29 results about "Marching cubes" patented technology

The present invention relates to the field of rock structure measurement, and disclosed are a three-dimensional pore reconstruction method and system based on a rock image, for use in solving the problem that using a machine learning method to reconstruct pores in a rock image requires relatively high mathematical or computer science expertise, involves high labor and economic costs, and is difficult to apply to small-scale rock pore reconstruction projects. The method comprises: step 1: grouping slice images of each rock sample into one set, and performing preprocessing; step 2: performing grayscale processing, histogram equalization and normalization processing on an image; step 3: using a Harris corner detection algorithm to perform corner detection, sorting Harris response values, selecting key points by setting the number of key points, and using a non-maximum suppression method to screen the selected key points; step 4: segmenting the boundaries of pores in the image by means of an adaptive threshold selection method, and extracting pore features; and step 5: using a marching cubes algorithm to construct a three-dimensional pore reconstruction model. The present invention is used for three-dimensional reconstruction of pores in a rock image.

The invention discloses a three-dimensional visualization head and neck tumor preoperative imaging system based on deep learning. The method comprises the following steps: image analysis and preprocessing; carrying out image denoising and spatial-temporal feature extraction based on a deep learning convolutional neural network architecture, and introducing a u-Net model to realize image segmentation and semantic annotation; performing three-dimensional reconstruction and surface rendering by adopting a Marking Cubes algorithm, and realizing visual interaction image presentation and personalized operation by combining volume rendering and surface rendering; and data is output through multiple ports, and data viewing and interactive operation of a web terminal and a mobile terminal are supported. The method has the advantages that according to the technical scheme, through fusion of a three-dimensional attention mechanism and multi-phase data, the tumor segmentation Dice coefficient reaches 0.92 + / -0.03, and the method is superior to a traditional threshold segmentation method.

The invention discloses an ocean terrain digital display method based on three-dimensional modeling. The ocean terrain digital display method comprises the following steps of 1, constructing a multi-source ocean data set; 2, performing space fusion by adopting a self-adaptive interpolation method based on position weight to generate a three-dimensional space density field; 3, performing contour surface extraction by adopting an improved Marching Cubes algorithm to generate an initial three-dimensional terrain model; 4, performing local modeling by adopting a Delaunay triangulation network reconstruction method, and constructing a complete three-dimensional terrain model; 5, obtaining a texture terrain model; step 6, constructing an LOD tree structure and realizing dynamic loading in combination with view cone cutting and a screen space error control strategy; and step 7, based on a dynamic loading result of the LOD tree structure, driving node switching through a time controller, and performing visual display. According to the method, three-dimensional modeling and an LOD tree structure are fused, and high-precision reconstruction and evolution display of the ocean topography are achieved.

The invention relates to the technical field of medical image processing and three-dimensional reconstruction, and particularly discloses a reconstruction and analogue simulation method based on a cardiac image. The method comprises the following steps: firstly, preprocessing an input CT or MRI medical image, and automatically segmenting a heart multi-tissue structure by using an nnUNetv2 model, including a plurality of anatomical regions such as atrium, ventricle, myocardial tissue and blood pool; and then performing topology repair, smoothing processing and label resampling on the segmentation result to obtain a three-dimensional label body with a continuous structure and a complete boundary. And based on the optimized tag body, constructing a heart three-dimensional surface model by adopting a Marching Cubes algorithm, and generating a multi-material volume grid suitable for finite element analysis. According to an electrode position and a radio frequency parameter set by a user, a radio frequency ablation simulation model based on a Pennes biological heat conduction equation and an Arrhenius model is established, a temperature field and a tissue thermal damage range are calculated, and a simulation result is displayed in a three-dimensional mode in an overlapping mode. According to the method, the integrated process from medical image automatic segmentation to heart three-dimensional reconstruction and thermal simulation is realized, the segmentation precision, the modeling efficiency and the simulation reliability are improved, and the method can be applied to application scenes such as surgical planning, preoperative evaluation and medical teaching.

The application discloses an analysis system and method based on first principle and quantum oscillation test, and the analysis system comprises a data input module, a preprocessing module, an operation module and an output module, wherein the operation module comprises an extreme value orbit visualization unit used for visualizing the position of an extreme value orbit in a Fermi surface, an extreme value curve visualization unit used for outputting a frequency change diagram of the corresponding extreme value orbit in a calculation process, and a Fermi surface selected area calculation unit used for effectively improving the calculation efficiency for a complex Fermi surface. The application can quickly and accurately screen and determine the quantum oscillation frequency and effective mass information of relevant materials based on the marching cubes, slice-to-slice orbit matching in SKEAF and hdbscan algorithm, and further determine the Fermi surface characteristics of the materials, which has important significance for the research on the electronic structure and physical properties of quantum materials.

The invention provides a complex tubular structure modeling method based on 2D segmented algebraic splines, and relates to the technical field of geometric modeling and computer graphics. The method comprises the following steps: firstly, constructing a section control polygon and generating a high-order algebraic spline primary function; calculating a Frenet frame corresponding to the guide curve, and mapping the section control polygon along the curve to generate a multi-section structure; and constructing an implicit field based on the parameters, and extracting contour surfaces by using a Marching Cubes algorithm to generate a triangular mesh model. The method is small in number of control points, high in structural continuity and suitable for efficient modeling and batch generation of complex tubular structures.

The application discloses a brain glioma segmentation and three-dimensional visualization model training method and system based on multi-modal fusion, and comprises the following steps: inputting multi-modal medical image data into a Laplacian pyramid multi-modal fusion learning model to generate a fusion graph, then inputting the fusion graph into a U-net segmentation model comprising an encoder and a decoder to obtain a segmentation mask under the multi-modal fusion graph; and finally, based on the multi-modal fusion graph and the segmentation mask, adopting a Marching cubes (MC) algorithm to perform three-dimensional reconstruction visualization of brain regions and brain tumor region labeling. The brain glioma segmentation and three-dimensional visualization model training method and system based on multi-modal fusion provided by the application optimizes the disadvantages of insufficient information and incomplete modeling under single modal, fully utilizes the advantages of the completeness of multi-modal data, performs multi-modal fusion, segmentation and three-dimensional visualization, can more comprehensively assist surgeons in treatment, and improves the accuracy in the operation process.

The invention relates to a medical image processing and three-dimensional reconstruction simulation system. The medical image processing and three-dimensional reconstruction simulation system comprises image loading, image processing, three-dimensional reconstruction, simulation analysis and multi-view display. The system loads various medical image formats, provides real-time display of axial, sagittal and coronal slices, extracts information of key parts by using an image segmentation strategy, generates a high-precision three-dimensional model based on a Marching Cubes algorithm, supports thermal and electrophysiology analysis, can automatically input physiological parameters, simulates heat distribution and current distribution of tissues, supports interactive adjustment, and is high in accuracy and high in accuracy. And a user can conveniently observe image details and analysis results. By using the system, the medical image processing efficiency and the three-dimensional reconstruction precision can be improved, meanwhile, the multi-dimensional analysis requirement of a complex anatomical structure is met through a simulation analysis function, and the system can be widely applied to medical diagnosis, surgical planning and medical teaching scenes.

The application discloses a kind of tropical cyclone temperature body structure calculation method and system, wherein the method includes: based on satellite microwave detection data and tropical cyclone position information, the three-dimensional temperature data of tropical cyclone area is smoothed, and temperature grid with tropical cyclone center as origin is formed.Three-dimensional temperature anomaly value is calculated according to the smoothed temperature grid.Based on three-dimensional temperature anomaly value, three-dimensional region growth method is used to extract tropical cyclone warm core main area.Based on the extracted tropical cyclone warm core main area, the isosurface is calculated using Marching Cubes algorithm.According to the isosurface, the light model is rendered to draw the three-dimensional visualization structure of tropical cyclone warm core.Based on the three-dimensional temperature anomaly value of tropical cyclone warm core main area, the tropical cyclone warm core intensity index is calculated.Thus, the application effectively solves the problem of insufficient overall understanding of tropical cyclone temperature structure in the prior art, and provides the ability of systematic quantitative analysis.

The invention relates to the technical field of tunnel face surrounding rock grading in railway construction, and provides a drilling and blasting method tunnel face surrounding rock three-dimensional refined grading method based on a seismic wave reflection method, which comprises the following steps: collecting three-dimensional inversion dot matrix data of an area in front of a tunnel face; the data coordinate system is converted into a tunnel design coordinate system, the surrounding rock basic quality index BQ value of each point is calculated according to the wave velocity, and a space BQ value dot matrix database is generated; performing spatial region division on the BQ value dot matrix by adopting a three-dimensional region self-growth method to obtain a connected region with similar BQ value characteristics; after the representative value of the region boundary is determined, using a Marking Cubes algorithm to extract contour surfaces to construct a surrounding rock three-dimensional grid model; and finally, mapping the model vertex BQ value with a preset grading standard, and outputting a three-dimensional refined grading model with a surrounding rock grading label. Automatic processing and three-dimensional visualization of surrounding rock classification are achieved, and classification accuracy and construction decision reliability are improved.

The invention discloses a non-rigid object three-dimensional reconstruction method based on a single-view neural radiation field, and the method comprises the steps: firstly inputting a single RGB image, extracting global features through a pre-trained deep learning backbone network, and generating a mask image of an object in combination with a semantic segmentation network; secondly, predicting volume density and color distribution by adopting a neural radiation field variant pixelNeRF, embedding a dynamic deformation field and a parameterized template to constrain shape rationality, and solving the problem of shape ambiguity in single-view reconstruction; thirdly, extracting a grid model through a Marking Cubes algorithm, complementing shielding and details by using a reconstruction algorithm, and enhancing the sense of reality of the surface by using a texture super-resolution technology; and finally, providing a user interaction interface for local optimization, and correcting a motion posture through a differentiable physical engine to generate a drivable three-dimensional animation model. By combining the single-view neural radiation field and the dynamic deformation modeling technology, the method has the advantages of high precision, high efficiency, support of interactive editing, capability of generating driving animations and the like.

The application discloses a multi-radar power envelope fusion method, comprising: single-radar envelope modeling, constructing envelope fusion foundation; determining non-coherent field fusion, constructing multi-radar envelope model; constructing gradient diffusion model, realizing envelope fusion smooth transition; displaying Euler method iteration convergence, realizing multi-radar envelope complete fusion; adopting adaptive octree structure to dynamically divide three-dimensional space. Advantageous effects: the multi-radar power envelope fusion method firstly calculates single-station detection probability of each radar unit at a grid point in a three-dimensional space voxel according to a radar equation, a propagation loss model and target scattering characteristics; then adopts a probability union or a more complex dependent relationship model to generate a full-space joint detection probability field; finally, an isosurface extraction (such as an improved Marching Cubes algorithm) and a probability gradient driven color / transparency mapping are adopted to generate a visual continuous, mathematically accurate fusion envelope surface.

This invention discloses a method and system for constructing a mesh model based on the spatial parameters of a 3D Gaussian surface model, belonging to the field of 3D reconstruction technology. It includes at least the following steps: establishing an initial 3D Gaussian elliptical surface model based on the sparse point cloud of the reconstructed object; constructing a local implicit function field for each Gaussian elliptical surface model; for a single Gaussian elliptical surface model, the implicit function value of a vertex is the signed distance from the voxel vertex to the corresponding Gaussian elliptical surface; performing global weighted fusion based on the local implicit function field of the current Gaussian elliptical surface model; and extracting 0-isosurfaces using the Marching Cubes algorithm based on the vertex function values ​​of the global voxel field, thereby constructing the mesh model of the reconstructed object. This invention establishes a voxel field consistent with the distribution of the 3D Gaussian elliptical surface through the 3D spatial parameters of the 3D Gaussian elliptical surface model, enabling accurate and unobstructed surface representation and achieving high-precision 3D reconstruction of the mesh model.

The application provides a shadow-aware satellite image neural radiance field ground surface three-dimensional reconstruction method, comprising: optimizing an RPC model based on multi-temporal satellite images and generating a sparse point cloud, and constructing a unified coordinate system through an outer enclosing sphere normalization; sampling three-dimensional points along a pixel line-of-sight direction and constructing a training set; an SDF network predicts a signed distance value of a spatial point, a color network fuses SDF features and a line-of-sight direction to output color, and a shadow color network generates a shadow vector in combination with a sun direction; a ray color is predicted through volume rendering integration, a shadow probability is calculated based on twice ray casting of the sun direction, a loss function is constructed in combination with an irradiance error to jointly optimize network parameters; and a three-dimensional grid is generated by using a Marching Cubes algorithm after sampling points in the unified coordinate system by using the trained SDF network. The application can improve the convergence stability of a neural radiance field model on satellite data and the success rate and precision of scene three-dimensional surface reconstruction.

Disclosed herein are an apparatus and method for compressing 3D volume data. The apparatus for compressing three-dimensional (3D) volume data includes one or more processor, and execution memory configured to store at least one program that is executed by the one or more processors, wherein the at least one program is configured to receive 3D volume data for data compression, select necessary voxels that contribute to formation of a geometric surface from the 3D volume data using a marching cubes algorithm, represent location information of the selected necessary voxels by a tree structure, and binarize the location information of the necessary voxels by serializing the tree structure, and quantize truncated signed distance values to surfaces adjacent to the necessary voxels constituting a volume from the 3D volume data using the binarized location information of the necessary voxels.

The invention relates to the technical field of tunnel advanced geological forecast, and discloses a railway tunnel construction tunnel face front surrounding rock water abundance three-dimensional model construction method comprising the following steps: collecting a plurality of transient electromagnetic survey lines in a tunnel face direction, and uniformly mapping two-dimensional inversion results to a right-handed coordinate system according to survey line angles, mileages and survey point distances; and constructing a spatial water-rich database. After geometric parameters of the tunnel face are obtained, a three-dimensional cube grid is established, and apparent resistivity values are given to nodes. A kd tree is constructed based on node distribution, a normal vector and curvature are calculated, and a mark array is initialized. And completing node clustering by adopting a three-dimensional region self-growth algorithm in combination with attribute difference, a region threshold and curvature constraint to form a water-rich region. And a binary volume data field is constructed for each clustering region, contour surfaces are extracted by using Marching Cubes, and a surrounding rock water yield three-dimensional model is generated. According to the method, high-precision three-dimensional modeling of the water-rich area in front of the tunnel face is effectively constructed, and the accuracy of advanced prediction of the tunnel is remarkably improved.

The invention discloses a multi-radar power envelope fusion method, which comprises the steps of performing single-radar envelope modeling, and constructing an envelope fusion foundation; incoherent field fusion is determined, and a multi-radar envelope model is constructed; constructing a gradient diffusion model to realize envelope fusion smooth transition; euler method iterative convergence is displayed, and multi-radar envelope complete fusion is realized; and dynamically dividing a three-dimensional space by adopting a self-adaptive octree structure. The multi-radar power envelope fusion method has the beneficial effects that firstly, according to a radar equation, a propagation loss model and target scattering characteristics, the single-station detection probability of each radar unit at a grid point is calculated in a three-dimensional space voxelization mode; a probability union set or a more complex dependency relationship model is adopted, and a total-space joint detection probability field is generated through fusion; finally, through contour surface extraction (such as an improved March i ng Cubes algorithm) and color / transparency mapping driven by probability gradient, a fusion envelope curved surface with continuous vision and accurate mathematics is generated.

This invention discloses a three-dimensional thermal environment neural field reconstruction method and system for diagnosing building facade defects, belonging to the field of building energy conservation technology. It includes: simultaneously acquiring multi-view image data using an integrated thermal infrared and visible light dual-modal device; achieving cross-modal camera pose calibration using an improved Structure for Motion Restoration (SfM) algorithm; constructing an implicit thermal radiation model using neural radiation fields (NeRF) to introduce temperature information and enhance thermal environment characterization; and generating an explicit triangular mesh model with both geometric details and thermal properties using the signed distance function (SDF) and Marching Cubes algorithm. The invention's multi-modal neural radiation field fusion framework complements thermal infrared and visible light data to achieve three-dimensional thermal environment reconstruction. This method can be widely applied to building facade safety diagnosis, energy-saving renovation, and low-carbon operation and maintenance.

The invention discloses a three-dimensional thermal environment neural field reconstruction method and system for building facade defect diagnosis, and belongs to the technical field of building energy saving. The method comprises the following steps: synchronously acquiring multi-view image data through an integrated thermal infrared and visible light bimodal device, and realizing cross-modal camera pose calibration by using an improved motion recovery structure (SfM) algorithm; the method comprises the following steps: constructing a thermal radiation implicit model by using a neural radiation field (NeRF), introducing temperature information to enhance thermal environment representation, and generating an explicit triangular mesh model with both geometric details and thermophysical properties through a signed distance function (SDF) and a Marching Cubes algorithm; according to the multi-mode neural radiation field fusion framework, thermal infrared data and visible light data are complementary, and thermal environment three-dimensional reconstruction is achieved. The method can be widely applied to safety diagnosis, energy-saving transformation and low-carbon operation and maintenance of building facades.

The invention belongs to the technical field of three-dimensional modeling, and particularly relates to an unsupervised non-watertight point cloud high-fidelity reconstruction method. Aiming at the problems of inconsistent gradient directions, local detail loss, artifact fragmentation and the like in open surfaces, non-watertight point clouds and complex geometric scenes in the existing method, the invention provides a normalization and density adaptive resampling mechanism; symmetric point pairs are generated through local normal estimation, and prior constraints with consistent UDF values and opposite gradients are applied; designing a multi-scale Gabor code with spatial locality and direction selectivity, and combining the combined feature input of a symmetric invariant base and a symmetric variant base; gradually converging from global topology to local details by adopting progressive multi-scale sampling and composite unsupervised loss function optimization; and a UDF curved surface extraction method based on gradient discrimination and improved Marking Cubes is provided.

The application discloses a kind of marine topography digital display methods based on three-dimensional modeling, comprising the following steps: step one: constructing multi-source marine data set;Step two: using the adaptive interpolation method based on position weight to carry out space fusion, generate three-dimensional space density field;Step three: using the improved Marching Cubes algorithm to carry out isosurface extraction, generate initial three-dimensional terrain model;Step four: using Delaunay triangulation reconstruction method to carry out local modeling, build complete three-dimensional terrain model;Step five: obtain texture terrain model;Step six: construct LOD tree structure and realize dynamic loading in combination with view frustum clipping and screen space error control strategy;Step seven: based on the dynamic loading result of LOD tree structure, through time controller drive node switching, carry out visual display.The application fuses three-dimensional modeling and LOD tree structure, realizes marine topography high-precision reconstruction and evolution display.

A kind of orthopedic surgery navigation registration method based on UWB marker point, UWB module is used as calibration point around the patient's affected part to calibrate, DICOM data is obtained by scanning medical imaging equipment, convolutional neural network is combined with integrated learning algorithm Xgboost to pre-process medical imaging data, finally, three-dimensional reconstruction is carried out by surface rendering Marching Cubes method, the spatial coordinate system of calibration point is obtained by UWB positioning algorithm, according to the number of registration points in actual spatial coordinate system and virtual spatial coordinate system, the optimal transformation matrix is found, so that the transformed coordinates are closest to virtual coordinate system coordinates.Translation transformation matrix and rotation transformation matrix are solved by SVD matrix decomposition method, scale transformation parameter is solved by quaternion algorithm combined with homogeneous coordinate theory, when marker point moves, repeat the steps to complete real-time registration and registration based on UWB marker point.Avoid the movement of intraoperative non-active factors and need to recalibrate.

The application discloses a kind of turning processing workpiece surface three-dimensional topography prediction method, comprising the following steps: step one: the Marching Cubes contour surface of target surface is constructed: according to the range of target surface, constructs a cubic grid;The value of implicit function is operated on each grid point in cubic grid;According to the sign of implicit function value, based on Marching Cubes algorithm, generate triangular patches and the corresponding vertex value of triangle;Step two: the target surface is uniformly randomly sampled: according to the established triangular contour surface grid, uniformly randomly in each triangular grid inside sampling point is obtained, to realize the uniform random sampling of target surface;Step three: three-dimensional topography prediction is carried out to target surface: based on the tool position data of complex free surface path planning, tool path is modeled, and residual height on each random sampling point is calculated, realizes three-dimensional topography prediction to target surface.

A gingival three-dimensional model shell extraction method and device based on the Marching Cube algorithm, the method comprising: (1) importing an initial gingival three-dimensional model and performing relevant pretreatment; (2) performing extrusion offset on non-point; (3) constructing a signed distance field by adopting a grid diffusion + hierarchical construction strategy; (4) deleting redundant repeated vertices and corresponding face sheets in the model; (5) traversing the vertices of the inner shell gingival model in the manifold, and removing topological errors; (6) stitching the bottom surface of the height compensation gingival model to obtain a closed outer shell gingival model; (7) combining the standard inner shell gingival model and the standard outer shell gingival model together to obtain an open shell gingival model; and (8) obtaining a complete and closed standard shell gingival model. The present application provides effective pretreatment and post-treatment processes before and after Marching Cube reconstruction, accurately repairs the construction defects of the Marching Cube result, and finally generates a standard shell gingival model without error when imported into CAD printing software, and the mesh quality is uniform and healthy.

In order to obtain high-precision industrial CT isosurface mesh for reverse engineering, the present application proposes a surface extraction method based on dual octree of adaptive volume data of industrial CT, directly constructs adaptive octree volume data grid from industrial X-ray computed tomography projection graph sequence, and based on this, high-precision and crack-free interpolation isosurface mesh is obtained. In terms of extracting isosurface mesh, the Marching Cubes method is easy to generate ambiguous surfaces with aliasing and oscillation around the accurate surface, and such methods based on CT discrete volume data cannot avoid the influence brought by CT reconstruction artifacts and discretization errors. The method firstly recursively constructs adaptive volume data grid, the voxel value at the node is obtained by reconstructing the projection graph by FDK algorithm, then the vertex is calculated and the dual grid is constructed by QEFs technology, finally the surface profile is accurately extracted by using Marching Cubes lookup table.

The invention discloses a tropical cyclone temperature body structure calculation method and system, and the method comprises the steps: carrying out the smoothing of the three-dimensional temperature data of a tropical cyclone region based on satellite microwave detection data and tropical cyclone position information, and forming a temperature grid with the center of the tropical cyclone as an original point. And calculating a three-dimensional temperature anomaly according to the smoothed temperature grid. And based on the three-dimensional temperature anomaly, a three-dimensional region growing method is adopted to extract a tropical cyclone warm nucleus main body region. And on the basis of the extracted tropical cyclone warm nucleus main body region, a Marching Cubes algorithm is utilized to calculate a contour surface. And according to the contour surface, rendering by using an illumination model to draw a three-dimensional visual structure of the tropical cyclone warm nucleus. And calculating a tropical cyclone warming core strength index based on the three-dimensional temperature anomaly of the tropical cyclone warming core main body area. In this way, the problem that in the prior art, overall cognition on the tropical cyclone temperature structure is insufficient is effectively solved, and the systematic quantitative analysis capacity is provided.

The application relates to the field of point cloud generation and discloses a diversity shape completion method based on component analysis and decomposition, which comprises the following steps: representing an object as an abstract template structure, predicting and matching template structure parameters of a defective point cloud through a deep learning network, and deducing a complete structure representation of the object; performing component decomposition on the object, obtaining projection details of internal points of each component on a plane, inputting the projection details into a network, and generating solid details of each component; performing three-view completion, combining the details of symmetrical components, deducing complete component details of the object, comparing the three views of the defective and complete objects to identify a missing area, judging whether diversity generation is needed, masking the area needing diversity generation in the complete detail map, performing image-to-image diffusion generation, and obtaining diversified complete detail three views; reconstructing a three-dimensional area according to the three views, obtaining a corresponding signed distance field, and obtaining a watertight completion result through a marching cubes algorithm.

The invention discloses a preoperative focus recognition system based on multi-modal image fusion, and belongs to the technical field of medical image processing and computer-aided surgery. The system comprises an OCT image modeling module which is used for reconstructing an OCT three-dimensional tissue volume model through an FFT (Fast Fourier Transform) method, structure tensor enhancement and a Marching Cubes algorithm; the microscope image registration module adopts SIFT (scale invariant feature transform) feature extraction, K-D tree matching, RANSAC (random sample consensus) and ICP (inductively coupled plasma) optimization and Gamma correction to realize sub-pixel-level space and luminosity registration of a microscope image and an OCT (optical coherence tomography) model; the multi-modal fusion modeling module is used for adaptively fusing OCT structural features and microscope texture features by using a dual-channel deep neural network and a cross-modal attention mechanism; the focus segmentation and positioning module is used for realizing accurate segmentation and three-dimensional positioning of the focus by adopting an improved U-Net network and a joint loss function; and the preoperative planning module automatically generates an operation point and an orbit injection scheme of a surgical instrument through ellipsoid fitting and a genetic algorithm. The lesion recognition precision and the operation planning efficiency are remarkably improved.

The invention discloses a magnetic combined aluminum formwork and a special-shaped curved surface generation method and a special-shaped curved surface construction method.The magnetic combined aluminum formwork comprises 18 kinds of unit modules which can be spliced to form any special-shaped curved surface and are consistent in modulus size, and the shapes of the 18 kinds of unit modules correspond to 15 kinds of shapes specified by a Marching cubes algorithm; a spherical magnet used for being attracted to other unit modules is movably arranged in each right angle of the 18 kinds of unit modules. Any curved surface shape is fitted through combination and splicing of 20 kinds of unit modules, the special-shaped concrete template is suitable for pouring of various special-shaped concrete structures, the manufacturing process is simple, the construction period is short, and the tedious procedure of a traditional customized curved surface template is omitted; all unit modules can be completely reused, so that resource waste is avoided; the modulus size of the unit modules is adjusted according to requirements, the precision of any curved surface shape formed by combining and splicing the unit modules can be controlled, and the smaller the modulus size is, the higher the precision is.