Method and apparatus for reconstructing three-dimensional model

Abstract

The present disclosure relates to the technical field of three-dimensional model construction, and provides a method and apparatus for reconstructing a three-dimensional model, which may solve the problems of the structure being complex and modeling efficiency being low for existing three-dimensional models. The method for reconstructing a three-dimensional model of the present disclosure comprises: according to data features of a group of initial three-dimensional models, acquiring point cloud data of each initial model; selecting one model in the group of initial three-dimensional models as a standard three-dimensional model, and acquiring a plurality of discrete deformation fields by using difference values between points in the standard three-dimensional model and points in the other initial three-dimensional models; converting the plurality of discrete deformation fields into a statistical morphological model by using a multi-dimensional Gaussian process; and resampling the statistical morphological model, and acquiring a target three-dimensional model.

Description

Three-dimensional model reconstruction method and device Technical Field

[0001] The present disclosure belongs to the technical field of three-dimensional model construction, and particularly relates to a three-dimensional model reconstruction method and device. Background Art

[0002] With the continuous development of 3D modeling technology, 3D models have been widely used in more and more fields. Among them, polygon optimization and smoothing of 3D models are very important parts of the 3D development process. Most 3D models require the number of polygons to be increased or decreased according to the application scenario and usage to achieve the purpose of optimization and smoothing, so that the model can adapt to more scenarios.

[0003] However, currently, model editing is primarily done manually, requiring a significant workload comparable to rebuilding a 3D model. Furthermore, while deformers are a widely used and effective model animation solution in the 3D field, they lack automated operations for smoothing and optimizing models. This results in a significant increase in the workload associated with deformer-related model optimization. Therefore, automating deformer processing to reconstruct 3D models is a pressing issue in the industry.

[0004] Summary of the Invention

[0005] The present disclosure aims to solve at least one of the technical problems existing in the prior art and provides a method and device for reconstructing a three-dimensional model.

[0006] In a first aspect, an embodiment of the present disclosure provides a method for reconstructing a three-dimensional model, wherein the method for reconstructing a three-dimensional model includes:

[0007] Acquiring point cloud data of each of the initial three-dimensional models according to data features of the initial three-dimensional models;

[0008] Selecting one of the initial three-dimensional models as a standard three-dimensional model, and obtaining a plurality of discrete deformation fields using differences between points in the standard three-dimensional model and points in other initial three-dimensional models;

[0009] Using a multidimensional Gaussian process, the plurality of discrete deformation fields are converted into a statistical morphological model;

[0010] The statistical morphological model is resampled to obtain a target three-dimensional model.

[0011] Optionally, obtaining point cloud data of each of the initial three-dimensional models according to data features of the initial three-dimensional models includes:

[0012] placing the deformer of the initial three-dimensional model in an average state and obtaining an average state model;

[0013] obtaining limit state models of multiple deformation components of the deformer according to the average state model;

[0014] storing the average state model and the limit state model as readable data;

[0015] The readable data are sorted to obtain point cloud data of the initial model.

[0016] Optionally, the limit state model includes: a positive limit model and a negative limit model.

[0017] Optionally, the readable data includes: polygon file format data.

[0018] Optionally, selecting one of the set of initial three-dimensional models as a standard three-dimensional model and obtaining a plurality of discrete deformation fields using differences between points in the standard three-dimensional model and points in other initial three-dimensional models includes:

[0019] According to the point cloud data, obtaining a grid corresponding to the point cloud data;

[0020] Calculating, based on the grid, a difference between a point corresponding to the point cloud data and a reference point;

[0021] A plurality of discrete deformation fields are obtained according to the difference between the point corresponding to the point cloud data and the reference point.

[0022] Optionally, the converting the plurality of discrete deformation fields into a statistical morphological model using a multidimensional Gaussian process comprises:

[0023] According to the points corresponding to the point cloud data, the mean value of each point and the covariance of each point are obtained;

[0024] Obtaining a multidimensional Gaussian process according to the mean and the covariance;

[0025] According to the multidimensional Gaussian process, the statistical morphology model is obtained.

[0026] Optionally, resampling the statistical morphological model to obtain a target three-dimensional model includes:

[0027] According to the points in the statistical morphological model, some of the points are selected;

[0028] The points of the part are integrated to obtain the target three-dimensional model.

[0029] In a second aspect, an embodiment of the present disclosure provides a three-dimensional model reconstruction device, wherein the three-dimensional model reconstruction device includes:

[0030] a point cloud data acquisition module configured to acquire point cloud data of each of the initial three-dimensional models based on data features of the initial three-dimensional models;

[0031] a deformation field conversion module configured to select one of the initial three-dimensional models as a standard three-dimensional model and obtain a plurality of discrete deformation fields by using differences between points in the standard three-dimensional model and points in other initial three-dimensional models;

[0032] a statistical morphological model conversion module, configured to convert the plurality of discrete deformation fields into a statistical morphological model using a multidimensional Gaussian process;

[0033] The target three-dimensional model acquisition module is configured to resample the statistical morphological model to acquire a target three-dimensional model.

[0034] Optionally, the point cloud data acquisition module includes:

[0035] an average state model acquisition submodule, configured to place the deformer of the initial three-dimensional model in an average state and acquire an average state model;

[0036] a limit state model acquisition submodule, configured to acquire limit state models of multiple deformation components of the deformer based on the average state model;

[0037] A readable data storage submodule is configured to store the average state model and the limit state model as readable data;

[0038] The sorting submodule is configured to sort the readable data to obtain point cloud data of the initial model.

[0039] Optionally, the deformation field conversion module includes:

[0040] A grid acquisition submodule is configured to acquire a grid corresponding to the point cloud data based on the point cloud data;

[0041] a difference calculation submodule, configured to calculate the difference between the point corresponding to the point cloud data and a reference point according to the grid;

[0042] The interpolation submodule is configured to obtain a plurality of discrete deformation fields according to the difference between the point corresponding to the point cloud data and the reference point.

[0043] Optionally, the statistical morphology model conversion module includes:

[0044] The mean and covariance acquisition submodule is configured to obtain the mean of each point and the covariance of each point according to the point corresponding to the point cloud data;

[0045] a multidimensional Gaussian process acquisition module, configured to acquire a multidimensional Gaussian process according to the mean and the covariance;

[0046] The statistical morphological model acquisition submodule is configured to acquire the statistical morphological model according to the multidimensional Gaussian process.

[0047] Optionally, the target three-dimensional model acquisition module includes:

[0048] A selection submodule is configured to select some of the points in the statistical morphological model according to the points in the statistical morphological model;

[0049] The integration submodule is configured to integrate the points of the portion to obtain the target three-dimensional model.

[0050] In a third aspect, an embodiment of the present disclosure provides an electronic device, including:

[0051] at least one processor; and

[0052] a memory communicatively connected to the at least one processor; wherein,

[0053] The memory stores one or more computer programs that can be executed by the at least one processor, and the one or more computer programs are executed by the at least one processor to enable the at least one processor to perform the three-dimensional model reconstruction method provided above.

[0054] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the computer program implements the three-dimensional model reconstruction method provided above. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] FIG1 is a flow chart of a method for constructing a three-dimensional model provided in an embodiment of the present disclosure.

[0056] FIG2 is a flow chart of a method for obtaining point cloud data of an initial model.

[0057] FIG3 is a flow chart of a method for transforming a deformation field.

[0058] FIG4 is a schematic structural diagram of an exemplary three-dimensional model.

[0059] FIG5 is a flow chart of a method for transforming a statistical morphological model.

[0060] FIG6 is a schematic structural diagram of another exemplary three-dimensional model.

[0061] FIG7 is a flow chart of a method for acquiring a target three-dimensional model.

[0062] FIG8 is a schematic structural diagram of a three-dimensional model reconstruction device provided by an embodiment of the present disclosure.

[0063] FIG9 is a schematic structural diagram of an electronic device provided in some embodiments of the present disclosure. DETAILED DESCRIPTION

[0064] In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, the present disclosure is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0065] Unless otherwise defined, the technical or scientific terms used in this disclosure should have the usual meanings understood by people with ordinary skills in the field to which this disclosure belongs. The words "first", "second" and similar words used in this disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. Similarly, words such as "one", "an" or "the" do not indicate a quantity limitation, but rather indicate the existence of at least one. Words such as "include" or "comprise" mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Words such as "connect" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0066] In 3D model construction technology, the number of polygons in a model needs to be increased or decreased based on the application scenario and usage to achieve optimization and smoothing, making the model suitable for a wider range of scenarios. However, currently, model editing is mainly done manually, which requires a workload similar to rebuilding a 3D model, which is a huge amount of work. Furthermore, deformers, a widely used and effective model animation solution in the 3D field, lack relevant automated operations for deformers during model smoothing and optimization, resulting in a doubling of the workload when dealing with deformer-related model optimization.

[0067] In order to solve at least one of the above-mentioned technical problems, the embodiment of the present disclosure provides a method and device for reconstructing a three-dimensional model. The method and device for reconstructing a three-dimensional model provided by the embodiment of the present disclosure will be further described in detail below in combination with the accompanying drawings and specific implementation methods.

[0068] It should be noted that in order to better describe the three-dimensional model construction method and device provided in the embodiments of the present disclosure, the terms involved therein are explained here.

[0069] Statistical Shape Models (SSMs) are geometric models that describe a set of semantically similar objects in a very compact way. SSMs represent the average shape of a set of three-dimensional objects and their shape variation. Creating an SSM requires a correspondence mapping, which can be achieved using a parameterization of their respective samples. If a corresponding parameterization can be established for all shapes, the variation between individual shape features can be mathematically studied.

[0070] A blend shape is a technique for deforming a single mesh to achieve any combination of predefined shapes. In Maya / 3ds Max, this is called a morph target. For example, a single mesh can be the default base shape (e.g., a neutral face), and other shapes within that base shape can be used for blending / deforming, creating different expressions (smiling, frowning, closed eyelids). These are collectively referred to as blend shapes or morph targets. In practice, this involves transforming a group of models with different shapes but the same topology, such as model points and wiring, into a specific shape using a series of parameters.

[0071] Polygon smoothing optimization is to optimize the number of faces and face structure of the model by reshaping the polygons of the model. If the face of the model needs to be changed by increasing the number of model faces, it is polygon smoothing of the model. If the face of the model needs to be changed by reducing the number of model faces, it is polygon optimization of the model.

[0072] In a first aspect, an embodiment of the present disclosure provides a method for constructing a three-dimensional model. FIG1 is a flow chart of a method for constructing a three-dimensional model provided by an embodiment of the present disclosure. As shown in FIG1 , the method for constructing a three-dimensional model includes the following steps S101 to S104.

[0073] S101, acquiring point cloud data of each initial model according to data features of a group of initial three-dimensional models.

[0074] In the above step S101, the initial three-dimensional model is specifically a three-dimensional model of the human body, such as the entire human body or a specific part of the human body, which can be a hand, foot, head or internal organs. Taking the initial three-dimensional model as a three-dimensional model of a hand as an example, its data features may include: the length and width of each finger, the range of movement of each finger in three-dimensional space, etc. The initial three-dimensional model can be stored in formats such as RVT, 3DS, DWG, FBX, IFC, OSGB, OBJ, etc., which can be read and processed by software such as 3DMAX, SoftImage, Maya, UG and AutoCAD. Multiple point cloud data can form a mesh, which is often composed of triangles, quadrilaterals or other simple convex polygons. The final mesh can form a three-dimensional model. Power supply data can be stored in binary file formats such as LAS, LAZ, PCD, PCAP, PLY, PTS, etc.

[0075] S102 : selecting one of a group of initial three-dimensional models as a standard three-dimensional model, and obtaining a plurality of discrete deformation fields by using the difference between points in the standard three-dimensional model and points in other initial three-dimensional models.

[0076] In step S102, the converted point cloud data has a one-to-one correspondence between grids. For example, the tenth point cloud data in each initial 3D model represents the middle finger. One of the initial 3D models is selected as a standard 3D model. This standard 3D model has a moderate morphology, for example, a 3D model of a finger in a normal extended state. Each point cloud data only represents a point in the hand, and the points are discontinuous. The difference between points in the standard 3D model and points in other initial 3D models can be calculated to form multiple discontinuous deformation fields, i.e., discrete deformation fields.

[0077] S103, using a multi-dimensional Gaussian process, converting the multiple discrete deformation fields into a statistical morphological model.

[0078] In the above step S103, the data of multiple discrete deformation fields are brought into the Gaussian process formula, and the multiple discrete deformation fields can be interpolated to form a continuous domain deformation field. The multiple continuous domain deformation fields can constitute a statistical morphological model.

[0079] S104: resample the statistical morphological model to obtain a target three-dimensional model.

[0080] In the above step S104, for the statistical morphological model with a continuous domain, nearest neighbor interpolation, bilinear interpolation, cubic convolution interpolation, etc. can be used to select some points therein to form a new three-dimensional model, i.e., the target three-dimensional model.

[0081] In the three-dimensional model reconstruction method provided by the embodiments of the present disclosure, point cloud data of the initial three-dimensional model can be first obtained, wherein each point cloud data corresponds to a point in a grid. The point cloud data can be converted into a deformation field in the grid, and the space between grid points can be processed using interpolation calculation to form multiple discrete deformation fields. Then, using a multidimensional Gaussian process, the multiple discrete deformation fields can be converted into a statistical morphological model of a deformation field with a continuous domain. According to the application scenario, the statistical morphological model is resampled to obtain a deformer with a specific number of grid points to achieve smooth optimization of the deformer model and form a target three-dimensional model. In this way, non-functional characteristics of the three-dimensional model can be improved without changing the functional characteristics of the original model, such as improving the readability of the three-dimensional model and reducing the complexity of the original model. The three-dimensional model can be reconstructed through automated means, thereby greatly reducing the workload and effectively improving the efficiency of three-dimensional model reconstruction.

[0082] In some embodiments, Figure 2 is a flow chart of a method for obtaining point cloud data of an initial model. As shown in Figure 2, the above step S101 obtains point cloud data of each initial model based on data features of a group of initial three-dimensional models, including the following steps S1011 to S1014.

[0083] S1011, placing the deformer of the initial three-dimensional model in an average state, and obtaining an average state model.

[0084] S1012: Obtain limit state models of multiple deformation components of the deformer according to the average state model.

[0085] S1013: The average state model and the limit state model are stored as readable data.

[0086] S1014 , sorting the readable data to obtain point cloud data of the initial model.

[0087] Specifically, in this step, it is mainly necessary to convert a three-dimensional model containing a deformer into separate models for the next software processing. The deformer can be placed in an average state first, and then the maximum value of each deformation component of the deformer can be selected separately to export this model. In a specific example, the deformer has 5 deformation components, and 10 separate models need to be exported, that is, the limit state model of the positive and negative parts of each deformation classification for the average state. At the same time, the initial average state model needs to be exported as a reference. It should be noted here that, with the horizontal direction as a deformation component, the limit position to which the grid point in the three-dimensional model moves to the left in the horizontal direction can be defined as its negative limit state, and the limit position to which it moves to the right in the horizontal direction can be defined as its positive polarity state. The deformation components in other directions can also refer to the above description and will not be listed one by one here.

[0088] To facilitate the reading of 3D models, the polygonal file format (ply) is used to export the mesh information of the 3D model. After exporting the model, the model data needs to be processed. The main purpose is to convert the complete model information into a point cloud state, that is, only the point coordinate data in the model file is saved in sequence to obtain the point cloud data of the initial model.

[0089] In some embodiments, Figure 3 is a flow chart of a method for converting a deformation field. As shown in Figure 3, the above step S102 selects one of a group of initial three-dimensional models as a standard three-dimensional model, and uses the difference between the points in the standard three-dimensional model and the points in other initial three-dimensional models to obtain multiple discrete deformation fields, including the following steps S1021 to S1023.

[0090] S1021: Obtain a grid corresponding to the point cloud data based on the point cloud data.

[0091] S1022: Calculate the difference between the point corresponding to the point cloud data and the reference point based on the grid.

[0092] S1023 , obtaining a plurality of discrete deformation fields according to the difference between the point corresponding to the point cloud data and the reference point.

[0093] Specifically, the point cloud data converted in the previous stage has a characteristic that all grids are corresponding. That is, they all have the same number of points, and points with the same address in the grid represent the same point / area in the grid.

[0094] Any mesh corresponding to this reference can be represented as a deformation field, which is defined on this reference mesh; that is, the points of the reference mesh are the domain that defines the deformation field. The deformation can be calculated by taking the difference between the corresponding points of the mesh and the reference points.

[0095] The deformation field obtained in the above steps is discrete because it is defined only on the grid points. This is not ideal because the real-world objects being modeled are continuous and the discretization of the grid is quite arbitrary. In the scheme of the embodiment of the present disclosure, it is necessary to obtain a continuous representation of the deformation field by interpolation. The interpolation can be found by finding the nearest point on the surface for each point in the three-dimensional space, and using the corresponding deformation as the deformation of the given point. The points on the surface are then obtained by interpolating the centroids of the corresponding vertices. As a result of the interpolation, the deformation field of the entire real space is obtained, that is, the deformation field corresponding to any point can be obtained.

[0096] Figure 4 is a schematic diagram of the structure of an exemplary three-dimensional model. As shown in Figure 4, a deformation field is formed by interpolating the positions of corresponding points of a figure from a reference model. In Figure 4, the model shape of a hand composed of contour points is shown. The black dots are in an averaged state. The black dot contour can be used as a basis. The other hand contour is the contour indicated by the arrow in the figure. The direction indicated by the arrow is expressed as a vector from the black dot (the reference point of the point) to the actual position of the other hand contour point. This operation can be used for every point of each model, which is a deformation field.

[0097] Continuing to refer to Figure 4 above, the existing contour in this three-dimensional hand model is composed of a group of points, so the model actually expressed is a discrete contour. When we connect these points into a line, a continuous contour is formed instead of a series of points with distances. In this way, the contour continuity process is realized, that is, it can be represented by a group of y(x) functions. Similarly, the two continuous hand contours are subjected to the above-mentioned deformation field extraction operation, that is, the corresponding coordinates of the two hand contours in Figure 4 are subtracted to generate the deformation field of the continuous domain.

[0098] In some embodiments, FIG5 is a flow chart of a method for converting a statistical morphological model. As shown in FIG5 , the above step S103 utilizes a multidimensional Gaussian process to convert multiple discrete deformation fields into a statistical morphological model, including the following steps S1031 to S1033.

[0099] S1031, according to the points corresponding to the point cloud data, obtain the mean of each point and the covariance of each point.

[0100] S1032: Obtain a multidimensional Gaussian process based on the mean and covariance.

[0101] S1033: Obtain a statistical morphological model according to a multidimensional Gaussian process.

[0102] Specifically, taking a deformer with 5 deformation components as an example, a total of 11 models were obtained. The positive maximum model and the negative maximum model corresponding to each deformation component were taken as a group to calculate the statistical morphological model of the continuous domain. In this step, the following logic will be used: first, the mean μ of each point in the model of each group is calculated (Formula 1), and then the covariance of any corresponding point is calculated (Formula 2).

[0103] Finally, μ and ∑ in the above are brought into a multidimensional Gaussian process (Formula 3).

[0104] In this way, a multi-dimensional Gaussian process is obtained. This process has a lot of information and properties that can be extracted, but in the embodiment of the present disclosure, this Gaussian process is mainly used as a sum of the above-mentioned model deformation field. In summary, in the above-mentioned mathematical calculations, multiple models are serialized and then reintegrated together. Taking the hand shape model as an example, as shown in Figure 6, the black outline in this figure is the hand shape that we set as the average model, and the black shadows around the hand shape are the possible positions of the hand outline in different models under different circumstances. It can be understood as a series of hands of different shapes flattened and overlapped together to form a superimposed shadow. In this process, this Gaussian process learns a kind of information, that is, when part of the hand outline point appears at a certain position in the shadow, the position of each point of the entire outline can be obtained.

[0105] After the above explanation, by inputting the positions of a finite number of points in the positive maximum value model corresponding to a certain deformation component into the above Gaussian process, the expression of the entire contour is obtained, and the continuous expression of the entire contour is obtained.

[0106] In some embodiments, FIG7 is a flowchart of a method for acquiring a target three-dimensional model. As shown in FIG7 , the above step S104 , resampling the statistical morphological model to acquire the target three-dimensional model, includes the following steps S1041 to S1042 .

[0107] S1041: Select some of the points in the statistical morphological model.

[0108] S1042: Integrate some of the points to obtain a target three-dimensional model.

[0109] Specifically, after completing the above steps, we can obtain the expression of the positive maximum value model corresponding to a certain deformation component in the continuous space. After implementing this step, we only need to resample the model according to actual needs to obtain the new point position of the model. Taking the hand shape model as an example, after obtaining the contour line of a certain shape of the hand, when we need a hand contour point map composed of 100 points, we only need to select the required 100 points in the entire contour line as needed to realize the smoothing or optimization process of the entire three-dimensional model.

[0110] The resulting new deformer, a combination of models with different shapes, operates the same way as above, requiring only resampling within the standard model of the Gaussian process. Based on the consistent position of each point across the entire model, a new deformer is constructed from the same selected region of points with different shapes. The original deformer had a densely populated center and many points. After these operations, a smoothing optimization of the resampling was performed, resulting in a deformer with fewer points.

[0111] In the second aspect, an embodiment of the present disclosure provides a three-dimensional model reconstruction device. Figure 8 is a structural schematic diagram of a three-dimensional model reconstruction device provided by an embodiment of the present disclosure. As shown in Figure 8, the three-dimensional model reconstruction device includes: a point cloud data acquisition module 801, a deformation field conversion module 802, a statistical morphological model conversion module 803 and a target three-dimensional model acquisition module 804.

[0112] The point cloud data acquisition module 801 is configured to obtain point cloud data of each initial model based on the data features of a group of initial three-dimensional models; the deformation field conversion module 802 is configured to select one of the group of initial three-dimensional models as the standard three-dimensional model, and use the difference between the points in the standard three-dimensional model and the points in other initial three-dimensional models to obtain multiple discrete deformation fields; the statistical morphological model conversion module 803 is configured to use a multi-dimensional Gaussian process to convert multiple discrete deformation fields into a statistical morphological model; the target three-dimensional model acquisition module 804 is configured to resample the statistical morphological model to obtain the target three-dimensional model.

[0113] The three-dimensional model reconstruction device provided in the embodiment of the present disclosure is used to implement the three-dimensional model reconstruction method provided in any of the above embodiments. For specific related descriptions, please refer to the description of the three-dimensional model reconstruction method in any of the above embodiments, which will not be repeated here.

[0114] Specifically, the point cloud data acquisition module 801 includes: an average state model acquisition submodule 8011, which is configured to place the deformer of the initial three-dimensional model in an average state and obtain the average state model; a limit state model acquisition submodule 8012, which is configured to obtain the limit state model of multiple deformation components of the deformer based on the average state model; a readable data storage submodule 8013, which is configured to store the average state model and the limit state model as readable data; and a sorting submodule 8014, which is configured to sort the readable data to obtain the point cloud data of the initial model.

[0115] The deformation field conversion module 802 includes: a grid acquisition submodule 8021, which is configured to obtain a grid corresponding to the point cloud data based on the point cloud data; a difference calculation submodule 8022, which is configured to calculate the difference between the point corresponding to the point cloud data and the reference point based on the grid; and an interpolation submodule 8023, which is configured to obtain the deformation field of the continuous domain based on the difference between the point corresponding to the point cloud data and the reference point.

[0116] The statistical morphological model conversion module 803 includes: a mean and covariance acquisition submodule 8031, which is configured to obtain the mean of each point and the covariance of each point based on the point corresponding to the point cloud data; a multidimensional Gaussian process acquisition module 8032, which is configured to obtain a multidimensional Gaussian process based on the mean and covariance; and a statistical morphological model acquisition submodule 8033, which is configured to obtain a statistical morphological model based on the multidimensional Gaussian process.

[0117] The target three-dimensional model acquisition module 804 includes: a selection submodule 8041 configured to select some points in the statistical morphological model according to the points; and an integration submodule 8042 configured to integrate some points to obtain the target three-dimensional model.

[0118] In the third aspect, an embodiment of the present disclosure provides an electronic device. Figure 9 is a structural diagram of the electronic device provided in some embodiments of the present disclosure. As shown in Figure 9, the electronic device includes: one or more processors 901; a memory 902, on which one or more programs are stored. When the one or more programs are executed by one or more processors, the one or more processors implement the three-dimensional model reconstruction method provided in any of the above embodiments; one or more I / O interfaces 903, connected between the processor and the memory, and configured to implement information interaction between the processor and the memory.

[0119] Among them, the processor 901 is a device with data processing capabilities, including but not limited to a central processing unit (CPU); the memory 902 is a device with data storage capabilities, including but not limited to random access memory (RAM, more specifically SDRAM, DDR, etc.), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), and flash memory (FLASH); the I / O interface (read-write interface) 903 is connected between the processor 901 and the memory 902, and can realize information interaction between the processor 901 and the memory 902, including but not limited to a data bus (Bus), etc.

[0120] In some embodiments, the processor 901 , the memory 902 , and the I / O interface 903 are connected to each other via a bus, and further connected to other components of the computing device.

[0121] In a fourth aspect, this embodiment provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the three-dimensional model reconstruction method provided in any of the above embodiments.

[0122] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In hardware implementations, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or temporary medium). As is well known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable, and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those skilled in the art that communication media typically embodies computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0123] It is understood that the above embodiments are merely exemplary embodiments for illustrating the principles of the present disclosure, and the present disclosure is not limited thereto. Those skilled in the art may make various modifications and improvements without departing from the spirit and substance of the present disclosure, and such modifications and improvements are also considered to be within the scope of protection of the present disclosure.

Claims

1. A method for reconstructing a three-dimensional model, wherein: The three-dimensional model reconstruction method comprises: According to data features of a group of initial three-dimensional models, obtaining point cloud data of each of the initial models; Selecting one of the initial three-dimensional models as a standard three-dimensional model, and obtaining a plurality of discrete deformation fields by using the difference between points in the standard three-dimensional model and points in other initial three-dimensional models; Using a multidimensional Gaussian process, a plurality of discrete deformation fields are converted into a statistical morphological model; The statistical morphological model is resampled to obtain a target three-dimensional model.

2. The three-dimensional model reconstruction method according to claim 1, wherein: The step of obtaining point cloud data of each of the initial three-dimensional models according to data features of the initial three-dimensional models includes: placing the deformer of the initial three-dimensional model in an average state and obtaining an average state model; According to the average state model, obtaining the limit state model of multiple deformation components of the deformer; storing the average state model and the limit state model as readable data; The readable data are sorted to obtain point cloud data of the initial model.

3. The method for reconstructing a three-dimensional model according to claim 2, wherein: The limit state model includes: a positive limit model and a negative limit model.

4. The method for reconstructing a three-dimensional model according to claim 2, wherein: The readable data includes: polygon file format data.

5. The method for reconstructing a three-dimensional model according to claim 2, wherein: The step of selecting one of the initial three-dimensional models as a standard three-dimensional model and obtaining a plurality of discrete deformation fields by using the difference between points in the standard three-dimensional model and points in other initial three-dimensional models includes: According to the point cloud data, obtaining a grid corresponding to the point cloud data; Calculating the difference between the point corresponding to the point cloud data and the reference point according to the grid; A plurality of discrete deformation fields are obtained according to the difference between the point corresponding to the point cloud data and the reference point.

6. The method for reconstructing a three-dimensional model according to claim 5, wherein: The method of converting the plurality of discrete deformation fields into a statistical morphological model by using a multidimensional Gaussian process includes: According to the points corresponding to the point cloud data, obtain the mean of each point and the covariance of each point; Obtain a multidimensional Gaussian process according to the mean and the covariance; According to the multidimensional Gaussian process, the statistical morphology model is obtained.

7. The method for reconstructing a three-dimensional model according to claim 6, wherein: Resampling the statistical morphological model to obtain a target three-dimensional model includes: According to the points in the statistical morphological model, select some of the points; The points of the parts are integrated to obtain the target three-dimensional model.

8. A three-dimensional model reconstruction device, wherein: The three-dimensional model reconstruction device comprises: A point cloud data acquisition module is configured to acquire point cloud data of each of the initial models according to data features of a group of initial three-dimensional models; A deformation field conversion module is configured to select one of the initial three-dimensional models as a standard three-dimensional model, and obtain a plurality of discrete deformation fields by using the difference between points in the standard three-dimensional model and points in other initial three-dimensional models; A statistical morphological model conversion module is configured to convert a plurality of discrete deformation fields into a statistical morphological model using a multi-dimensional Gaussian process; The target three-dimensional model acquisition module is configured to resample the statistical morphological model to acquire the target three-dimensional model.

9. The three-dimensional model reconstruction device according to claim 8, wherein: The point cloud data acquisition module includes: an average state model acquisition submodule, configured to place the deformer of the initial three-dimensional model in an average state and acquire an average state model; A limit state model acquisition submodule is configured to acquire a limit state model of a plurality of deformation components of the deformer according to the average state model; A readable data storage submodule, configured to store the average state model and the limit state model as readable data; The sorting submodule is configured to sort the readable data to obtain the point cloud data of the initial model.

10. The three-dimensional model reconstruction device according to claim 9, wherein: The deformation field conversion module comprises: A grid acquisition submodule is configured to acquire a grid corresponding to the point cloud data according to the point cloud data; a difference calculation submodule, configured to calculate the difference between the point corresponding to the point cloud data and the reference point according to the grid; The interpolation submodule is configured to obtain a plurality of discrete deformation fields according to the difference between the point corresponding to the point cloud data and the reference point.

11. The three-dimensional model reconstruction device according to claim 10, wherein: The statistical morphology model conversion module comprises: The mean and covariance acquisition submodule is configured to acquire the mean of each point and the covariance of each point according to the point corresponding to the point cloud data; A multidimensional Gaussian process acquisition module is configured to acquire a multidimensional Gaussian process according to the mean and the covariance; The statistical morphological model acquisition submodule is configured to acquire the statistical morphological model according to the multidimensional Gaussian process.

12. The three-dimensional model reconstruction device according to claim 11, wherein: The target three-dimensional model acquisition module includes: A selection submodule is configured to select some of the points in the statistical morphological model according to the points in the statistical morphological model; The integration submodule is configured to integrate the points of the portion to obtain the target three-dimensional model.

13. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores one or more computer programs executable by the at least one processor, and the one or more computer programs are executed by the at least one processor to enable the at least one processor to perform the three-dimensional model reconstruction method according to any one of claims 1 to 7.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the computer program implements the three-dimensional model reconstruction method according to any one of claims 1 to 7.

Images