3D Mesh Research Articles

Automated 2D and 3D finite element overclosure adjustment and mesh morphing using generalized regression neural networks

Computer representations of three-dimensional (3D) geometries are crucial for simulating systems and processes in engineering and science. In medicine, and more specifically, biomechanics and orthopaedics, obtaining and using 3D geometries is critical to many workflows. However, while many tools exist to obtain 3D geometries of organic structures, little has been done to make them usable for their intended medical purposes. Furthermore, many of the proposed tools are proprietary, limiting their use. This work introduces two novel algorithms based on Generalized Regression Neural Networks (GRNN) and 4 processes to perform mesh morphing and overclosure adjustment. These algorithms were implemented, and test cases were used to validate them against existing algorithms to demonstrate improved performance. The resulting algorithms demonstrate improvements to existing techniques based on Radial Basis Function (RBF) networks by converting to GRNN-based implementations. Implementations in MATLAB of these algorithms and the source code are publicly available at the following locations: https://github.com/thor-andreassen/femors; https://simtk.org/projects/femors-rbf; https://www.mathworks.com/matlabcentral/fileexchange/120353-finite-element-morphing-overclosure-reduction-and-slicing

Efficient prediction of individual head-related transfer functions based on 3D meshes

Individual head-related transfer functions (HRTFs) are critical for binaural spatial audio rendering. In contrast to anthropometric parameters and pinnae images, 3D meshes allow for a more direct and comprehensive representation of the anthropometric structure, which provides highly effective inputs for modeling individualized HRTFs. This paper presents a neural network-based method for predicting individualized HRTFs in full space based on 3D meshes. Unlike many previous methods that estimate HRTF spectra at sampling grids or frequencies separately, the proposed model predicts the HRTF spectra of each vertical plane by considering the spectral correlation and continuity across adjacent sampling grids and frequencies. Evaluation results indicate that the proposed method enhances the prominence of peaks and notches in the obtained HRTF spectra and improves the speed and accuracy of HRTF individualization. The log spectral distortion of the proposed method is lower than that of state-of-the-art methods using anthropometric parameters and pinnae images. Further evaluation confirms that the proposed method requires significantly fewer points in 3D meshes when compared to numerical simulation methods. The evaluation based on localization models demonstrates that the HRTFs predicted by the proposed method are perceptually similar to the measured HRTFs.

MeshCL: Towards robust 3D mesh analysis via contrastive learning

Recent years have witnessed the success of deep learning in 3D mesh analysis, including classification and segmentation. Previous works mainly rely on supervised learning, which trains models on the labeled 3D mesh data with the cross-entropy loss. However, in the supervised training, the very limited amount of 3D mesh data restricts the feature’s discriminative power and the model’s robustness. To address or at least alleviate this issue, we propose to regularize the learned representations by associating the data points in feature space via contrastive learning. To achieve this, we first explore mesh samples with diverse geometric structures, and then apply contrastive regularization into the positive and negative pairs sampled from the feature space. In this way, our method, MeshCL, can learn robust feature representations invariant to the geometric deformation for 3D manifold mesh. In detail, to augment the geometric diversity, we examine different data augmentation techniques for mesh data. Furthermore, considering the geometric characteristics of 3D mesh, we develop more efficient positive/negative pair sampling strategies for the dense prediction task. To verify the effectiveness of MeshCL, we study two typical mesh analysis tasks, classification and segmentation. Experimental results demonstrate that MeshCL not only achieves state-of-the-art performance on the two tasks mentioned above but also significantly improves the model’s robustness.

Convergence analysis of virtual element method for the electric interface model on polygonal meshes with small edges

A conforming virtual element method is studied for approximating the solution of a pulsed electric interface model on polygonal meshes with small edges or faces, which traditional virtual and finite element methods cannot easily handle. One of the significant advantages of this virtual element method is to generate interface-conforming meshes efficiently exploiting polygonal meshes and hanging nodes. The virtual element method employs specific projection operators to establish optimal error estimates with reasonable assumptions regarding the solution's regularity. The voltage potential of the pulsed electric model through the physical media is approximated by using a fully discrete virtual element approach based on Crank-Nicolson temporal discretization. Numerical examples are used to illustrate the expected order of convergence. The efficiency and robustness of the proposed method are displayed on interface-independent/dependent background-fitted meshes with small edges and large magnitudes of discontinuities across the interface by these experiments.

Ultra-dense Motion Capture: An exploratory full-automatic approach for dense tracking of breast motion in 4D.

Understanding the dynamic deformation pattern and biomechanical properties of breasts is crucial in various fields, including designing ergonomic bras and customized prostheses, as well as in clinical practice. Previous studies have recorded and analyzed the dynamic behaviors of the breast surface using 4D scanning, which provides a sequence of 3D meshes during movement with high spatial and temporal resolutions. However, these studies are limited by the lack of robust and automated data processing methods which result in limited data coverage or error-prone analysis results. To address this issue, we identify revealing inter-frame dense correspondence as the core challenge towards conducting reliable and consistent analysis of the 4D scanning data. We proposed a fully-automatic approach named Ulta-dense Motion Capture (UdMC) using Thin-plate Spline (TPS) to augment the sparse landmarks recorded via motion capture (MoCap) as initial dense correspondence and then rectified it with a sophisticated post-alignment scheme. Two downstream tasks are demonstrated to validate its applicability: virtual landmark tracking and deformation intensity analysis. For evaluation, a dynamic 4D human breast anthropometric dataset DynaBreastLite was constructed. The results show that our approach can robustly capture the dynamic deformation characteristics of the breast surfaces, significantly outperforms baselines adapted from previous works in terms of accuracy, consistency, and efficiency. For 10 fps dataset, average error of 0.25 cm on control-landmarks and 0.33 cm on non-control (arbitrary) landmarks were achieved, with 17-70 times faster computation time. Evaluation was also carried out on 60 fps and 120 fps datasets, with consistent and large performance gaining being observed. The proposed method may contribute to advancing research in breast anthropometry, biomechanics, and ergonomics by enabling more accurate tracking of the breast surface deformation patterns and dynamic characteristics.

Physically Informed Deep Learning Technique for Estimating Blood Flow Parameters in Four-Vessel Junction after the Fontan Procedure

This study introduces an innovative approach leveraging physics-informed neural networks (PINNs) for the efficient computation of blood flows at the boundaries of a four-vessel junction formed by a Fontan procedure. The methodology incorporates a 3D mesh generation technique based on the parameterization of the junction’s geometry, coupled with an advanced physically regularized neural network architecture. Synthetic datasets are generated through stationary 3D Navier–Stokes simulations within immobile boundaries, offering a precise alternative to resource-intensive computations. A comparative analysis of standard grid sampling and Latin hypercube sampling data generation methods is conducted, resulting in datasets comprising 1.1×104 and 5×103 samples, respectively. The following two families of feed-forward neural networks (FFNNs) are then compared: the conventional “black-box” approach using mean squared error (MSE) and a physically informed FFNN employing a physically regularized loss function (PRLF), incorporating mass conservation law. The study demonstrates that combining PRLF with Latin hypercube sampling enables the rapid minimization of relative error (RE) when using a smaller dataset, achieving a relative error value of 6% on the test set. This approach offers a viable alternative to resource-intensive simulations, showcasing potential applications in patient-specific 1D network models of hemodynamics.

A posteriori error analysis and mesh adaptivity for a virtual element method solving the Stokes equations

We investigate an adaptive mesh strategy for the conforming virtual element method (VEM) of the Stokes equations proposed in Manzini and Mazzia (2022). The VEM generalizes the finite element approach to polygonal and polyehedral meshes in the framework of Galerkin approximation. The scheme of Manzini and Mazzia (2022) is inf-sup stable, converges optimally in the L2 and energy norm for all polynomial orders k≥1, and the Stokes velocity is weakly divergence-free at the machine precision level. Our adaptive mesh strategy is based on a suitable residual-based a posteriori indicator. A posteriori analysis shows that such an indicator is theoretically efficient and reliable. Our numerical experiments show that it can be an efficient tool for solving scientific and engineering problems by applying it to a set of representative situations, including the case of a weakly singular solution as that of the “L-shape” domain.

An automatic mesh generator for coupled 1D–2D hydrodynamic models

Abstract. Two-dimensional (2D), depth-averaged shallow water equation (SWE) models are routinely used to simulate flooding in coastal areas – areas that often include vast networks of channels and flood-control topographic features and/or structures, such as barrier islands and levees. Adequately resolving these features within the confines of a 2D model can be computationally expensive, which has led to coupling 2D simulation tools to less expensive one-dimensional (1D) models. Under certain 1D–2D coupling approaches, this introduces internal constraints that must be considered in the generation of the 2D computational mesh used. In this paper, we further develop an existing automatic unstructured mesh generation tool for SWE models, ADMESH+, to sequentially (i) identify 1D constraints from the raw input data used in the mesh generation process, namely the digital elevation model (DEM) and land–water delineation data; (ii) distribute grid points along these internal constraints, according to feature curvature and user-prescribed minimum grid spacing; and (iii) integrate these internal constraints into the 2D mesh size function and mesh generation processes. The developed techniques, which include a novel approach for determining the so-called medial axis of a polygon, are described in detail and demonstrated on three test cases, including two inland watersheds with vast networks of channels and a complex estuarine system on the Texas, USA, coast.

A bubble VEM-fully discrete polytopal scheme for mixed-dimensional poromechanics with frictional contact at matrix–fracture interfaces

This article addresses the discretisation of fractured/faulted poromechanical models using 3D polyhedral meshes in order to cope with the geometrical complexity of faulted geological models. A new polytopal scheme is proposed for contact-mechanics, based on a mixed formulation combining a fully discrete space and suitable reconstruction operators for the displacement field together with a face-wise constant approximation of the Lagrange multiplier, accounting for the surface tractions along the fracture/fault network. To ensure the inf–sup stability of the mixed formulation, a bubble-like degree of freedom is included in the discrete space of displacements and used in the reconstruction operators. This fully discrete scheme for the displacement is equivalent to a low-order Virtual Element scheme, with a bubble enrichment of the virtual space. This P1-bubble VEM–P0 mixed discretisation is combined with an Hybrid Finite Volume scheme for the Darcy flow. The proposed approach is adapted to complex geometry with a network of planar faults/fractures which can include corners, tips and intersections; it leads to efficient semi-smooth Newton solvers for the contact-mechanics, and preserves the dissipative properties of the fully coupled model. The scheme is numerically investigated in terms of convergence and robustness on several 2D and 3D test cases, using either analytical or numerical reference solutions and considering both the stand alone static contact-mechanics model and the fully coupled poromechanical model.

Global dust density in two-dimensional complex plasma

The driven-dissipative Langevin dynamics simulation is used to produce a two-dimensional (2D) dense cloud, which is composed of charged dust particles trapped in a quadratic potential. A 2D mesh grid is built to analyze the center-to-wall dust density. It is found that the local dust density in the outer region relative to that of the inner region is more nonuniform, being consistent with the feature of quadratic potential. The dependences of the global dust density on equilibrium temperature, particle size, confinement strength, and confinement shape are investigated. It is found that the particle size, the confinement strength, and the confinement shape strongly affect the global dust density, while the equilibrium temperature plays a minor effect on it. In the direction where there is a stronger confinement, the dust density gradient is bigger.

SpATr: MoCap 3D human action recognition based on spiral auto-encoder and transformer network

Recent technological advancements have significantly expanded the potential of human action recognition through harnessing the power of 3D data. This data provides a richer understanding of actions, including depth information that enables more accurate analysis of spatial and temporal characteristics. In this context, We study the challenge of 3D human action recognition. Unlike prior methods, that rely on sampling 2D depth images, skeleton points, or point clouds, often leading to substantial memory requirements and the ability to handle only short sequences, we introduce a novel approach for 3D human action recognition, denoted as SpATr (Spiral Auto-encoder and Transformer Network), specifically designed for fixed-topology mesh sequences. The SpATr model disentangles space and time in the mesh sequences. A lightweight auto-encoder, based on spiral convolutions, is employed to extract spatial geometrical features from each 3D mesh. These convolutions are lightweight and specifically designed for fix-topology mesh data. Subsequently, a temporal transformer, based on self-attention, captures the temporal context within the feature sequence. The self-attention mechanism enables long-range dependencies capturing and parallel processing, enabling scalability for long sequences. The proposed method is evaluated on three prominent 3D human action datasets: Babel, MoVi, and BMLrub, from the Archive of Motion Capture As Surface Shapes (AMASS). Our results analysis demonstrates the competitive performance of our SpATr model in 3D human action recognition while maintaining efficient memory usage. The code and the training results are publicly available at https://github.com/h-bouzid/spatr.

Improving the Automated Diagnosis of Breast Cancer with Mesh Reconstruction of Ultrasound Images Incorporating 3D Mesh Features and a Graph Attention Network

This study proposes a novel approach for breast tumor classification from ultrasound images into benign and malignant by converting the region of interest (ROI) of a 2D ultrasound image into a 3D representation using the point-e system, allowing for in-depth analysis of underlying characteristics. Instead of relying solely on 2D imaging features, this method extracts 3D mesh features that describe tumor patterns more precisely. Ten informative and medically relevant mesh features are extracted and assessed with two feature selection techniques. Additionally, a feature pattern analysis has been conducted to determine the feature’s significance. A feature table with dimensions of 445 × 12 is generated and a graph is constructed, considering the rows as nodes and the relationships among the nodes as edges. The Spearman correlation coefficient method is employed to identify edges between the strongly connected nodes (with a correlation score greater than or equal to 0.7), resulting in a graph containing 56,054 edges and 445 nodes. A graph attention network (GAT) is proposed for the classification task and the model is optimized with an ablation study, resulting in the highest accuracy of 99.34%. The performance of the proposed model is compared with ten machine learning (ML) models and one-dimensional convolutional neural network where the test accuracy of these models ranges from 73 to 91%. Our novel 3D mesh-based approach, coupled with the GAT, yields promising performance for breast tumor classification, outperforming traditional models, and has the potential to reduce time and effort of radiologists providing a reliable diagnostic system.

NeuroEditor: a tool to edit and visualize neuronal morphologies.

The digital extraction of detailed neuronal morphologies from microscopy data is an essential step in the study of neurons. Ever since Cajal's work, the acquisition and analysis of neuron anatomy has yielded invaluable insight into the nervous system, which has led to our present understanding of many structural and functional aspects of the brain and the nervous system, well beyond the anatomical perspective. Obtaining detailed anatomical data, though, is not a simple task. Despite recent progress, acquiring neuron details still involves using labor-intensive, error prone methods that facilitate the introduction of inaccuracies and mistakes. In consequence, getting reliable morphological tracings usually needs the completion of post-processing steps that require user intervention to ensure the extracted data accuracy. Within this framework, this paper presents NeuroEditor, a new software tool for visualization, editing and correction of previously reconstructed neuronal tracings. This tool has been developed specifically for alleviating the burden associated with the acquisition of detailed morphologies. NeuroEditor offers a set of algorithms that can automatically detect the presence of potential errors in tracings. The tool facilitates users to explore an error with a simple mouse click so that it can be corrected manually or, where applicable, automatically. In some cases, this tool can also propose a set of actions to automatically correct a particular type of error. Additionally, this tool allows users to visualize and compare the original and modified tracings, also providing a 3D mesh that approximates the neuronal membrane. The approximation of this mesh is computed and recomputed on-the-fly, reflecting any instantaneous changes during the tracing process. Moreover, NeuroEditor can be easily extended by users, who can program their own algorithms in Python and run them within the tool. Last, this paper includes an example showing how users can easily define a customized workflow by applying a sequence of editing operations. The edited morphology can then be stored, together with the corresponding 3D mesh that approximates the neuronal membrane.

EVALUATION OF NERF 3D RECONSTRUCTION FOR ROCK ART DOCUMENTATION

Abstract. Digital documentation of rock art traditionally relies on a point cloud captured by a terrestrial laser scanner (TLS) or derived from an oriented image obtained using photogrammetry. In modern photogrammetry, the dense point cloud is generated using multi-view stereo (MVS) and subsequently used to generate a photorealistic 3D model. A recent method to reconstruct 3D models from images is Neural Radiance Fields (NeRF), which uses volume density to render the scenes through neural networks. The advantage of NeRF is that it can construct 3D models faster without using high computer processors and memory. NeRF has been studied in various applications, including cultural heritage, but not specifically for rock art documentation. Therefore, this paper evaluates three-dimensional (3D) reconstruction techniques using NeRF on Nerfstudio platform on two rock art datasets and compares them with the point cloud and 3D mesh models obtained from TLS and photogrammetry/MVS. The results have shown that NeRF does not match MVS in achieving geometric precision and texture quality. However, its learning-based approach accelerates reconstruction and offers potential enhancements to complement photogrammetric workflow.

ACCURACY RESEARCH OF THE EFFECT OF LOCAL SURFACE MODELING METHODS ON THE GENERATION OF 3D MODELS OF HISTORICAL OBJECTS WITH ROCK-BASED BUILDING STRUCTURES USING OPEN-SOURCE SOFTWARE

Abstract. The paper describes the effect of the local surface modelling methods known as plane, quadric and triangulation on the generation of the 3D mesh models of the historical objects with rock-based building structures using Cloudcompare, an open-source software for point cloud data processing. The methodology begins with data collection of the 3D test objects using terrestrial laser scanning technology, the pre-processing of the point cloud data, the point cloud subsampling process, the process of generating of the local surface modelling data, the process of generating the 3D mesh models and finally, the analysis, which involves the 3D surface deviation analysis between the generated 3D mesh models. The overall results shows that there are no significant differences between the 3D mesh models that was generated from all the three local surface modelling methods. The histogram analysis shows that the plane and the quadric local surface modelling methods is the best methods to be used in the research where the 3D test object to be modelled contains curvy surfaces without sharp edges and corners.

Deep Spectral Meshes: Multi-Frequency Facial Mesh Processing with Graph Neural Networks

With the rising popularity of virtual worlds, the importance of data-driven parametric models of 3D meshes has grown rapidly. Numerous applications, such as computer vision, procedural generation, and mesh editing, vastly rely on these models. However, current approaches do not allow for independent editing of deformations at different frequency levels. They also do not benefit from representing deformations at different frequencies with dedicated representations, which would better expose their properties and improve the generated meshes’ geometric and perceptual quality. In this work, spectral meshes are introduced as a method to decompose mesh deformations into low-frequency and high-frequency deformations. These features of low- and high-frequency deformations are used for representation learning with graph convolutional networks. A parametric model for 3D facial mesh synthesis is built upon the proposed framework, exposing user parameters that control disentangled high- and low-frequency deformations. Independent control of deformations at different frequencies and generation of plausible synthetic examples are mutually exclusive objectives. A Conditioning Factor is introduced to leverage these objectives. Our model takes further advantage of spectral partitioning by representing different frequency levels with disparate, more suitable representations. Low frequencies are represented with standardised Euclidean coordinates, and high frequencies with a normalised deformation representation (DR). This paper investigates applications of our proposed approach in mesh reconstruction, mesh interpolation, and multi-frequency editing. It is demonstrated that our method improves the overall quality of generated meshes on most datasets when considering both the L1 norm and perceptual Dihedral Angle Mesh Error (DAME) metrics.

Design Optimization of Blade Tip in Subsonic and Transonic Turbine Stages—Part I: Stage Design and Preliminary Tip Optimization

Abstract Rotor blade tip has significant influence on turbine stage aerodynamics and heat transfer. Most previous efforts have been based on low-speed cascade settings. However, more recent research on transonic blade tips exhibits distinctive flow features with qualitatively different performance sensitivities. These prompt two key issues of interest on the related flow conditioning. First, the contrast between a low-speed flow and a transonic regime highlights the less studied high-subsonic flow regime, closely relevant to many realistic turbine designs. Second, the relative casing movement and upstream inflow conditions, known to have non-negligible effects, indicate the need to examine a rotor blade tip in a realistic stator–rotor stage environment, which is also lacking. To elaborate the Mach number effect in the flow regimes of practical interest, we aim to examine a high subsonic stage in a direct and consistent comparison with a transonic one. To this end, a high subsonic stage (exit Mach number of 0.7) and a transonic (exit Mach number of 1.1) are designed at the same Reynolds number with a three-dimensional parameterization and meshing system. The tip squealer height is used as a representative parameter to investigate the sensitivity of the stage aerothermal performance. The multi-objective optimization using the Kriging surrogated model is employed to identify the Pareto fronts for the stage efficiency and the heat transfer. The comparison of the optimized results between these two stages shows distinctively different trends in the performance variation with the squealer height. The efficiency of the subsonic stage increases with the squealer height reaching a plateau. In contrast, the efficiency in the transonic stage first increases and then drops to the level comparable to that of a flat tip. Significantly, the present results indicate, for the first time, that the squealer tip in a transonic stage may not be as effective as in a subsonic stage. On the other hand, for heat transfer, sensitivity variations are more complex. The overall heat load and the local nonuniformity lead to qualitatively different sensitivities with the squealer height, as well as completely incomparable Pareto fronts. These observed heat transfer sensitivities raise the question on how to effectively conduct a combined aerodynamic and heat transfer performance design optimization. The authors subsequently resort further aerothermal physics analyses described in a companion paper as Part II of the two-part article. In Part II, the physical interpretation of the contrasting aero-efficiency sensitivities for the two stages, as well as a physical understanding leveraged selection of the objective function for such combined blade tip aerothermal optimization will be presented.

Advancing Breath‐Based Diagnostics: 3D Mesh SERS Sensor Via Dielectrophoretic Alignment of Solution‐Processed Au Nanoparticle‐Decorated TiO2 Nanowires

AbstractSurface enhanced Raman spectroscopy (SERS) is becoming an attractive analytical technique for the next generation of breath diagnostics. However, current SERS substrates present challenges related to fabrication cost, complexity, signal uniformity, and reproducibility. Here, a low‐cost, label‐free SERS sensor based on fully solution‐processed decoration of TiO2 nanowires is demonstrated (NW) with plasmonic Au nanoparticles (NP) followed by the dielectrophoretic self‐assembly into a 3D mesh with high signal to noise ratio. The sensor performance is tested using 4‐aminothiophenol (4‐ATP) as a model analyte in gas phase, at concentrations down to 10 ppbv, and in solution, with limit of detection ≈2.4 pM. Finally, to explore the sensor capability for breath‐based diagnostics, a proof‐of‐concept experiment is performed with exhaled breath condensates (EBCs). The possibility to discriminate EBCs of individuals with upper respiratory tract infection (URTI) from healthy ones is demonstrated. Multiple SERS spectra (n≈50) from each sample are analyzed using orthogonal partial least squares discriminant analysis (OPLS‐DA), which identifies spectral features representative of URTI in up to 80% of the infection‐related spectra. These results demonstrate the applicability and potential of 1D nanomaterials together with state‐of‐the‐art solution‐processed techniques for the development of low‐cost and compact SERS breath‐based diagnostic platforms for clinical point‐of‐care applications.

3DPMesh: An enhanced and novel approach for the reconstruction of 3D human meshes from a single 2D image

Currently, virtual reality, the gaming industry, cloth parsing, surveillance and security systems, and avatars have gained much interest in computer vision. To fulfil the growing demand in this field, an accurate 3D human mesh plays an important role. We proposed a novel approach, 3DPMesh, to reconstruct more accurate 3D meshes for humans from a single 2D image. 3DPMesh stands for 3DPose estimation, 3DPose enhancement and Mesh articulation. We achieve enhanced results of 3D meshes with the ultimate fusion of these three modules. The depth predictor regresses the 3D keypoints of the human body, whereas the pose alignment enhances the accuracy of the predicted 3D joint locations. Finally, the mesh articulation module articulates the parametric mesh so that it accurately matches the 3D pose of a person. The proposed method offers more accurate 3D human meshes and better alignment to the content of an input image when compared to the state-of-the-art methods on the challenging UP-3D, Human3.6M, and 3DPW datasets. Additionally, there is no need for a training phase in the proposed model, therefore, it does not require an extensive collection of datasets to train the model, so it saves a lot of computational time and ultimately generates a 3D mesh in less inference time. The experimental results and analysis of the proposed method indicate that it advances the mesh fitting state-of-the-art methods at any scale variation.

Research and Application of Key Technologies for Assembling Design and Construction of Reinforcement in Super High Concrete Bridge Tower

Concrete bridge towers are widely used in long-span bridges. Aiming at the reinforcement process which requires the largest labor force and has the highest safety risk during the construction of the super high concrete bridge tower. The key technology of assembling design and construction of super high concrete bridge tower reinforcement characterized by “industrialized production of mesh, rapid section forming and on-site assembling installation” is creatively proposed. In terms of design, the three-dimensional bridge tower reinforcement is discretized into planar or 3D mesh. In order to facilitate the reinforcement segment, a new flush lapping is used for the circumferential connection of stirrup, and the tapered sleeve splicing is used for the connection of the reinforcement parts between the segments, and the reliability of the connection is verified by experiments. In terms of construction, industrial assembly lines are used for the production of steel mesh and overall bending forming. The bridge tower reinforcement segments are assembled on the special jig, and finally installed on the tower by the tower crane. At present, this technology has been successfully applied to projects such as Lingdingyang Bridge, reducing labor intensity, improving construction efficiency by 30%, and ensuring project quality.