78 results about "Digital reconstruction" patented technology

The invention discloses a digital human rendering method based on Gaussian splashing and multi-scale characteristic field distillation, and belongs to the field of three-dimensional human body digital reconstruction. According to the method, feature extraction is carried out through a Vision Transformer encoder, based on an SMPL model, through a cross-modal parameter estimation module and dynamic human body modeling of three-dimensional Gaussian splashing and semantic feature rendering, three-dimensional Gaussian is projected to a two-dimensional image plane to calculate a covariance matrix and color mixing, and after a rendered color image and an initial feature field image are output, a three-dimensional image is obtained. A student feature map is obtained through a convolution acceleration module, a teacher feature map is obtained after feature extraction is carried out through a two-dimensional basic model, the constructed model is trained, and optimization training is completed through comprehensive total loss function calculation; according to the method, the problems of fuzzy semantics, detail missing, low rendering efficiency and inaccurate human body-scene separation in the existing method are effectively solved, so that more efficient, fine and robust monocular or multi-view human body three-dimensional reconstruction is realized.

The invention belongs to the technical field of customer service, and particularly relates to an intelligent customer service practical training system and method, and the system comprises a business scene digital reconstruction layer which is used for constructing a business training scene to form a training scene library, managing the business training scene library, and reconstructing a training scene according to scene configuration data; the AI simulation interaction core layer is used for receiving and recognizing voice and text input of students and generating behavior strategies of AI partner training according to input content on the basis of a current training scene; the training execution layer is used for generating scene configuration data according to a student selection instruction and sending the scene configuration data to the business scene digital reconstruction layer, and the scene configuration data content comprises a training scene, training difficulty and role setting; the training difficulty is dynamically adjusted according to the real-time performance of the trainee; and the quantitative evaluation layer is used for performing multi-dimensional analysis on the training data and generating a quantitative evaluation result and an improvement suggestion. The problems that existing customer service training is low in efficiency and high in cost are solved.

The invention provides a parallel battlefield environment construction method and system based on digital twinning. The method comprises the steps that a laser echo set reflected by the earth surface is obtained; aggregating the intensity characteristic of each laser echo in the laser echo set and the time difference between the departure time and the arrival time of each laser echo, and generating point cloud data; performing space coordinate association on a three-dimensional elevation model constructed based on the point cloud data and a preset battlefield terrain material feature map to generate a battlefield comprehensive terrain entity grid; performing spatial feature analysis on the battlefield comprehensive terrain entity grid to generate a dynamic threat distribution representation; performing spatio-temporal evolution deduction on the dynamic threat distribution representation to generate enemy unit maneuvering path prediction data; and inputting the enemy unit maneuvering path prediction data into a preset digital twinborn battlefield environment to drive the preset digital twinborn battlefield environment to dynamically update. According to the invention, accurate digital reconstruction of the complex battlefield terrain is realized, and a real-time and high-confidence battlefield situation deduction platform is provided for command decision.

The invention discloses a portable X-ray back scattering security check system based on AI image enhancement. The system belongs to the technical field of radiation imaging, and comprises a sensing unit used for sensing physical situation information in an imaging process so as to generate time-sequenced physical prior data; the prediction unit is used for predicting a future scanning behavior based on the physical prior data to generate a feed-forward control instruction and prediction confidence; the regulation and control unit is used for adaptively regulating and controlling scanning parameters based on the feedforward control instruction and recording execution information including regulation and control residual errors; and the AI reconstruction processor is used for performing enhanced reconstruction on the original X-ray image based on the physical priori data, the prediction confidence and the execution information. By constructing an intelligent closed loop of perception, prediction, regulation and control and reconstruction, physical uncertainty information in the imaging process is quantized and used for guiding AI reconstruction, so that the problem of information fault between physical acquisition and digital reconstruction is solved, and the quality and reliability of a final image are improved.

The invention discloses a three-dimensional point cloud-based power distribution network digital reconstruction method, which comprises the following steps of: automatically acquiring high-precision point cloud and image data of a power distribution network site through an unmanned aerial vehicle, and eliminating noise points, simplifying data volume and improving data quality in combination with a gross error elimination algorithm and a voxel filtering technology; a pyramid type hierarchical storage and octree indexing technology is utilized to realize regional management and rapid space retrieval of massive point clouds; and performing automatic semantic classification on the point cloud based on a PointNet deep learning algorithm, performing high-precision alignment on a classification result and a real scene in combination with a homonymy matching method, and constructing a real and restored power distribution network field three-dimensional working condition model. According to the method, the problems of low field investigation efficiency and incomplete information of a traditional power distribution network are solved, digital and automatic investigation of non-power-cut operation of the power distribution network is realized, transformation from operation and maintenance of the power distribution network to scientific and technical transformation is promoted, manpower and material resources are saved, and power supply reliability and operation and maintenance management efficiency are remarkably improved.

The invention discloses an existing bridge high-precision three-dimensional reconstruction method based on laser point cloud data, and relates to the technical field of bridge engineering digital reconstruction. Multi-source laser point cloud cooperatively collects detail point cloud of key components of an existing bridge, point cloud of a bridge floor and an upper structure and point cloud of a shielding part; multi-source point cloud coarse registration is completed based on a target ball of a preset specification, and then fine registration is carried out by adopting an improved iterative nearest point algorithm of integrated normal vector constraint; and adopting a layered denoising strategy to carry out statistical filtering denoising on the flat area, and carrying out adaptive radius filtering on the vulnerable area to reserve cracks and peel off disease points. Through collaborative design of multi-source acquisition, precise processing, intelligent segmentation, parameterized modeling and hierarchical verification, data integrity improvement, consideration of point cloud processing precision and disease retention, semantic segmentation and instance identification precision optimization, BIM model engineering value improvement and achievement adaptability enhancement are realized.

The invention relates to the technical field of three-dimensional digital reconstruction, in particular to an archaeological site reconstruction method based on image point cloud cross-modal alignment, and the method comprises the steps: obtaining an optical image of an archaeological site and archaeological site point cloud data; obtaining a point cloud distribution characterization value based on the archaeological site point cloud data, determining a thinning mode of the archaeological site point cloud according to the point cloud distribution characterization value, and carrying out thinning processing based on the thinning mode; performing cross-modal alignment on the thinned archaeological site point cloud data and the optical image of the archaeological site; determining the qualification of the alignment effect according to the cross-modal alignment rate, and determining an adjustment strategy when the alignment effect is unqualified according to the mismatch rate of the key region when the alignment effect is unqualified; and fusing the archaeological site optical image with the qualified alignment effect with the archaeological site point cloud data, and executing dense point cloud reconstruction, three-dimensional grid generation and texture mapping to complete three-dimensional reconstruction of the archaeological site. According to the method, the cross-modal alignment efficiency of archaeological site reconstruction is improved.

The invention discloses an environment situation scanning three-dimensional reconstruction method based on infrared and SLAM, and belongs to the technical field of computer vision and three-dimensional reconstruction. According to the method, thermal radiation information and environment image information of a target environment are obtained through infrared sensing and visual imaging respectively, after feature extraction and matching processing are conducted on the two kinds of information, synchronous positioning and map construction are completed, and environment pose data and environment map data are generated; a sensing uncertainty probability model is established, after multi-source data denoising is completed through Bayesian filtering, feature level fusion is carried out to obtain fusion environment data, and finally a three-dimensional reconstruction model of a target environment is constructed based on the fusion environment data. According to the method, the full-process closed loop of environment situation scanning and three-dimensional reconstruction is realized, the application limitation of single sensing information is made up, the comprehensiveness of environment sensing data and the scene adaptability of three-dimensional reconstruction are improved, and the digital reconstruction requirements in various complex environments can be met.

The invention provides a construction method of a virtual electromagnetic environment model and a test device. The construction method comprises the following steps: acquiring multi-dimensional test data of a mobile carrier in a target scene; taking a timestamp of the baseband signal data as a reference, performing synchronous fusion on the positioning data and the attitude data, and generating a synchronous data set; performing parameter estimation on the synchronous data set, and separating a relative channel parameter of the multipath component in a mobile carrier coordinate system; converting the relative channel parameters into absolute space parameters under a geographic coordinate system based on the instantaneous space state, and determining physical scattering source characteristics corresponding to the multipath component based on the relative channel parameters and the absolute space parameters; and generating a virtual electromagnetic environment model based on the absolute space parameters and the physical scattering source characteristics. In the mode, a structured virtual electromagnetic environment model containing absolute space attributes can be generated, and then high-fidelity digital reconstruction and reproduction of a complex dynamic electromagnetic environment in a real scene are realized.

The invention belongs to the field of geological monitoring, and discloses a loess macropore seepage observation device and method based on a loess digital reconstruction body, and the loess macropore seepage observation device comprises the loess digital reconstruction body, a coloring agent container and a water discharging bottle; the bottom of the loess digital reconstruction body is provided with a permeable stone layer, and the top is provided with a hermetically connected top cover; a plurality of image acquisition devices are arranged around the loess digital reconstruction body; one sides of the drainage bottle and the permeable stone layer are communicated with the bottom of the coloring agent container; the first end of the three-way pipe is communicated with the interior of the top cover, the second end of the three-way pipe is provided with a pressure meter, and the third end of the three-way pipe is communicated with the atmosphere. The loess digital reconstruction body duplicates the macropore structure of the loess sample, the water flow migration process in the macropores of the loess digital reconstruction body can be rapidly observed, the water-driven gas process in the macropores of the loess digital reconstruction body can be rapidly determined, the working efficiency is high, the device is flexible and detachable, and the cost is low. The problem that the three-dimensional migration process of the water flow in the unsaturated loess macropores is difficult to accurately capture at present can be effectively solved.

The invention discloses a quantitative analysis method for ardealite leachate-carbonate rock fracture corrosion effect, and belongs to the technical field of material analysis. The quantitative analysis method for the ardealite leachate-carbonate rock fracture corrosion effect comprises the following steps: preparing a carbonate sample; preparing ardealite leachate; mounting the sample in a seepage device in a sealing manner, pouring the leachate into the seepage device, and carrying out a seepage test; performing CT scanning on the sample before and after corrosion to obtain a two-dimensional section slice image of the sample, and preprocessing the slice image based on a gray value and threshold segmentation method; and performing three-dimensional digital reconstruction on the preprocessed CT slice data, and calculating pore structure parameters, fracture structure parameters, fractal dimensions and permeability of the sample to complete quantitative analysis. By adopting the quantitative analysis method for the ardealite leachate-carbonate rock fracture corrosion effect, the problem of low analysis accuracy of the ardealite leachate corrosion effect by an existing analysis method can be solved.

The invention relates to a safety assessment method for pressure-bearing special equipment with crack defects based on health monitoring and digital twinning. Various sensors are installed at key parts of the equipment to collect running state data in real time, a parameterized digital reconstruction technology is adopted to establish high-precision twin bodies of the equipment, and stress distribution and crack propagation trend of the structure are automatically simulated and analyzed by means of cloud computing. A two-stage Bayesian optimization algorithm is innovatively used to realize self-correction of simulation result errors, and adaptive accurate mapping of a digital model to physical equipment is realized. And in a crack defect-containing area, an instability critical dimension prediction model is constructed based on an extended finite element method and a failure evaluation diagram technology, and the crack instability critical dimension is dynamically calculated. According to the invention, real-time, efficient and automatic safety assessment and early warning can be carried out on pressure-bearing special equipment, and scientific data support and decision basis are provided for safe operation and maintenance of the equipment.

The present application relates to the technical field of rock mass engineering, and particularly relates to a rock mass fracture intelligent identification and digital reconstruction method, comprising: collecting a rock mass fracture image; manually marking a color block; based on transfer learning, introducing an already-labeled medical image cell membrane sample and a road fracture sample, and jointly training a fracture identification model; inputting the rock mass fracture image into the fracture identification model to segment the fracture, and building a snake convolution network to extract multi-scale features of the image; digitally reconstructing a binary image to obtain a box set and a polyline set data; and reducing the digital data from a pixel scale to an engineering scale. The present application realizes full-process automation from intelligent identification of a rock mass fracture image to digital reconstruction, and can be widely applied to fields such as geotechnical engineering design, slope disaster identification, underground space stability analysis and the like.

Disclosed is a medical imaging conversion method, automatically converting: at least one or more real x-ray images of a patient, including at least a first anatomical structure of the patient and a second anatomical structure of the patient, into at least one digitally reconstructed radiograph (DRR) of the patient representing the first anatomical structure without representing the second anatomical structure, by a single operation using either one convolutional neural network (CNN) or a group of convolutional neural networks (CNN) which is preliminarily trained to, both or simultaneously: differentiate the first anatomical structure from the second anatomical structure, and convert a real x-ray image into at least one digitally reconstructed radiograph (DRR).

The application discloses a kind of based on space sampling consistency determination pipe robot multi-source perception full true simulation test system, belong to pipe detection simulation test technical field, it includes mechanical execution system, space state perception gate and physical trigger system, multi-sensor sampling unit, true value reconstruction system.Mechanical execution system provides the motion reference along the axial direction of pipeline, and allow to introduce controlled nonlinear disturbance, space state perception gate and physical trigger system are used for the arbitration of sampling opportunity when FPGA motion control card, step modulation and physical trigger;Multi-sensor sampling unit is used to execute sampling instruction and generate original data with position label;True value reconstruction system is used to construct the standardized pipeline true value containing position, texture and geometric information;The application realizes high-fidelity three-dimensional digital reconstruction to the inner wall of pipeline by introducing IMU posture latch, micro time delay elastic capture, binocular vision forward variable frequency modulation and reverse geometry correction algorithm.

This application relates to a method and apparatus for digital transformation of ODN facilities. The method includes: obtaining a character array of box names and a character array of port names based on an actual box image; identifying the box under maintenance and installing box tags and fiber optic cable segment tags directly connected to the box; obtaining a port status array based on a box image after all fiber optic cable segment tags have been installed; locating and identifying each installed fiber optic cable segment tag, and obtaining a port name-bound tag object array based on the located and identified box image, port name character array, and port status array; constructing a meter-length pairing structure object array between box ports based on the meter-scale image of the fiber optic cable segment; and constructing a tag pairing relationship between box ports based on the box topology connection table, the meter-length pairing structure object array between box ports, and the port name-bound tag object array. This application allows a single person to quickly assemble tags for boxes or fiber optic cable segments and complete the digital transformation, with high efficiency and low construction difficulty.

The application belongs to the technical field of three-dimensional measurement and digital reconstruction, and specifically discloses a pipeline reconstruction and detection method based on a virtual adapter, which comprises the following steps: a multi-view vision system collects multi-angle images of a pipeline to be detected, reconstructs the pipeline and determines a bending point; the pipeline is aligned and matched with the bending point, and an initial conversion matrix of a measurement and digital model coordinate system is obtained based on an SVD decomposition method; a digital model pipeline accessory is projected to a camera image, and an adjustment matrix is obtained by optimization of minimum pixel sum and maximum gradient; a threshold is set to screen edge points, and a final conversion matrix is obtained by optimization of maximum gradient after projection; the accessory is converted to the measurement coordinate system, and is spliced with the reconstructed pipeline to form a complete model, and the bending point and the accessory feature deviation are compared. The application reconstructs the pipeline with complex accessories such as flanges and valves through initial calibration of coordinate conversion, projection optimization and edge point fine adjustment, restores the pipeline and the accessory features completely, and improves the pipeline detection precision.

This invention provides a method and apparatus for rapid and non-destructive digital reconstruction of cultural relics and ancient buildings. The reconstruction method includes: acquiring an image set of the target object and corresponding depth data; performing quality improvement processing on the depth data and performing pose calculation on the image set to obtain pose data; identifying different feature regions on the surface of the target object based on the image features of the image set, and assigning corresponding fusion weights to the processed depth data according to the characteristics of each feature region; using the image set, the weighted depth data, and the pose data as input data, performing three-dimensional reconstruction of the target object through a reconstruction model, and completing scale calibration during the reconstruction process; and outputting a scale-calibrated three-dimensional model of the target object. The technical problem solved by this invention is that limitations in the performance indicators of object acquisition and processing make it difficult to coordinate and optimize the speed, accuracy, and direct usability of the overall reconstruction method, resulting in less than ideal digital reconstruction results.

Apparatuses and methods can be related to testing signals. An apparatus can be located on a memory die having a probe-based interface. The apparatus can include an input buffer located on the apparatus and configured to receive an input signal, a latch signal, and a buffer reference voltage, characterize the input signal based on the latch signal and the buffer reference voltage, and provide output data comprising a digital reconstruction of the input signal to an output driver located on the memory die.

An image registration method for a surgical robot, comprising: capturing a two-dimensional image of a surgical object and determining first guiding information in the two-dimensional image; acquiring a three-dimensional image of the surgical object and preoperative planning information, and determining second guiding information in the three-dimensional image; calculating first pose information according to the first guiding information and the second guiding information; adjusting the three-dimensional image according to the first pose information, and acquiring a digitally reconstructed two-dimensional image in the adjusted three-dimensional image; calculating the similarity between the two-dimensional image and the digitally reconstructed two-dimensional image; when the similarity meets a preset condition, projecting the preoperative planning information onto the two-dimensional image according to the first pose information; when the similarity does not meet the preset condition, updating the first pose information to obtain second pose information, and adjusting the three-dimensional image according to the second pose information.

A system for the image acquisition and three-dimensional digital reconstruction of human anatomical shapes is described, comprising a plurality of elements for the instant and synchronized acquisition of images and for sending, optionally wirelessly, information associated with the acquired images, a central processing unit for the reconstruction of a three-dimensional digital model starting from this information, a control device which is able to transmit the information associated with the acquired images to said central processing unit and / or to receive information relating to the obtained three-dimensional digital model, as well as at least one unit for transmitting. optionally wirelessly, the information associated with the images acquired by said plurality of elements for the instant and synchronized acquisition of images towards the control device; a process for the image acquisition and three-dimensional digital reconstruction of human anatomical shapes is also described.

This invention discloses an end-to-end 3D digital reconstruction method and system based on traffic accident event streams, relating to the fields of traffic accident processing and digital twin technology. The method includes: acquiring traffic accident video streams, performing semantic analysis, segmenting static traffic elements, and constructing a digital road network space; modeling targets as spatiotemporal nodes using a dynamic graph neural network, extracting 3D motion trajectories through stereo vision, and parsing semantic behavioral labels using human pose estimation; mapping trajectories and behavioral labels to the digital road network space based on a spatiotemporal alignment mechanism to generate a 3D digital reconstruction model; using an evaluation function containing a reconstruction error evaluation function and a physical reliability evaluation function, combined with an objective function, to evaluate the quality of the reconstruction model, and using the evaluation results as feedback signals to adjust model parameters, objective function weights, and trajectory optimization strategies, guiding end-to-end reconstruction learning. This invention can output high-precision, physically consistent 3D accident reconstruction results.

The application discloses a museum collection three-dimensional digital reconstruction and virtual interaction display method and system, relates to the technical field of digital cultural relics and historical sites, and comprises the following steps: acquiring surface image data and surface topography data of a target object; performing trace layering processing on the surface image data and the surface topography data, generating trace candidate data, and classifying the trace candidate data to generate trace layering data. The scheme can distinguish surface attached traces and entity deformation traces according to sources and interlayer relationships, improve trace recognition levels, and generate developed correction data by distinguishing after extracting covering relationships, boundary coincident relationships and layer position overlying relationships of the surface image data and the surface topography data, so that the shielding mode of a cover layer to lower layer traces can be recognized and the exposure range can be corrected, and the accuracy of lower layer information development can be improved.

The invention belongs to the technical field of oral medicine, and particularly relates to a method for measuring the overall in-position error of a personalized titanium mesh. The invention provides a method for measuring the overall in-position error of a personalized titanium mesh, which comprises the following steps of: reconstructing the postoperative personalized titanium mesh by adopting a three-dimensional digital reconstruction technology, registering the position of the titanium mesh designed before the operation with the position of the actual titanium mesh after the operation, and importing the registered pre-operation and post-operation titanium mesh models into Magics software to obtain the overall in-position error of the personalized titanium mesh. And the gravity center position of each titanium mesh model is determined, and the overall deviation condition of the titanium mesh is represented by calculating the gravity center deviation amount between the titanium mesh models. In order to visually display the in-position error, the center-of-gravity shift is presented in a vector form and is decomposed into three-dimensional components of X (mesiodistal median direction), Y (lip palate direction) and Z (crown root direction), direct guidance is provided for clinicians, and the method has visual clinical guidance significance.

The application belongs to the technical field of three-dimensional modeling, and particularly relates to a three-dimensional scanning modeling method based on kinematics cooperation and sub-group fuzzy ant colony, which comprises the following steps: based on the kinematics theory of multi-body system, a global coordinate system of a scanning system and a homogeneous transformation matrix of each motion axis are established; a depth-of-field constraint is constructed; for a workpiece to be measured, a dynamic edge profile of the workpiece is acquired through the scanning system, a rough visual shell model of the workpiece is constructed, the normal vector distribution of each point on the surface of the workpiece is calculated, and a probe task set is distributed; the probe task set is split into discrete scanning task strips and solved; according to the optimal scanning task strip sequence obtained by solving, the scanning system is used to collect data of the workpiece; an adaptive sampling strategy based on a dynamic chord height error is used for point cloud gridding processing, a bidirectional curve fitting method is used to generate a NURBS surface, and a three-dimensional model of the workpiece is output. The application can realize full-automatic and high-precision digital reconstruction of a complex surface.

The invention relates to the technical field of rock mass engineering, in particular to a rock mass fracture intelligent identification and digital reconstruction method, which comprises the following steps: acquiring a rock mass fracture image; manually marking color blocks; based on transfer learning, introducing a labeled medical image cell membrane sample and a road fracture sample, and carrying out joint training on the fracture recognition model; a rock mass fracture image is input into the fracture recognition model to segment fractures, and a snakelike convolutional network is established to perform multi-scale feature extraction on the image; performing digital reconstruction on the binary image to obtain square frame set and broken line set data; and restoring the digitized data from the pixel scale to the engineering scale. According to the method, the full-process automation from intelligent identification of the rock mass crack image to digital reconstruction is realized, and the method can be widely applied to the fields of geotechnical engineering design, slope disaster identification, underground space stability analysis and the like.

The invention discloses a medical navigation method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining a preoperative three-dimensional medical image, segmenting the preoperative three-dimensional medical image, and obtaining a three-dimensional positioning structure image for registration; obtaining an intraoperative two-dimensional medical image based on reference equipment parameters, and segmenting the intraoperative two-dimensional medical image to obtain a corresponding two-dimensional positioning structure image; performing rigid registration based on the three-dimensional positioning structure image and the two-dimensional positioning structure image to obtain a rigid registration result; obtaining a time sequence two-dimensional image based on the intraoperative real-time parameters, and performing digital reconstruction radiation imaging projection on the three-dimensional positioning structure image based on the rigid registration result and the intraoperative real-time parameters to obtain a projection image; and performing frame-by-frame elastic registration based on the projection image and the time sequence two-dimensional image to obtain an elastic registration result, and performing medical navigation based on the elastic registration result. According to the method, the registration precision and efficiency can be improved, and medical navigation can be carried out in real time.

The invention discloses an old and ancient book cover image retrieval method based on dynamic partitioning and multi-modal feature fusion, and relates to the technical field of image recognition, old and ancient book cover images are used as query vectors to be input into a retrieval model, and a matched book list corresponding to old and ancient books is obtained according to the query vectors; the training process of the retrieval model is as follows: performing visual feature extraction and text feature extraction on the cover image of the ancient and old book, and partitioning; inputting the blocked visual features and text features into a cross attention mechanism, and generating an original high-dimensional fusion feature vector including a layout text structure and image layout information; performing subspace decomposition and quantization coding on the original high-dimensional fusion feature vector by adopting a product quantization algorithm to obtain a compressed feature piece vector, and constructing an ancient and old book feature database according to the compressed feature piece vector; taking cover images of the old and ancient books as input query vectors, indexing the old and ancient book database by adopting a three-level progressive query architecture, and returning a matched book list; the trainable parameters of the retrieval model are jointly optimized by constructing a total loss function; according to the retrieval method, multi-level understanding and digital reconstruction of ancient book covers are achieved, and rapid retrieval is achieved through efficient index construction.

The application discloses a kind of based on improved GRAF algorithm heart three-dimensional reconstruction method, system and program product.The three-dimensional reconstruction method includes acquisition and pre-processing generates different patients, different contrast types and normal and abnormal heart anatomical structure digital reconstruction radiogram;Improved GRAF algorithm model is constructed and trained;By inputting a small amount of CT image data to the trained GRAF algorithm model, continuous image frame is rendered, and is used for three-dimensional reconstruction and the like steps.The improved GRAF algorithm model obtained by the method based on data enhancement transformation and the combination of multiple training heads can be trained under limited CT image input, to realize the accurate reconstruction of heart three-dimensional structure.

The invention provides a free-angle 3DGS voice digital human driving method based on generative prior, and the method comprises the following steps: S1, preprocessing: carrying out the preprocessing of original video data, including the decoding of pictures and audios, the cutting of the pictures, the unification and estimation of camera parameters, the extraction of original audio features, and the division of a training set and a test set; s2, proprietary generation of a generative model: converting an out-of-domain 3DGS digital human generation model which can only generate an unreal domain into a digital human space of a more real training set; s3, flow design and training: reconstructing and redefining the 3DGS into a hidden space navigation task of a generative model by utilizing the 3DGS generative model; and S4, carrying out decoupling training by adopting a cross-view mask training method. According to the method, a generative 3DGS model is introduced into digital reconstruction and driving of a specific person, reconstruction is redefined into a hidden space navigation task of the generative model, and free-angle digital person driving is achieved.