted contours, and missing structural details, reducing the accuracy and robustness of downstream text recognition models. Existing STISR methods mainly focus on visual texture reconstruction or text-related semantic guidance and remain inadequate for recovering character contours and fine-grained structures under severe degradation. Although diffusion models show strong detail generation capability, without explicit structural constraints, the denoising process may introduce local artifacts, structural shifts, or incomplete stroke recovery when processing highly degraded text images. To address these problems, a Latent Structure Prior-Prompted Diffusion Model (LSPDiff) was constructed for scene text image super-resolution. LSPDiff comprised a Latent Structure Generation Model (LSGM) and a Prior-Prompted Diffusion Model (PPDM). The LSGM took bicubic-interpolated low-resolution text images as input, learned multi-scale latent structural features through an encoder-decoder architecture, and used multiscale supervision to enhance character strokes, edge contours, local textures, and spatial arrangements at different resolutions. Multi-scale supervision is introduced to encourage the generated structural representations to approach the corresponding high-resolution text images. During reverse denoising, the PPDM jointly incorporated the low-resolution conditional image, noisy image, and structural prompts, and a prior-prompt attention module was designed to model the interaction between diffusion intermediate features and latent structural priors, enabling structural information to guide noise estimation and character reconstruction. Since the two models had different optimization objectives, LSGM focuses on structural prior learning while the PPDM focuses on diffusion-based noise prediction, their optimization objectives are not fully consistent. a two-stage training strategy was therefore adopted. The LSGM was first trained independently to obtain stable structural priors, after which its parameters were frozen, and the PPDM was optimized with fixed structural guidance to improve convergence stability and prior transfer effectiveness. Experiments were conducted on the TextZoom benchmark dataset, which contains 21 740 low-resolution and high-resolution scene text image pairs, including 17 367 pairs for training and 4 373 pairs for testing. The test set is divided into easy, medium, and hard subsets according to image degradation difficulty, allowing a detailed evaluation under different restoration conditions. Three scene text recognition models, namely ASTER, MORAN, and CRNN, were used to evaluate recognition accuracy on the restored images. Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) were also adopted to measure reconstruction quality. LSPDiff achieved average recognition accuracies of 66.2%, 63.3%, and 57.0% under ASTER, MORAN, and CRNN, respectively, as well as a PSNR of 22.32 dB and an SSIM of 0.792 2, demonstrating favorable performance in both recognition-oriented restoration and image reconstruction quality. Compared with representative STISR methods, LSPDiff obtains competitive overall results, especially in recognition accuracy and PSNR. Ablation experiments further verify the role of latent structural guidance. Compared with the PPDM-only variant, showed that introducing the LSGM increased PSNR from 21.40 dB to 22.32 dB and recognition accuracy from 60.5% to 66.2%. These results indicate that multi-scale latent structural priors provide effective guidance for recovering character strokes and maintaining text layout consistency. Multi-scale latent structural priors can effectively constrain the diffusion-based reconstruction process and improve stroke continuity, contour clarity, and text readability in low-resolution scene text images, and provide an effective approach to STISR under complex degradation conditions and has potential application value in scene text recognition, intelligent document processing, and visual information understanding.

To balance inference efficiency, animation control, and relighting realism in monocular 3D face reconstruction, a feed-forward inverse-rendering framework was proposed. 3D Gaussian Splatting (3DGS) was combined with the FLAME prior, and Physically Based Rendering (PBR) attributes were introduced into the Gaussian primitives. A shared image encoder jointly predicts 3DGS features and an HDR illumination map, unifying the inference of geometry, materials, and lighting from a single image. To alleviate material-lighting ambiguity, IC-Light and Sapiens are introduced for screen-space pseudo-supervision of albedo and normal. An animation-aware normal updating mechanism and temporal consistency constraints further stabilize training. On the VFHQ dataset, the method achieved 22.34 dB PSNR and 0.855 SSIM for self-reenactment, outperforming the LAM baseline by 0.93 dB PSNR while maintaining sub-10 ms per-frame latency. Comparisons under the VFHQ-Test protocol showed significant advantages in reconstruction fidelity, driving accuracy, and novel-illumination consistency. The reconstructed PBR-3DGS representation supported natural FLAME-driven animation while preserving rigorous specular highlights and smooth shading under novel lighting, effectively balancing quality, efficiency, and physical consistency.

In minimally invasive spinal surgery, a single imaging modality cannot simultaneously provide sufficient information on both osseous structures and soft tissues, making intraoperative spatial localization challenging. Conventional cross-modality registration methods often struggle to accommodate the intensity differences between CT and MRI, inconsistent scan coverage, imaging artifacts, and the distinct motion characteristics of multiple spinal tissues, which may lead to registration failure and vertebral localization errors. To address these issues, a multi-stage spinal CT-MRI registration method based on multi-vertebra structural modeling was proposed. In the coarse registration stage, anatomical constraints were constructed using vertebral segmentation results. Common vertebral labels were extracted from the two modalities, and a representative vertebra was selected as an anchor to initialize rigid and affine transformations, thereby reducing large-scale pose differences and preventing non-rigid registration from becoming trapped in local optima. In the fine registration stage, a deep neural network integrating multi-vertebra affine modeling and diffeomorphic deformation was developed. The multi-vertebra affine branch characterized the approximately rigid motion of different vertebral segments, while velocity-field integration was employed to generate a diffeomorphic deformation field for modeling the nonlinear deformation of intervertebral discs and surrounding soft tissues, thereby achieving high-precision non-rigid registration. Experiments were conducted on a self-built spinal CT-MRI dataset. The proposed method achieved Dice, HD₉₅, and ASD values of 91.43%, 2.61 mm, and 1.16 mm, respectively, in the lumbar region, outperforming all comparison methods. Quantitative results demonstrated that the proposed method improved the regional overlap and boundary alignment accuracy of lumbar vertebral structures. The visualization results further demonstrated that it could effectively coordinate the local near-rigid motion of vertebral segments with the continuous deformation of surrounding tissues, maintaining good stability and anatomical rationality. This provided a technical foundation for preoperative multimodal image fusion and subsequent surgical navigation.

With the increasing demand for user personalization, diversity objectives in recommender systems have received growing attention. Graph Neural Networks (GNNs) provide a new paradigm for alleviating recommendation homogeneity and improving result diversity by modeling high-order associations between users and items. However, existing GNN-based methods mainly rely on local neighbor aggregation, which tends to introduce redundant information and lacks the ability to distinguish heterogeneous interest structures. In contrast, Transformers have strong global modeling capabilities and can capture long-range dependencies, but their computational complexity is relatively high, and they have difficulty to effectively capturing users’ latent diverse interests. To address these problems, a diversity-aware recommendation model named SGMDRec (Synergistic Graph-Mamba for Diversified Recommendation) was proposed to jointly optimize the accuracy and diversity of recommendation results. The model was composed of two core modules: dynamic programming neighbor selection based on high-dimensional prefix sums and a GNN-Mamba alternating network. The former constructed an efficient diverse subset selection strategy based on high-dimensional prefix sums. Through low-dimensional mapping, symbolic encoding, and category sparsity weighting, diverse neighbor subsets with category coverage potential were selected from candidate neighbors, and a category-aware representation enhancement mechanism was incorporated to characterize fine-grained differences among items within the same category. The latter first employed GNN to perform message passing between nodes and aggregate local structural information in the user-item interaction graph. Then, to overcome the computational efficiency bottleneck of traditional Transformer, a Mamba module was introduced for lightweight global sequence modeling. In particular, to enhance the graph-structure awareness of Mamba, a graph-aware state transition mechanism was further designed. By integrating node degree information, the parameter configuration of the State-Space Model (SSM) was dynamically adjusted, enabling the model to capture both local collaborative relations and global interest evolution characteristics. Experimental results showed that SGMDRec could effectively improve the category coverage and diversity of recommendation results while maintaining recommendation accuracy, achieving a better balance between accuracy and diversity. The results verified the effectiveness and scalability of the proposed method in diversified recommendation tasks and provide an efficient graph-aware modeling scheme for balancing accuracy and diversity in recommender systems.

To address the limitations of current digital heritage conservation practices, including the high level of professional expertise required for manual drafting, low drawing efficiency, operator-dependent errors, and the risk of secondary damage to cultural relics during physical measurement, as well as the technical bottlenecks of line-drawing techniques in digital heritage conservation，a non-contact method for the automatic generation of archaeological line drawings was proposed, and an automated archaeological drawing system was developed based on the Grasshopper parametric platform. Using three-dimensional models reconstructed from preprocessed point-cloud data, the proposed method enabled the automated generation of archaeological line drawings through four core modules. First, the geometric correction module analyzed the covariance matrix of the point cloud and extracted its eigenvectors to construct a rotation matrix, thereby correcting deviations caused by scanning posture. An affine transformation was subsequently applied to normalize the spatial reference and reset the model origin. Second, the geometric segmentation module performed accurate model sectioning through the parametric construction of cutting planes and Boolean operations. Third, the projection generation module applied orthographic projection algorithms to produce half-sectional and plan views. Fourth, the dynamic annotation module established an adaptive mapping relationship between model dimensions and annotation font size, thereby enabling intelligent annotation. The system was primarily intended for artifacts with relatively well-defined morphological characteristics. With manual intervention during model preprocessing, it established an efficient archaeological line-drawing workflow centered on parametric processing and automated output. The proposed approach reduced visual errors caused by manual operation, avoided secondary damage to cultural relics, and substantially improved drawing efficiency and adherence to academic standards. It therefore provided a theoretically rigorous and practically applicable technical paradigm for the digital conservation of cultural heritage. Future work will further develop the texture-mapping module by incorporating high-resolution photographs and texture images of cultural relics to enhance the realism and detail representation of three-dimensional models, thereby extending the application value of the system in digital heritage documentation, presentation, and conservation.

To address the challenges of blurry rendering in low-texture regions, pervasive floating artifacts, and large model storage in novel view synthesis of complex scenes using 3D Gaussian Splatting (3DGS), an optimization method integratingjoint spatial-frequency supervision with multi-view consistency pruning was proposed. In terms of geometry, a multi-view consistency pruning algorithm was designed, computing the local confidence of each Gaussian primitive based on a cross-view projection consistency criterion. This is further combined with voxelized neighborhood consensus information to smooth the geometric space and periodically eliminate redundant Gaussians. For supervision, a joint spatial-domain and frequency-domain strategy was introduced. In the spatial domain, a spatially adaptive L1 loss function was devised, in which a feature extractor was constructed to generate a spatial weight map that enhances gradient contributions in high-error regions, while an energy-conservation scaling factor is employed to control the training magnitude. In the frequency domain, a magnitude regularization method based on the Dual-Tree Complex Wavelet Transform (DTCWT) was proposed, which leverages its approximate shift-invariance and multi-directional properties to constrain multi-scale subband magnitude differences during the fine-tuning stage. This smoothly adjusts the Gaussian attributes under prior guidance, mitigating the oscillation and surge in point counts caused by phase constraints. Experimental results on three public datasets, including Mip-NeRF360, demonstrated that the proposed method achieves effective improvements in both rendering quality and model compactness over the baseline 3DGS model: Structural Similarity (SSIM) is improved from 0.851 to 0.860, Peak Signal-to-Noise Ratio (PSNR) increased from 23.75 dB to 24.17 dB, and Learned Perceptual Image Patch Similarity (LPIPS) decreased from 0.169 to 0.148. In terms of model storage, the model size on the Deep Blending dataset was reduced from 628 MB to 417 MB, achieving a compression rate of approximately 33.6%. Overall, the proposed joint spatial-frequency supervision and physically grounded pruning method enabled both high-fidelity reconstruction and compact storage.

Local editing of 3D scenes constitutes a fundamental task in content generation, with substantial application value in domains such as post-production, digital twins, and game development. 3D Gaussian Splatting (3DGS), as a representative technique, has achieved pivotal breakthroughs in real-time rendering, thereby providing robust support for efficient editing workflows. Nevertheless, existing diffusion-based methods for local texture editing of 3D Gaussian scenes are commonly plagued by semantic leakage and background drift, which limit editing fidelity. To tackle these challenges, a semantically guided region-adaptive approach was proposed for local texture editing of 3D scenes. The proposed method was designed to deliver high-quality texture editing while rigorously preserving the geometric and appearance consistency of non-edited regions. Concretely, a high-confidence 3D semantic mask generation module was first introduced, which incorporates a vision-language foundation model, employs LangSAM in conjunction with a confidence-filtering strategy, and constructs a robust 3D semantic field via multi-view voting, thereby eliminating semantic leakage at its source. Subsequently, a saliency-guided anchor view propagation module was devised, which automatically selects optimal anchor views based on semantic saliency scores and integrates a multi-view attention consistency mechanism to effectively mitigate temporal flickering. Finally, a region-adaptive gradient update strategy was formulated, which decouples the foreground editing loss from the background anchoring loss, enabling fine-grained scene optimization. Experiments conducted on five editing tasks across the Instruct-NeRF2NeRF and Mip-NeRF 360 datasets demonstrated that, when compared with baseline methods such as GaussianEditor and DGE, the proposed method performed better overal on quantitative metrics such as CLIP Score, Background LPIPS, Background PSNR, and Temporal LPIPS. Qualitative evaluation further showed that the proposed method not only exhibited overall advantages in complex semantic decoupling and background preservation tasks, but also achieved a favorable balance between texture quality and geometric integrity.

For large-span architectural curved gridshells, a method for generating and optimizing quadrilateral grids aligned with a regular orthogonal direction field on surfaces was proposed. Based on global parameterization constructed using periodic transition functions, the spline surface was reparameterized to align its isoparametric lines with the target orthogonal direction field. Using the reparameterized coordinates of vertices on the discretized surface as variables, an energy functional was formulated to measure the deviation between the gradient of the reparameterization and the direction field. The functional was discretized into a weighted edge-based energy form. Energy minimization yielded a discrete reparameterized domain aligned with the direction field. Based on the periodic transition relationships between adjacent triangular parameter domains, continuous parametric triangles were reconstructed in the reparameterized domain. Isoparametric lines were then extracted and mapped back onto the surface to generate an initial quadrilateral grid satisfying the directional constraints. A mapping between isoparametric line spacing and target grid size was established, enabling direct control of the grid density and rapid generation of grids with different sizes without repeated parameterization. To satisfy the geometric requirements of architectural curved gridshells, a multi-objective optimization model considering grid planarity, smoothness, and size uniformity was further developed. Vertex positions were iteratively optimized in the original parametric domain of the surface to obtain grid layouts compatible with architectural design requirements. Numerical experiments demonstrated that the method efficiently generated quadrilateral grids aligned with arbitrary regular orthogonal direction fields, and outperformed comparison methods in planarity, uniformity, and overall grid quality. The method also exhibited robustness to different initial triangulations and enabled real-time control of grid sizes through adjusting isoparametric lines. Structural analysis of gridshells further showed that, under identical member cross-sections and material properties, the optimized grid layouts effectively reduced the maximum member stress and improved the overall stress distribution, while maintaining almost unchanged steel consumption, demonstrating the effectiveness and engineering applicability of the method for architectural curved gridshell design.

Mesh generation is a key step in the preprocessing stage of numerical simulation, and mesh quality directly affects the computational accuracy, convergence stability, and computational efficiency of simulation results. Flow-field simulation of fuel solenoid valves involves complex structures such as inlet narrow slots, narrow channels, transition corners, and near-wall regions. These regions have small local scales and obvious geometric variations, and velocity gradients and pressure changes are likely to occur when fluid passes through them. Therefore, local meshes with smaller element sizes, denser distributions, and more stable quality are required. Traditional mesh generation methods often suffer from insufficient resolution in key regions, inadequate boundary-layer representation, and unstable local element quality when dealing with complex small-scale flow-channel structures, which affects the reliability of flow-field simulation. To address these problems, a mesh generation method for flow-field simulation of fuel solenoid valves was designed. First, topological inspection and defect repair were performed on the geometric model to eliminate geometric errors that may affect mesh generation, and key regions such as inlet narrow slots, narrow channels, and near-wall regions were identified. Then, a spatial size field was constructed to guide controlled surface discretization, so that the mesh size can be adjusted according to geometric scales and flow-sensitive regions. Transition elements were also arranged layer by layer along the normal direction near the wall surface to form a high-resolution boundary-layer mesh, thereby improving the ability to capture near-wall flow characteristics. Furthermore, adaptive local refinement was carried out by combining geometric gradients and error indicators. Key channels, narrow slots, and local abrupt regions through which fuel flows were refined, and high-quality tetrahedral meshes were generated. Finally, mesh smoothing and element quality optimization were used to improve element shape, size transition, and distribution uniformity. The generated mesh was applied to the flow-field simulation of the fuel solenoid valve, and its simulation efficiency and accuracy were compared with those of traditional meshing. The experimental results showed that the mesh generation method for flow-field simulation of fuel solenoid valves reduced preprocessing time by 73.62% and overall workflow time by 40.75%. While improving mesh generation and simulation efficiency, the simulation accuracy remained consistent with that of traditional methods, showing good application value.

To support real-time walkthroughs in Web-based 3D indoor scenes such as virtual museums and digital exhibition halls, a fine-grained dynamic loading method integrating mesh saliency analysis was proposed to address long initial loading time, slow response to viewpoint changes, and low resource allocation efficiency under limited bandwidth caused by complex models and high-resolution textures. Guided by users’ visual-attention characteristics, the method incorporated saliency information throughout the entire pipeline of 3D scene preprocessing, multiresolution-structure construction, and online loading scheduling, thereby improving transmission and rendering efficiency while preserving the visual quality of key regions. First, considering that viewpoints during indoor roaming are constrained by navigation paths and that regions of interest are relatively concentrated, a guided-path-based multi-view projection sampling strategy was adopted to acquire scene images. During saliency prediction, a multi-input encoder-decoder architecture was combined with adversarial training to regress saliency maps for the projected images, and a texture-enhancement module was introduced to strengthen the representation of high-resolution textures. The resulting multi-view 2D saliency maps were then mapped back onto the 3D mesh, and neighborhood-smoothing optimization was performed to generate mesh saliency with spatially coherent distribution. Based on this distribution, saliency-priority boundaries were introduced into the V-Partitions multiresolution modeling framework, so that regions with high visual attention and semantic continuity preserved within the same triangular patch as much as possible, forming a hierarchical Directed Acyclic Graph (DAG) structure with saliency attributes. During online execution, a two-level scheduling mechanism was designed: At the scene level, global importance was computed according to the spatial relationship between objects and the user viewpoint; at the object level, local importance was determined by jointly considering saliency and screen-space error. A global priority queue is further employed to realize a patch-level dynamic scheduling framework. Experimental results showed that the method significantly reduced initial loading time and viewpoint-switching response latency, while improving both the overall PSNR of rendered images and the ROI-PSNR of salient regions. The online system supported interactive rendering of 3D scenes larger than 500 MB, providing an effective solution for Web-based 3D scene visualization.

Drawing review records generated during detailed design are often colloquial. Their semantic ambiguity, inconsistent engineering terminology, and strong dependence on drawing context make engineering experience difficult to reuse. The objective was to develop a structured translation method that converts unstructured engineering text into computable drawing-review rules, reduces the cost of manual rule authoring, and improves the efficiency of accumulating reusable engineering experience. Building Information Modeling (BIM) checking requirements were used as the operational constraint. Each candidate rule was represented as a quadruple comprising a review subject, a review attribute, a logical operator, and a threshold or object value. A domain semantic dictionary was constructed to align colloquial and nonstandard construction terms with Industry Foundation Classes (IFC) entities, and attributes. The implementation workflow contained six connected stages. First, raw records were cleaned, irrelevant meeting information was removed, separated questions and replies were merged, and compound entries were divided into minimal issue units. Second, BIM checkability was determined for each unit or requiring human judgment according to the availability of explicit objects, geometric relations, measurable attributes, spatial constraints, or existence requirements. Third, BIM-checkable units were decomposed into the four semantic slots while preserving thresholds, units, reference objects, and uncertainty markers. Fourth, extracted entities and properties were disambiguated using the semantic dictionary and IFC terminology; ambiguous matches were retained as candidates for review. Fifth, the logical operator was matched to available computational functions for attribute, spatial, and existence constraints. Sixth, the completed slots were instantiated as Structured Natural Language (SNL) statements with prerequisite conditions and executable checking clauses. A Large Language Model (LLM) prompt architecture incorporating chain-of-thought reasoning was used for staged reasoning and self-checking of results. Two hundred samples were selected from the company’s historical drawing review records. Five experienced engineers created the expert reference answers. A separate panel of ten senior engineers independently scored anonymized outputs for BIM checkability, component naming, component attributes, logical operators, and SNL conformity. A general-model group using a simple prompt was compared with the proposed chain-of-thought-enhanced group under the same base-model environment. The chain-of-thought-enhanced group obtained score rates of 98.0% for checkability, 91.7% for component names, 91.3% for component attributes, 92.0% for logical operators, and 94.3% for SNL conformity. Compared with the general-model group, the corresponding gains were 27.0, 34.4, 38.6, 33.6, and 43.9 percentage points. The largest improvement occurred in SNL conformity, indicating that the coordinated use of the semantic dictionary, rule quadruple, computational-function constraints, and staged prompts substantially reduced uncontrolled expression. Score rates for simple, routine, and complex samples were 97.3%, 93.9%, and 86.6%, respectively. The improvement was more pronounced for complex samples. Processing all 200 samples required approximately 20 minutes for the general-model group and 1.5 hours for the enhanced group. Processing by the enhanced group remained markedly faster than expert reference construction, which required approximately 50 person-hours, or more than 10 hours with five experts working in parallel. Coordinating controlled semantic slots, a domain semantic dictionary, SNL templates, and staged reasoning can improve the consistency, conformity, and practical efficiency of translating drawing review records into reusable BIM checking rules. The workflow provides an implementable route for converting dispersed engineering experience into structured digital assets while retaining uncertainty and manual-review flags.

Ecological governance in the Yellow River Basin involves coordinating multiple agents, tasks, and mechanisms. Multi-source texts including policy documents, corporate reports, and public-opinion texts contain numerous implicit governance relationships and clues regarding conflict transmission. Converting this information into computable graph structures to identify and measure conflicts represents a key challenge in complex network governance analysis. To address the limitations of existing methods in organizing unstructured governance semantics, expressing conflict topology, and quantifying conflict intensity, a complex network conflict measurement method integrating knowledge graphs with Relational Graph Attention Networks (R-GAT) was proposed, drawing on approaches to complex-object modeling, topological relationship representation, and visual analysis in graphics. The semantic structures and directed relational networks provided by knowledge graphs were used as structural foundations and semantic priors for R-GAT, through which, through which the impact intensity of different relational edges on conflict formation, were learned and complex governance relationships were transformed from textual semantics into graph structures, topological analysis, and quantitative measurement. Experimental results demonstrated favorable performance in ranking conflict intensity and identifying strong conflict relationships, and key conflict relationships within the complex network of Yellow River ecological governance were effectively identified. A knowledge-graph-driven graph-learning method was thereby provided for graph-structural modeling of complex relational objects and multi-agent governance conflict measurement.

As critical load-bearing components of aircraft wings, wing ribs play an important role in improving the overall performance of aircraft through lightweight design. Traditional wing rib design mainly relies on engineering experience to determine structural dimensions and lacks systematic optimization methods. Structural optimization methods based on finite element analysis have limitations in geometric representation accuracy and design-analysis integration. Based on weakly coupled isogeometric analysis and density-based topology optimization, a topology optimization design framework for aircraft wing ribs in complex configurations composed of skin, ribs, and spars was established. Firstly, a precise geometric model of the wing structure was constructed using non-uniform rational B-splines. On this basis, an isogeometric analysis framework based on the Reissner-Mindlin plate-shell theory and a penalty-based weak coupling method was established. Its accuracy and reliability were verified by comparison with computational results obtained using the commercial finite element software Abaqus. Secondly, with compliance used as the objective function and the optimality criteria method adopeted as the optimization algorithm, a density-based isogeometric topology optimization model was developed, and the effects of the density penalization factor and filter radius on the optimization results were analyzed. Finally, weakly coupled isogeometric analysis and density-based topology optimization were applied to the concurrent topology optimization design of wing ribs in two complex wing configurations, namely the RAE-2822 and Davis B-24 wings. The optimized and original models were subjected to finite element reanalysis and comparison using Abaqus to verify the effectiveness of the design results. The results showed that, the established weakly coupled isogeometric topology optimization method can optimize the material distribution in the wing-rib design domain and reduce structural weight while keeping the overall displacement response of the wing nearly unchanged, thereby providing a feasible technical approach for lightweight optimization design of complex thin-walled wing structures.

Currently, the design of lattice structures for additive manufacturing is disconnected from the subsequent slicing. Traditional methods rely on generic slicing algorithms based on STL (Standard Triangle Language) models, in which the entire lattice structure must first be discretized into a triangular mesh, and contours are then generated by intersecting the mesh with slicing plane. However, lattice structures contain a large number of members, resulting in massive STL model data volumes. Furthermore, the approximate representation of triangular patches is prone to introducing geometric errors, leading to reduced slicing accuracy. At the same time, the intersection calculation requires traversing a large number of redundant triangular patches, resulting in low computational efficiency and making it difficult to meet the demand for rapid preprocessing of large-scale lattice structures. To address these issues, a direct slicing method tailored for lattice structures was proposed to avoid the intermediate conversion to an STL model. This method first abstracted each connecting rod into a cylindrical or conical geometric shape and directly utilized the topological position data of the members generated during the design phase. to establish an analytical geometric relationship between the cutting plane and the members. By applying the closed-form formula for the intersection of a spatial line and a quadratic surface, the method directly calculated the intersection contour between the members and the cutting plane, yielding precise cross-sectional shapes. Geometric distortion caused by triangular-mesh discretization was thereby avoided, and improves contour accuracy was improved. To optimize computational efficiency, data-mapping technology was introduced to transform the analytical solution of conic-section contours into linear mappings and matrix operations based on the unit circle. By pre-computing transformation matrices, this method enabled contour reconstruction for each slice layer by simply updating the member center coordinates, effectively eliminating redundant calculations and accelerating the processing of large-scale grids. Furthermore, for complex contours formed by the intersection of multiple members at nodes, a polygon merging operation was employed for post-processing to eliminate contour intersections and redundant edges, thereby generating continuous, self-intersection-free slice contours. Experimental results demonstrated that the proposed algorithm exhibited good feasibility and effectiveness, capable of improving the efficiency and accuracy of preprocessing for additive manufacturing models of lattice structures.

In the design of overhead transmission lines, design changes frequently occur during inter-conductor clearance verification to satisfy stringent mechanical and insulation safety constraints. To address the low efficiency of variant design caused by switching among multiple software modules, repeated procedures, and excessive iterations in traditional methods, a fast variant design-supporting methodology for inter-conductor clearance verification was established. Firstly, based on the Top Basic Skeleton (TBS) modeling theory, a top-down design flow was proposed. In the proposed design flow, the tower design control information and top-level design requirements were transmitted to the conductor configuration parameters, enabling the clearance-verification process to respond rapidly to common design changes. Subsequently, in accordance with the model-granularity requirements of the design flow, a lightweight transmission-line skeleton model comprising conductors, towers, and their assembly relationships was developed. Specifically, a lightweight tower-shape skeleton was constructed by referencing general model definitions from industrial standards, enabling rapid creation and modification of towers. Meanwhile, a tower-positioning skeleton was built following the TBS lateral layout modeling approach, allowing efficient and accurate transfer of design changes. Additionally, a conductor skeleton model integrating both geometric and design information was established to facilitate data sharing for rapid mechanical calculation of wires. Furthermore, an assembly model incorporating the connections between towers and conductors was constructed to ensure data consistency in variant design and adaptability to complex line configurations. Finally, a corresponding prototype inter-conductor clearance verification system was developed based on the Qt platform, and various typical and special transmission line projects were used as case studies to demonstrate the application of variant design requirements. The results demonstrated that the proposed method effectively supported the four major types of rapid variant-designs: adjusting tower position, tower-top geometry, phase sequence arrangement, and custom operating conditions, while exhibiting good scalability for complex tension-section-level overhead transmission-line projects. Consequently, this research provided an efficient and flexible technical approach for increasing the efficiency of the insulation coordination design.

Augmented Reality (AR) interaction in digital cultural tourism has become an important means of influencing tourists’ information acquisition, emotional arousal, and visual engagement. However, the mechanisms through which different interaction modes affect tourists’ emotional experience and visual search behavior remain unclear, and empirical evidence for interaction optimization across different scenarios is still insufficient. To reveal the effects of scenario type and AR interaction mode on tourists’ emotional experience and visual search behavior, a measurement framework combining subjective rating scales and eye-tracking physiological measures was constructed. Two types of scenarios, namely knowledge-oriented learning and entertainment-oriented interaction, were used as experimental contexts. Using a thematic exhibition at the Shaanxi History Museum as the experimental content, four interaction conditions were designed, including traditional browsing, voice interaction, touch interaction, and gesture interaction. A 2 × 4 within-subject experimental with valid data design was adopted. PAD emotional rating data and eye-tracking data were collected from 24 participants under eight experimental conditions. The subjective indicators included pleasure, arousal, and dominance, while the objective indicators included fixation proportion, average pupil diameter, and saccade rate per unit time unit. Linear mixed-effects models were used to examine the effects of scenario type, interaction mode, and their interaction on the above indicators. The results showed that the effects of AR interaction modes on tourists’ emotional experience and visual search behavior varied across scenarios. In the knowledge-oriented learning scenario, gesture interaction showed certain advantages in arousal and in some pairwise comparisons of dominance. However, the eye-tracking results did not show a consistent advantage in visual attention stability, indicating that interaction optimization in knowledge-oriented tasks should take into account the clarity of information presentation, the stability of operational feedback, and the control of visual load. In the entertainment-oriented interaction scenario, gesture interaction produced a more positive emotional experience than touch interaction. It was also accompanied by a higher fixation proportion, larger average pupil diameter, and higher saccade rate per unit time. These findings suggested that gesture interaction could enhance visual engagement and emotional arousal, but it might also increase the intensity of dynamic content search. For knowledge-oriented scenarios, the organization of information and the stability of interaction should be emphasized. For entertainment-oriented scenarios, embodied operation and dynamic feedback can be moderately strengthened. The results not only revealed the differences in emotional experience and visual behavior characteristics produced by different AR interaction modes in knowledge-oriented and entertainment-oriented scenarios, but also provided empirical evidence and methodological guidance for scenario-based interaction configuration, emotional experience optimization, and design strategy development in digital cultural tourism AR applications.

To overcome traditional design methods’ inability to capture differentiated emotional needs, a new approach combining user cluster theory and swarm intelligence was developed. Based on online user review data, the RCLA model was used to identify high-value user groups; natural language processing and social network analysis were combined to identify heterogeneous user clusters; and the entropy weighting method was employed to quantify the weight of emotional needs. Second, morphological analysis was used to deconstruct product design elements and establish a coding system. Subsequently, type I quantification theory was used to calculate the contribution of design elements to emotional needs, and an ant colony algorithm was introduced to achieve intelligent configuration of design elements, combined with a user feedback mechanism to complete iterative optimization of the design. Using a vacuum cleaner as a case study, a semantic differential scale was employed to conduct satisfaction evaluations and validate the design solutions. The results demonstrated that the initial optimized solution achieved significantly greater satisfaction of emotional needs compared to traditional design samples. The final solution, refined through user feedback iterations further identified subtle differences in needs within clusters, continuously enhancing the design’s user adaptability and satisfaction. The study showed that this data-driven emotional design method, integrating intelligent algorithms, overcame the limitations of traditional small-sample analysis and offer new insights into product innovation under dynamic demand.

With the accelerating aging process in China, building an accessible and cognitive-friendly community environment has become a crucial initiative to enhance older adults’ quality of life, independent living capabilities, and sense of security. As the core vehicle for spatial guidance and information interaction, community environmental signage directly affects older adults’ mobility and social participation through its information accessibility and visual legibility. Grounded in the two key cognitive processes involved when older adults perceive and comprehend signs-icon recognition and semantic matching-the aim was to systematically reveal the differential characteristics and underlying factors influencing Chinese community-dwelling older adults’ ability to recognize typical environmental signs. To this end, pre- and post-test comparative experiments were conducted across four representative cities (Beijing, Shanghai, Chongqing, and Chengdu), and valid icon-recognition data were collected from 735 community-dwelling older adults. Recognition accuracy was evaluated for 26 typical community environmental signs across both icon recognition and semantic matching dimensions. The results showed that educational attainment was most strongly associated with performance in both tasks, with significantly higher recognition rates observed among participants with higher education in both icon recognition and semantic comprehension. In the semantic matching task, distinct inter-group differences emerged across age brackets: the semantic-recognition accuracy of participants in their 60s was significantly higher than that of participants in their 70s and 80s, while cognitive performance between the 70s and 80s groups was relatively comparable. Based on these findings, systematic age-friendly optimization redesign was performed for frequently misinterpreted community signs with pre-test accuracy rates below 80%. Subsequent re-testing revealed that the re-designed signs increased recognition accuracy among older adults by 12.1% to 25.3%. The findings further revealed the cognitive processing patterns and individual variations of older adults regarding visual environmental symbols, and provided practical scientific evidence and actionable design guidelines for the standardized development of age-friendly community signage systems and the development of cognitively friendly communities.