WO2022021309A1

WO2022021309A1 - Method and apparatus for establishing model, electronic device, and computer readable storage medium

Info

Publication number: WO2022021309A1
Application number: PCT/CN2020/106167
Authority: WO
Inventors: 黄文杰; 冯乐满; 黄胜
Original assignee: 深圳市大疆创新科技有限公司
Priority date: 2020-07-31
Filing date: 2020-07-31
Publication date: 2022-02-03

Abstract

A method and apparatus for establishing a model and a computer readable storage medium. The method comprises: obtaining description accuracy information of a first grid display layer of a scene three-dimensional model for a scene (101); determining a target geometric error of a second grid display layer of the scene three-dimensional model according to the description accuracy information (102); performing grid simplification processing on multiple grid surfaces in the first grid display layer to obtain new grid surfaces (103); and using the newly formed grid surfaces as the second grid display layer of the scene three-dimensional model for storage (104). According to the method, in the process of obtaining the grid display layers having decreasing levels of detail, the process of calculation of a perfect display layer is eliminated and calculation pressure is reduced; moreover, because more accurate physical size information and/or deviation information of sampling points are used for determining the target geometric error, the problem that when a global hausdorff distance is used as the geometric error, the level index of detail of the grid display layer is too large, thereby causing the subsequent rendering accuracy to decrease is solved.

Description

Method, apparatus, electronic device, and computer-readable storage medium for establishing a model

technical field

The present application relates to the technical field of model establishment, and in particular, to a method, apparatus, electronic device, and computer-readable storage medium for establishing a model.

Background technique

Levels of Detail (LOD, Levels of Detail) technology refers to determining the resource allocation when rendering objects of the 3D model according to the position and importance of the 3D model in the display environment, so as to reduce the amount of distant or non-important objects. The number of faces and the degree of detail, thus reducing the consumption of resources.

In the current scheme, LOD technology is often used in the field of computer graphics. Its purpose is to establish a three-dimensional model including multiple model display layers. Each model display layer is used to represent the same object, and different model display layers have different levels of detail. index. Specifically, in the application in the field of computer graphics, first, based on the known perfect display layer (ie, the finest layer), the perfect display layer is simplified into multiple model display layers with decreasing details, and each model display layer is The hausdorff distance between the perfect display layer and the perfect display layer is used as the detail level indicator of each model display layer and the iteration termination condition of the simplified operation, so that the subsequent renderer can select the corresponding model display layer according to the detail level indicator. For display, for example, the farther the object is from the viewpoint, the model display layer with lower level of detail is selected to display the object.

However, in the field of image-based reconstruction of 3D scenes, there is no known perfect display layer. Using the current solution requires calculation to generate a perfect display layer, which is time-consuming and labor-intensive. In addition, the hausdorff distance reflects the model display layer and the perfect display layer. The maximum distance value of the layer, and the hausdorff distance is used as the detail level indicator of the model display layer, which will cause the detail level indicator of the model display layer to be too large, resulting in a decrease in the subsequent rendering accuracy.

SUMMARY OF THE INVENTION

The present application provides a method, an apparatus and a computer-readable storage medium for establishing a model, which can solve the problems in the prior art that a perfect display layer needs to be calculated and generated, which leads to time-consuming and labor-intensive, and easily leads to a decrease in the subsequent rendering accuracy.

In a first aspect, an embodiment of the present application provides a method for establishing a model, including:

Acquiring description accuracy information of the scene by the first grid display layer of the 3D model of the scene, wherein the first grid display layer includes a plurality of grid surfaces used to display the surface of the 3D model of the scene, and the description accuracy The information includes: the physical size information of the scene corresponding to the sampling point in the first grid display layer, and/or the actual projection point of the position point of the scene displayed by the sampling point in the first grid display layer deviation information from the sampling point;

The target geometric error of the second mesh display layer of the three-dimensional scene model is determined according to the description accuracy information, wherein the target geometric error is used to characterize the relative relationship between the mesh surfaces in the second mesh display layer and the The geometric error of the surface of the 3D model of the scene;

Perform grid simplification processing on the plurality of grid surfaces in the first grid display layer to obtain a new grid surface, wherein the grid simplification processing includes: folding the grid surfaces of the plurality of grid surfaces. The edge forms a new point, and the formed new point is used as the corner point of the new mesh surface, and the actual geometric error of the new mesh surface relative to the surface of the three-dimensional model of the scene is equal to the target geometric error. The difference is less than the preset threshold;

The new mesh surface formed is stored as the second mesh display layer of the three-dimensional model of the scene.

In a second aspect, an embodiment of the present application provides an apparatus for establishing a model, including:

memory and processor;

The memory is used for acquiring description accuracy information of the scene by a first grid display layer of the scene 3D model, wherein the first grid display layer includes a plurality of grids used to display the surface of the scene 3D model On the other hand, the description accuracy information includes: physical size information of the scene corresponding to the sampling point in the first grid display layer, and/or, the position point of the scene displayed by the sampling point is displayed in the first grid deviation information between the actual projection point in the layer and the sampling point;

storing the formed new mesh surface as the second mesh display layer of the three-dimensional model of the scene;

The processor is used to:

Perform grid simplification processing on the plurality of grid surfaces in the first grid display layer to obtain a new grid surface, wherein the grid simplification processing includes: folding the grid surfaces of the plurality of grid surfaces. The edge forms a new point, and the formed new point is used as the corner point of the new mesh surface, and the actual geometric error of the new mesh surface relative to the surface of the three-dimensional model of the scene is equal to the target geometric error. The difference is less than the preset threshold.

In a third aspect, the present application provides an electronic device, comprising a processor, a memory, and a computer program stored on the memory and executable on the processor, the computer program implementing the above when executed by the processor the method described in the aspect.

In a fourth aspect, the present application provides a computer-readable storage medium, the computer-readable storage medium comprising instructions, which when executed on a computer, cause the computer to perform the method described in the above aspects.

In a fifth aspect, the present application provides a computer program product comprising instructions which, when run on a computer, cause the computer to perform the method described in the above aspects.

In the embodiment of the present application, the present application determines the geometric error in the first grid display layer by using the existing physical size information and/or deviation information of the sampling points in the first grid display layer, so as to utilize the geometric error The first grid display layer has a measure of the degree of detail. In the embodiment of the present application, the existing information of the sampling points in the first grid display layer obtained by the three-dimensional reconstruction of the scene is used, so that in the three-dimensional reconstruction scene, the measurement of the detail degree of the grid display layer is combined with the camera shooting process. The existing characteristics in the regions of each scene better match the actual situation that the degree of detail in different regions is different, improve the accuracy of the termination simplification conditions when the mesh surface is simplified, and reduce the use of The accuracy loss brought by the global hausdorff distance as the termination simplification condition makes it no longer rely on a known perfect display layer in the process of obtaining a multi-layer grid display layer with decreasing level of detail, eliminating the need to calculate the perfect display layer. The process reduces the calculation pressure, and because the more accurate physical size information and/or deviation information of the sampling point is used to determine the target geometric error, the detail level of the grid display layer generated when the global hausdorff distance is used as the geometric error is solved. The indicator is too large, which causes the problem of subsequent rendering accuracy decline.

Description of drawings

1 is a flowchart of a method for establishing a model provided by an embodiment of the present application;

2 is a schematic diagram of a grid display layer provided by an embodiment of the present application;

3 is a schematic diagram of a simplified process of a grid display layer provided by an embodiment of the present application;

4 is a flowchart of the specific steps of a method for establishing a model provided by an embodiment of the present application;

5 is a schematic block diagram of a grid display layer provided by an embodiment of the present application;

FIG. 6 is a block diagram of an apparatus for establishing a model provided by an embodiment of the present application.

detailed description

In order to make the objects, technical solutions and advantages of the present invention more apparent, exemplary embodiments according to the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some of the embodiments of the present invention, not all of the embodiments of the present invention, and it should be understood that the present invention is not limited by the example embodiments described herein. Based on the embodiments of the present invention described in the present invention, all other embodiments obtained by those skilled in the art without creative efforts shall fall within the protection scope of the present invention.

In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the present invention. It will be apparent, however, to one skilled in the art that the present invention may be practiced without one or more of these details. In other instances, some technical features known in the art have not been described in order to avoid obscuring the present invention.

It should be understood that the present invention may be embodied in different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used herein, the singular forms "a," "an," and "the/the" are intended to include the plural forms as well, unless the context clearly dictates otherwise. It should also be understood that the terms "compose" and/or "include", when used in this specification, identify the presence of stated features, integers, steps, operations, elements and/or components, but do not exclude one or more other The presence or addition of features, integers, steps, operations, elements, parts and/or groups. As used herein, the term "and/or" includes any and all combinations of the associated listed items.

For a thorough understanding of the present invention, detailed structures will be presented in the following description in order to explain the technical solutions proposed by the present invention. Alternative embodiments of the present invention are described in detail below, however, the invention is capable of other embodiments in addition to these detailed descriptions.

The method, device, and system for establishing a model of the present application will be described in detail below with reference to the accompanying drawings. The features of the embodiments and implementations described below may be combined with each other without conflict.

FIG. 1 is a flowchart of a method for establishing a model provided by an embodiment of the present application. As shown in FIG. 1 , the method may include:

Step 101: Acquire the description accuracy information of the scene by the first grid display layer of the three-dimensional model of the scene.

Wherein, the first grid display layer includes a plurality of grid surfaces used to display the surface of the three-dimensional model of the scene, and the description accuracy information includes: the scene corresponding to the sampling point in the first grid display layer Physical size information, and/or deviation information between the actual projection point of the scene represented by the sampling point and the sampling point in the first grid presentation layer.

In this embodiment of the present application, the method for building a model can be applied to the field of building a three-dimensional model of a scene based on an image of a scene. Specifically, the method for building a model can reconstruct the scene based on images collected by a movable platform in the scene. After the 3D scene model is established, the 3D scene model can be displayed on the terminal through the renderer, so as to realize the 3D reconstruction display of the scene. Among them, the movable platform may include: unmanned aerial vehicle, unmanned vehicle, unmanned boat, handheld shooting equipment, etc.

Specifically, the goal of building a 3D model of a scene is to build a multi-layer grid display layer, and the level of detail of the multi-layer grid display layer decreases. Each grid display layer can include multiple grid surfaces, and the grid surfaces are used for displaying The 3D model of the scene corresponds to the surface of the object.

For example, assuming that the corresponding object of the 3D model of the scene is a sphere with a smooth surface, the mesh display layer can be understood as a regular polyhedron that wraps the sphere. Different mesh display layers have different numbers of mesh faces. The higher the number of mesh faces, the higher the level of detail in the mesh display layer. Referring to FIG. 2 , it shows a schematic diagram of a grid display layer provided by an embodiment of the present application. For a grid display layer 10 and a grid display layer 11 , the grid display layer 10 is a regular hexahedron, and the grid display layer 10 is a regular hexahedron. Layer 11 is a regular dodecahedron. During the process of changing the grid display layer 10 to the grid display layer 11 , the number of grid surfaces of the grid display layer increases, and the grid display layer 11 is closer to a sphere than the grid display layer 10 . Therefore, the detail level of the grid display layer 11 is higher than that of the grid display layer 10 .

The first grid display layer can be the first layer with the highest level of detail in the three-dimensional model of the scene. In this case, the first grid display layer can be constructed based on the images collected by the movable platform in the scene. An implementation situation Motion recovery structure (SFM, structure from motion) technology can be used to realize the construction of the image collected by the movable platform in the scene as the first grid display layer. The SFM technology can be based on the camera of the movable platform when the camera captures the image. pose, extract the point cloud data of the image, and establish a textured mesh display layer based on the textureless mesh layer of the point cloud data component and the texture of the image, and the textured mesh display layer can be used as the detail The highest level of the first grid presentation layer.

It should be noted that the first grid display layer may be in other layers except the last layer in the 3D model of the scene. In this case, the first grid display layer may be constructed based on the previous grid display layer. For the specific construction method, reference may be made to the following process of obtaining the second grid display layer from the first grid display layer.

Specifically, the description accuracy information can be used to determine the detail level of the grid display layer, and the detail level of the grid display layer can be represented by a geometric error, which is used to characterize the mesh surface in the grid display layer relative to the scene. The geometric distance between the surfaces of the 3D model (that is, the surface of the object corresponding to the 3D model of the scene, and can also be understood as the surface of the mesh surface of the perfect mesh display layer). Generally speaking, the smaller the geometric error of the grid display layer, the greater the number of grid surfaces included in the grid display layer, and the smaller the geometric distance between the grid surface of the grid display layer and the surface of the 3D model of the scene. Since the number of mesh faces is large, and the mesh faces are closer to the perfect mesh display layer, the more similar the mesh display layer is to the perfect mesh display layer, the higher the level of detail of the mesh display layer.

For example, referring to FIG. 2 , when the object corresponding to the three-dimensional scene model is a sphere with a smooth surface, the grid display layer 10 is a regular hexahedron, and the grid display layer 11 is a regular dodecahedron. It can be seen that the greater the number of grid surfaces of the grid display layer, the smaller the geometric distance between the grid surface and the sphere surface, so that the geometric error of the grid display layer 11 is smaller than that of the grid display layer 11 .

In this embodiment of the present application, the description accuracy information of the scene on the first grid display layer of the 3D model of the scene is obtained, specifically, the physical size information of the scene corresponding to the sampling point in the first grid display layer may be obtained, and/or the sampling Deviation information between the actual projection point and the sampling point of the location point of the scene displayed by the point in the first grid display layer.

Among them, in the process of reconstructing the three-dimensional model of the scene according to the image collected by the camera of the mobile device, there is an intersection point between the grid surface of the grid display layer and the grid surface, and the intersection point is also called the corner point or sampling point of the grid surface. point, so that the sampling point can include the physical size information of the corresponding scene, and the physical size information can be called the ground sampling distance (gsd, ground sample distance), which can refer to the physical distance between the center point of adjacent pixels in the image corresponding to the ground .

The calculation method of the gsd of the sampling point v is: when the depth (v) of the sampling point v in the image and the focal length focal of the camera are known, the gsd of the sampling point v=depth(v)/focal.

Further, the deviation information between the actual projection point of the scene displayed by the sampling point and the sampling point in the first grid display layer can be referred to as the motion recovery structure (SFM, Structure) in the first grid display layer. From Motion) reconstruction residuals of sparse points, SFM is a technology that calculates the internal and external parameters of a group of pictures according to the principle of multi-view geometry. Some sparse spots.

For example, inputting two pictures, first extract the sampling points of each picture, and then match the sampling points of the two pictures, after obtaining the matching, you can conduct forward intersection to obtain the position point. The deviation between the two-dimensional projection point projected from the position point to the image and the actual sampling point is also called the reconstruction residual. The core of the SFM algorithm is to optimize the internal and external parameters of the camera and the three-dimensional point coordinates to minimize the reconstruction residual. The deviation information in this embodiment of the present application may include the maximum value or the average value of the reconstruction residuals of the SFM sparse points in the first grid display layer.

In the process of reconstructing the 3D model of the scene based on the images of the scene collected by the movable platform, the grid precision of each area of the scene is different. By using the first grid to display the physical size information corresponding to the sampling points in the layer, and/or , the deviation information between the actual projection point of the scene displayed by the sampling point and the sampling point in the first grid display layer is used as the description accuracy information of the scene by the first grid display layer to display the first grid. The geometric error of the layer is measured so that the first mesh display layer has a measure of the level of detail.

Compared with the prior art, a global hausdorff distance is used to measure the detail level of each grid display layer. Due to the different grid precisions corresponding to different areas in the actual scene, the deviation of many areas is not as large as the global hausdorff distance. , the use of the global hausdorff distance will ignore the differences in each area, resulting in the level of detail of each grid display layer being represented by a larger hausdorff distance, resulting in a larger level of detail in the model display layer. Therefore, in the embodiment of the present application, the existing information of the sampling points in the first grid display layer obtained by the three-dimensional reconstruction of the scene is used, so that the measurement of the detail level of the grid display layer is combined with the camera shooting process in the three-dimensional reconstruction scene. The existing features in each region better match the actual situation that the details of different regions are differentiated, and reduce the accuracy loss caused by using the global hausdorff distance.

Step 102: Determine the target geometric error of the second mesh display layer of the three-dimensional model of the scene according to the description accuracy information.

The target geometric error is used to represent the geometric error of the mesh surface in the second mesh display layer relative to the surface of the three-dimensional model of the scene.

In this embodiment of the present application, after obtaining the description accuracy information of the first grid display layer, the first grid display layer can be simplified by performing a simplified operation to reduce the number of grid surfaces in the first grid display layer, and the result is obtained In the second mesh display layer, since the number of mesh faces in the second mesh display layer is reduced, the detail level of the second mesh display layer is reduced, thereby meeting the requirements of the LOD technology.

Specifically, in the simplification operation, the target geometric error needs to be used as the termination condition of the simplification operation, that is, the geometric error between the new mesh surface formed after the simplification of the first mesh display layer and the surface of the three-dimensional model of the scene is continuously determined, until This geometric error is equal to or close to the target geometric error.

Since the concept of geometric error has been explained in step 101, the target geometric error of the second grid display layer is determined according to the description accuracy information of the first grid display layer. Specifically, the description accuracy of the first grid display layer can be determined first. The geometric error of the first grid display layer is obtained by information calculation, and then the geometric error of the second grid display layer is determined according to the geometric error of the first grid display layer.

Optionally, step 102 may specifically include:

Sub-step 1021: Determine the target geometric error of the first mesh display layer according to the description precision information and the first coefficient.

In the embodiment of the present application, since the description precision information and the geometric error are two different parameters, in an implementation manner, the product of the description precision information and the first coefficient may be determined as the target of the first grid display layer Geometric error, where the first coefficient may be a fixed conversion coefficient, or may be a coefficient established according to actual scene requirements, which is not limited in this embodiment of the present application.

Sub-step 1022: Determine the target geometric error of the second grid display layer according to the target geometric error of the first grid display layer.

In this embodiment of the present application, the target geometric error of the second grid display layer may be determined according to the target geometric error of the first grid display layer according to actual modeling requirements and simplification requirements. If the level of detail of the layer is low, the target geometric error of the second grid display layer can be set to a larger value based on the target geometric error of the first grid display layer; If the value is high, the target geometric error of the second grid display layer may be set to a smaller value based on the target geometric error of the first grid display layer.

Optionally, sub-step 1022 may specifically include:

Sub-step A1: Determine the target geometric error of the second grid display layer according to the target geometric error of the first grid display layer and the second coefficient.

Optionally, the second coefficient increases as the number of layers where the second mesh display layer is located in the three-dimensional model of the scene increases.

Specifically, the target geometric error of the second grid display layer is determined according to the target geometric error of the first grid display layer. Specifically, the product value of the target geometric error of the first grid display layer and the second coefficient may be determined as For the target geometric error of the second grid display layer, the second coefficient can be a value greater than 1, so that as the number of layers where the grid display layer is located in the 3D model of the scene increases, the target geometric error of the grid display layer also increases. Large, the detail level of the grid display layer is gradually reduced to meet the requirements of LOD technology.

For example, it is assumed that the three-dimensional model of the scene includes 3 mesh presentation layers m1, m2, and m3. The target geometric error of each grid display layer is the product of the target geometric error of the previous grid display layer and the second coefficient, and the second coefficient can be the number of layers where the grid display layer is located. The target geometric error is a, then the target geometric error of the grid display layer m2 is 2a, and the target geometric error of the grid display layer m2 is 3a. , the target geometric error of the grid display layer also increases, and the detail level of the grid display layer is gradually reduced.

It should be noted that the value of the second coefficient may be set according to actual needs, which is not limited in the embodiment of the present application.

Step 103: Perform grid simplification processing on the plurality of grid surfaces in the first grid display layer to obtain new grid surfaces.

The mesh simplification processing includes: folding the edges of the plurality of mesh surfaces to form new points, and using the formed new points as corner points of the new mesh surfaces, and the new mesh surfaces are relative to all the mesh surfaces. The difference between the actual geometric error of the surface of the three-dimensional model of the scene and the target geometric error is smaller than a preset threshold.

In this embodiment of the present application, the purpose of mesh simplification processing is to obtain the next mesh display layer by reducing the number of mesh faces in the current mesh display layer, so that the detail level of the next mesh display layer is lower .

Specifically, referring to FIG. 3 , it shows a schematic diagram of a simplified process of a grid display layer provided by an embodiment of the present application. An embodiment of the present application uses a relatively simple example to describe the grid simplification process. In FIG. 3 , , shows a mesh face set 12 and a mesh face set 13. The mesh face set 12 includes 9 mesh faces, and the mesh face set 13 includes 7 mesh faces. In a simplified operation, by folding The edge of the grid surface in the grid surface set 12, eliminate the corner points v1 and v2, and form a new corner point v3 to obtain the grid surface set 13, that is, through the collapse of the edge, the middle of the grid surface set 12 The corner point v1 and the corner point v2 are merged into a corner point v3, so that the two triangular mesh faces in the middle are eliminated, that is, the number of mesh faces is reduced.

It should be noted that, according to actual requirements, a new corner point may not be generated in the simplified operation, but one of the existing corner points may be used as the corner point of the new mesh surface, which is not made in this embodiment of the present application. limited. For example, in FIG. 3 , only the corner point v2 and the edge associated with the corner point v2 may be eliminated, and the obtained simplified mesh face set includes four triangular mesh faces and one pentagonal mesh face.

Further, the first grid display layer obtains the second grid display layer through grid simplification processing, and the entire grid simplification processing can be implemented by multiple simplified iterative operations. 103 The obtained target geometric error of the second mesh display layer can be calculated in each or preset simplified iterative operation, and the actual geometric error of the currently obtained new mesh surface relative to the surface of the three-dimensional model of the scene and the target Whether the difference between the geometric errors is smaller than a preset threshold, if it is smaller, it is considered that the grid simplification process has been completed, and the actual geometric error of the second grid display layer obtained at this time is very close to the target geometric error, which is in line with the current application scenario. and LOD technology requirements. If it is greater than this, it is considered that the mesh simplification process has not been completed, and the next simplification iteration operation needs to be continued.

It should be noted that the preset threshold may be set according to actual requirements. For example, if the requirement for the accuracy of grid simplification processing is high, the preset threshold may be set to a smaller value.

Step 104: Store the formed new mesh surface as the second mesh display layer of the three-dimensional scene model.

After the second grid display layer is obtained, the grid and texture data contained in the second grid display layer can be stored, specifically in the directory of the 3D model of the scene, so that the second grid display layer can be used as a scene A presentation layer for a 3D model.

After the multi-layer grid display layers with decreasing levels of detail are obtained and stored according to actual requirements, the 3D model of the scene is established, and then the renderer can be used to render one or more layers of grid display layers in the 3D model of the scene. Display, in order to achieve resource allocation through multi-layer grid display layers when rendering objects in the 3D model of the scene, so as to reduce the number of faces and details of distant or non-important objects, thereby reducing resource consumption. For example, the farther the object is from the viewpoint, the mesh display layer with lower level of detail is selected to display the object.

It should be noted that the second grid display layer may be the next grid display layer of the first grid display layer, and the second grid display layer may also be several layers of grids below the first grid display layer. The presentation layer, for example, the first grid presentation layer is the first grid presentation layer of the scene 3D model, and the second grid presentation layer may be the nth grid presentation layer of the scene 3D model, where n≥2. This embodiment of the present application does not limit this.

To sum up, a method for establishing a model provided by an embodiment of the present application uses the existing physical size information and/or deviation information of sampling points in the first grid display layer to determine the geometric error in the first grid display layer. A determination is made such that the first mesh presentation layer has a measure of the level of detail using the geometric error. In the embodiment of the present application, the existing information of the sampling points in the first grid display layer obtained by the three-dimensional reconstruction of the scene is used, so that in the three-dimensional reconstruction scene, the measurement of the detail degree of the grid display layer is combined with the camera shooting process. The existing characteristics in the regions of each scene better match the actual situation that the degree of detail in different regions is different, improve the accuracy of the termination simplification conditions when the mesh surface is simplified, and reduce the use of The accuracy loss brought by the global hausdorff distance as the termination simplification condition makes it no longer rely on a known perfect display layer in the process of obtaining a multi-layer grid display layer with decreasing level of detail, eliminating the need to calculate the perfect display layer. The process reduces the calculation pressure, and because the more accurate physical size information and/or deviation information of the sampling point is used to determine the target geometric error, the detail level of the grid display layer generated when the global hausdorff distance is used as the geometric error is solved. The indicator is too large, which causes the problem of subsequent rendering accuracy decline.

4 is a specific flowchart of a method for establishing a model provided by an embodiment of the present application, and the method may include:

Step 201: Acquire the description accuracy information of the scene by the first grid display layer of the three-dimensional model of the scene.

Specifically, for step 201, reference may be made to the foregoing step 101, and details are not repeated here.

Step 202: Determine the target geometric error of the second mesh display layer of the three-dimensional scene model according to the description accuracy information.

Specifically, for step 202, reference may be made to the foregoing step 102, and details are not repeated here.

Step 203: Perform grid simplification processing on the plurality of grid surfaces in the first grid display layer to obtain new grid surfaces.

Specifically, for step 203, reference may be made to the foregoing step 103, and details are not repeated here.

Step 204: Store the formed new mesh surface as the second mesh display layer of the three-dimensional scene model.

Specifically, for step 205, reference may be made to the foregoing step 104, and details are not repeated here.

Optionally, the first grid display layer includes a plurality of first grid blocks, and each of the first grid blocks includes a plurality of grid surfaces; the target geometry of the first grid display layer The error includes a geometric error of each of the first grid blocks, and the description accuracy information of the first grid display layer includes a description accuracy value of each of the first grid blocks; the second grid The display layer includes a plurality of second grid blocks, each of the second grid blocks includes at least one grid face; the target geometric error of the second grid display layer includes each of the second grids The geometric error of the block, the description precision information of the second grid display layer includes the description precision value of each of the second grid blocks.

In the embodiment of the present application, when the LOD strategy is specifically implemented, the entire grid display layer may be divided into multiple independent grid blocks. Referring to FIG. 5 , it shows the A block diagram of a grid display layer. Among them, three grid display layers are included: grid display layer m0, grid display layer m1, grid display layer m2, and from grid display layer m0 to grid display layer m2, the level of detail decreases in order, grid display layer m0 has 16 grid blocks, the grid display layer m1 has 4 grid blocks, and the grid display layer m2 has 1 grid block, each grid block is equivalent to a node, forming a four-fork In a tree structure, the nodes between different layers have a parent-child relationship. For example, the grid display layer m2 has four child nodes, that is, the four grid blocks of the grid display layer m1. It should be noted that the quad-tree structure is only a tree-shaped structure obtained by a block operation. According to different requirements, the tree-shaped structure may also include: a binary tree structure, a ternary tree structure, an octree structure, and the like.

Therefore, in the embodiment of the present application, there are several advantages to dividing the grid display layer: First, in the actual display process of the 3D model of the scene, in most cases, only a certain local area of the scene needs to be displayed, and it is not necessary to load the entire scene. Models and blocks can alleviate this problem and reduce resource consumption; second, the level of detail in different areas is inconsistent, and blocks can separate areas with different details for more accurate subsequent display.

Specifically, the description precision information of each grid display layer may specifically include a description precision value of each grid block in the layer, that is, the description precision information may specifically be a set of description precision values of each grid block. The target geometric error of each grid display layer may specifically include the geometric error of each grid block in the layer, that is, the target geometric error may specifically be a set of geometric errors of each grid block. Moreover, since each grid block can correspond to a certain local area in the scene, the geometric error of each grid block can accurately reflect the accuracy and error of its corresponding local area. Taken together, the entire grid display The target geometric error of the layer can include the existing characteristics of each local area in the scene, which better matches the actual situation that the degree of detail in different areas is different, and further improves the termination of simplification when the mesh surface is simplified. The precision of the condition.

Optionally, grid simplification processing is performed on the grid surface of each first grid block in the first grid display layer to obtain a new grid surface in each of the first grid blocks.

Specifically, in the process of simplifying the grid block, it is necessary to calculate the folded cost of the edges in the grid block, and continuously fold the edges in the grid block according to the order of the folded cost from small to large, until the current The difference between the actual geometric error of the obtained new mesh surface relative to the surface of the three-dimensional model of the scene and the target geometric error is smaller than a preset threshold. Wherein, the folded cost of the edge of the grid block is calculated according to the distance between the corner points of the new grid surface obtained after the grid simplification process and the target grid surface, the target grid surface For the mesh face adjacent to the corner point of the new mesh face in the mesh block before the mesh simplification process.

In this embodiment of the present application, the reduced cost may actually be a value of a distance property, which can reflect the distance between the corner points of the new mesh surface obtained after the mesh simplification process and the target mesh surface. 3. Assuming that the grid block 12 and the grid block 13 are included in FIG. 3, the grid block 12 is simplified to obtain the grid block 13. In an implementation manner, the reduced cost of the grid block 13 can be The sum of the squares of the distances between the corner v3 of the new mesh face and the target mesh face (9 mesh faces in mesh block 12).

Further, for the scheme in which the iteration termination error of each grid block in the grid simplification process is a value obtained by the same hausdorff distance, the iteration termination error of each grid block in the embodiment of the present application is determined by each grid block. Obtained from the respective description precision values of the blocks, and the description precision values can be the gsd value of the largest corner point in the grid block, the product of the gsd value of the largest corner point and the preset coefficient, and the reconstruction residual of the SFM sparse point. any of the . In this way, each grid block can retain its own characteristics, which better matches the actual situation that the details of different areas in the scene are different, and reduces the accuracy loss caused by the hausdorff distance scheme.

It should be noted that the texture resolution of the texture part in each grid block can also be adjusted according to the description precision value of the grid block during the grid simplification process, so as to simplify the details of the texture part. . In addition, the relevant information of each grid block can be specifically saved in a general LOD format file, such as an Open Scene Graph (OSG, OpenSceneGraph) format file.

Optionally, the number of the second grid blocks is smaller than the number of the first grid blocks.

In this embodiment of the present application, the number of grid blocks included in the grid display model at a higher layer is greater than the number of grid blocks included in the grid display model at a lower layer, that is, the number of grid blocks The smaller the number, the lower the level of detail of the grid display layer. The number of grid blocks decreases layer by layer to meet the requirements of LOD technology to realize the 3D model of the scene including the multi-layer grid display layer with decreasing detail.

Step 205: Obtain a three-dimensional model of the scene including at least two grid display layers when the number of all grid surfaces of the second grid display layer satisfies a preset condition.

In this embodiment of the present application, the termination condition for the establishment of the 3D model of the scene under the entire LOD architecture may be that the number of all mesh faces of the last mesh display layer is less than the preset number, that is, all meshes of the mesh display layer The number of grids is small enough so that the detail level of the grid display layer is low enough. At this point, the 3D model of the scene is established. The preset quantity value may be set according to actual requirements, which is not limited in the embodiment of the present application.

Step 206: Render and display the target mesh display layer in the three-dimensional model of the scene that is within the range of the visible window of the screen.

In this embodiment of the present application, after obtaining the 3D scene model, the 3D scene model can be displayed, wherein the screen of the display device has a range of visible windows. For ease of understanding, the 3D scene model can be imagined as a larger range The user casts the visual window of the screen to watch the 3D model of the scene, and the visual window of the screen is smaller than the scene, that is, the visual window of the screen maintains a relative positional relationship with the 3D model of the scene at a moment, so that The visible window of the screen can only display a part of the 3D model of the scene. To display other areas, the relative positional relationship between the visible window of the screen and the 3D model of the displayed scene can be changed.

In this step, according to the relative positional relationship between the current three-dimensional model of the scene and the visible window of the screen, a target mesh display layer in the three-dimensional model of the scene within the range of the visible window of the screen can be determined for rendering display. It should be noted that the visual window of the screen is equivalent to the user's viewpoint, and the target grid display layer within the range of the visual window of the screen is selected for display, and the following rules can be followed: If the scene is far away, the grid display layer with the lower level of detail is selected to display the scene, thereby reducing the consumption of rendering resources by using the lower face count and detail degree of the grid display layer of the distant scene.

Optionally, step 206 may specifically include:

Sub-step 2061: Calculate the maximum pixel error of each target grid display layer on the visible window in the order from the smallest to the largest number of grid faces included, and when the maximum pixel error is less than When the target grid display layer is equal to or equal to the preset error value, the target grid display layer whose maximum pixel error is less than or equal to the preset threshold is rendered and displayed.

In the process of displaying the 3D model of the scene, the renderer can use the target geometric error δ(L) of the grid display layer as the detail level indicator for selection and display. The renderer can calculate a grid display during real-time rendering. The maximum pixel error φ(L) of the layer's projection to the screen.

φ(L)=(f×δ(L))/D, where f is the focal length of the visible window, and D is the closest distance from the viewpoint to the grid display layer. The fidelity requirements for the renderer to display according to the LOD are: φ(L)<τ, τ is a preset threshold value of constant nature, generally set to 1, then select the target grid layer that satisfies the condition of φ(L)<τ to carry out Display it.

Specifically, referring to the three grid display layers included in FIG. 5: grid display layer m0, grid display layer m1, grid display layer m2, the specific process of rendering the scene in real time by the renderer is:

First, the renderer loads the grid display layer m2 with the lowest level of detail, determines whether the grid display layer m2 is within the range of the visible window, and if it is within the visible window, calculates the maximum pixel error φ of the grid display layer m2 projected to the screen (m2), φ(m2)=(f×δ(m2))/D(m2), where f is the focal length of the visible window, and D(m2) is the closest distance from the visible window to the grid display layer m2 ( The minimum distance between all corners and the visible window in the grid display layer m2), δ(m2) is the target geometric error of the grid display layer m2.

The display fidelity requirements of LOD technology: φ<τ, τ is generally taken as 1, that is, the maximum pixel error of the grid projected to the screen is less than one pixel. So if φ(m2)<1, then the renderer will select the grid display layer m2 for rendering, and no longer traverse its child nodes; if φ(m2)>1, then the renderer will then traverse its child nodes in turn, that is, The grid displays the four block grids in the layer m1, and judges whether the grid display layer m1 is in the visible window, and if it is in the visible window, go to the next step; calculate the maximum pixel error of the grid display layer m1 projected to the screen φ(m1), the subsequent process is the same as the above-mentioned processing for the grid display layer m2, until φ<1 or the leaf node is reached. Through this rule, the grid display layer that meets the fidelity requirements can be selected and rendered.

Optionally, after step 203, the method further includes:

Step 207: Under the condition that the difference between the actual geometric error of the new mesh surface relative to the surface of the three-dimensional model of the scene and the target geometric error is greater than or equal to the preset threshold The grid surface is used as the first grid display layer, and the process goes to step 203 .

In this embodiment of the present application, the first grid presentation layer obtains the second grid presentation layer through grid simplification processing, and the entire grid simplification processing can be implemented through multiple simplified iterative operations. How to determine when to stop the iteration, this application In the embodiment, according to the target geometric error of the second mesh display layer obtained in step 203, the actual geometric error of the currently obtained new mesh surface relative to the surface of the three-dimensional model of the scene may be calculated in each or a preset simplified iterative operation. Whether the difference between the geometric error of the target and the target is less than a preset threshold, if it is less than, it is considered that the grid simplification process has been completed, and the actual geometric error of the second grid display layer obtained at this time is very close to the target geometric error, Meet the requirements of current application scenarios and LOD technology. If it is greater than that, it is considered that the mesh simplification process has not been completed, and the process goes to step 203, and the next simplification iterative operation is continued until the difference between the actual geometric error and the target geometric error is smaller than the preset threshold.

Optionally, before step 201, it may further include:

Step B1: Acquire at least one image to be reconstructed.

Step B2, constructing and obtaining the first grid display layer according to the image to be reconstructed.

In the embodiment of the present application, if the first grid display layer is the first layer in the 3D model of the scene, the method for establishing the first grid display layer may be based on the images to be reconstructed collected in the scene by the movable platform, technology to obtain the first mesh display layer, so as to use it as the mesh display layer with the highest level of detail in the three-dimensional model of the scene.

In one implementation, the structure from motion (SFM, structure from motion) technology can be used to realize the construction of the image collected by the movable platform in the scene as the first grid display layer, and the SFM technology can be captured based on the camera of the movable platform. The camera pose at the time of the image, extract the point cloud data of the image, and establish a textured mesh display layer according to the textureless mesh layer of the point cloud data component and the texture of the image. The textured mesh display layer is Can be used as the first mesh display layer with the highest level of detail.

Optionally, step B2 may specifically include:

Step B21: Determine the camera pose and texture map corresponding to each of the to-be-reconstructed images.

In the embodiment of the present application, for each image to be reconstructed, the camera pose corresponding to each image to be reconstructed can be obtained based on the internal and external parameters of the image to be reconstructed by the camera, and a texture extraction model (convolutional neural network model is sufficient) ) can obtain the texture map corresponding to each image to be reconstructed.

Among them, the internal and external parameters of the camera refer to the internal parameters of the camera (camera focal length, camera center point offset, camera distortion parameters, etc.) and external parameters (the rotation matrix and translation vector of the camera in the world coordinate system).

Step B22: Perform point cloud matching on two target images in the plurality of images to be reconstructed according to the camera posture to obtain point cloud data, and there are overlapping parts in the frames of the two target images.

In this step, according to the known camera posture, point cloud matching is performed on two target images with overlapping parts in the multi-image to be reconstructed, and point cloud data can be obtained. view stereo) technology implementation. The point cloud data in the scene refers to the collection of point data on the appearance surface of the objects in the scene obtained by the measuring instrument in the reverse engineering.

Step B23, constructing a textureless mesh display layer according to the point cloud data.

In this step, the point cloud data includes the three-dimensional coordinates of each point in the scene. Therefore, according to the point cloud data, a textureless mesh display layer reflecting the three-dimensional structure of the scene can be constructed. Specifically, this process can be obtained by the Poisson surface reconstruction technique.

Step B24: Map the textureless mesh display layer to the texture map, so as to determine the texture area corresponding to each mesh surface in the textureless mesh display layer from the texture map, and map the textureless mesh display layer to the texture map. The texture area is added to the corresponding mesh surface to obtain the first mesh display layer.

Since the textureless mesh display layer is obtained in step B23, in order to obtain the final first mesh display layer with texture, the embodiment of the present application may map the textureless mesh display layer to the texture map as a whole, so as to obtain the texture map from the texture map. Determine the texture area corresponding to each mesh surface in the textureless mesh display layer, and add the texture area to the corresponding mesh surface to obtain the first mesh display layer. At this time, the obtained first mesh The presentation layer has textures and can be rendered directly.

FIG. 6 is a block diagram of an apparatus for establishing a model provided by an embodiment of the present application. As shown in FIG. 6 , the apparatus 300 for establishing a model may include: a memory 301 and a processor 302;

The memory 301 is configured to perform: acquiring description accuracy information of the scene by a first mesh display layer of the three-dimensional model of the scene, wherein the first mesh display layer includes a plurality of surfaces used to display the three-dimensional model of the scene. Grid plane, the description accuracy information includes: physical size information of the scene corresponding to the sampling point in the first grid display layer, and/or, the position of the scene displayed by the sampling point is in the first grid Deviation information between the actual projection point in the grid display layer and the sampling point; storing the formed new grid surface as the second grid display layer of the three-dimensional model of the scene;

The processor 302 is configured to execute:

Optionally, the processor 302 is specifically configured to:

determining the target geometric error of the first mesh display layer according to the description accuracy information and the first coefficient;

A target geometric error of the second grid presentation layer is determined according to the target geometric error of the first grid presentation layer.

Optionally, the processor 302 is specifically configured to:

The target geometric error of the second grid display layer is determined according to the target geometric error of the first grid display layer and the second coefficient.

Optionally, the first grid display layer includes a plurality of first grid blocks, and each of the first grid blocks includes a plurality of grid surfaces; the target geometry of the first grid display layer The error includes a geometric error of each of the first grid blocks, and the description precision information of the first grid display layer includes a description precision value of each of the first grid blocks;

The second grid display layer includes a plurality of second grid blocks, and each of the second grid blocks includes at least one grid surface; the target geometric error of the second grid display layer includes each The geometric error of the second grid block, and the description precision information of the second grid display layer includes a description precision value of each of the second grid blocks.

Optionally, the processor 302 is specifically configured to:

Perform grid simplification processing on the grid surface of each first grid block in the first grid display layer to obtain a new grid surface in each of the first grid blocks.

Optionally, the processor 302 is specifically configured to:

Under the condition that the difference between the actual geometric error of the new mesh surface relative to the surface of the three-dimensional model of the scene and the target geometric error is greater than or equal to the preset threshold, all the new mesh surfaces As the first mesh display layer, and enter the step of performing mesh simplification processing on the plurality of mesh surfaces in the first mesh display layer to obtain a new mesh surface.

Optionally, the processor 302 is specifically configured to:

Obtaining a three-dimensional scene model including at least two grid display layers when the number of all grid surfaces of the second grid display layer satisfies a preset condition;

The target mesh display layer in the three-dimensional model of the scene within the range of the visible window of the screen is rendered and displayed.

Optionally, the processor 402 is specifically configured to:

Calculate the maximum pixel error of each target grid display layer on the visible window in the order from the smallest to the largest number of grid faces included, and when the maximum pixel error is less than or equal to a preset value When the error value is the target grid display layer, the target grid display layer whose maximum pixel error is less than or equal to the preset threshold is rendered and displayed.

Optionally, the processor 402 is specifically configured to:

Obtain at least one image to be reconstructed;

The first grid display layer is constructed and obtained according to the to-be-reconstructed image.

Optionally, the processor 402 is specifically configured to:

Determine the camera pose and texture map corresponding to each of the to-be-reconstructed images;

According to the camera posture, point cloud matching is performed on two target images in the plurality of images to be reconstructed to obtain point cloud data, and there are overlapping parts in the frames of the two target images;

According to the point cloud data, construct a textureless mesh display layer;

The textureless mesh display layer is mapped to the texture map to determine a texture area corresponding to each mesh face in the textureless mesh display layer from the texture map, and the texture The regions are added to the corresponding mesh faces to obtain the first mesh display layer.

To sum up, the apparatus for establishing a model provided by the embodiment of the present application determines the geometric error in the first grid display layer by using the existing physical size information and/or deviation information of the sampling points in the first grid display layer , so that the first mesh display layer has a measure of the degree of detail by using the geometric error. In the embodiment of the present application, the existing information of the sampling points in the first grid display layer obtained by the three-dimensional reconstruction of the scene is used, so that in the three-dimensional reconstruction scene, the measurement of the detail degree of the grid display layer is combined with the camera shooting process. The existing characteristics in the regions of each scene better match the actual situation that the degree of detail in different regions is different, improve the accuracy of the termination simplification conditions when the mesh surface is simplified, and reduce the use of The accuracy loss brought by the global hausdorff distance as the termination simplification condition makes it no longer rely on a known perfect display layer in the process of obtaining a multi-layer grid display layer with decreasing level of detail, eliminating the need to calculate the perfect display layer. The process reduces the calculation pressure, and because the more accurate physical size information and/or deviation information of the sampling point is used to determine the target geometric error, the detail level of the grid display layer generated when the global hausdorff distance is used as the geometric error is solved. The indicator is too large, which causes the problem of subsequent rendering accuracy decline.

Embodiments of the present application further provide a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, each process of the above-mentioned method for establishing a model is implemented, and can achieve the same The technical effect, in order to avoid repetition, will not be repeated here. Wherein, the computer-readable storage medium, such as read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), magnetic disk or optical disk and so on.

The memory can be an interface for connecting an external control terminal with a model-building device. For example, the external control terminal may include a wired or wireless headset port, an external power (or battery charger) port, a wired or wireless data port, a memory card port, a port for connecting a control terminal with an identification module, an audio input /Output (I/O) ports, video I/O ports, headphone ports, and more. The memory may be used to receive input (eg, data information, power, etc.) from an external control terminal and transmit the received input to one or more elements within the apparatus for modelling or may be used in the apparatus for modelling and Transfer data between external control terminals.

For example at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device.

The processor is the control center of the control terminal. It uses various interfaces and lines to connect various parts of the entire control terminal, and executes control by running or executing the software programs and/or modules stored in the memory and calling the data stored in the memory. Various functions of the terminal and processing data, so as to carry out overall monitoring of the control terminal. The processor may include one or more processing units; preferably, the processor may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface and application programs, etc., and the modem processor Mainly deals with wireless communication. It can be understood that, the above-mentioned modulation and demodulation processor may not be integrated into the processor.

The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the various embodiments may be referred to each other.

It should be understood by those skilled in the art that the embodiments of the present application may be provided as a method, a control terminal, or a computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to the present application. It will be understood that each flow and/or block in the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to the processor of a general purpose computer, special purpose computer, embedded processor or other programmable data processing terminal equipment to produce a machine that causes the instructions to be executed by the processor of the computer or other programmable data processing terminal equipment A control terminal is generated for implementing the functions specified in a flow or flow of the flowchart and/or a block or blocks of the block diagram.

These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer readable memory result in an article of manufacture comprising the instruction to control the terminal, The instruction controls the terminal to implement the function specified in one flow or multiple flows of the flowchart and/or one block or multiple blocks of the block diagram.

These computer program instructions can also be loaded on a computer or other programmable data processing terminal equipment, so that a series of operational steps are performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby executing on the computer or other programmable terminal equipment The instructions executed on the above provide steps for implementing the functions specified in the flowchart or blocks and/or the block or blocks of the block diagrams.

While the preferred embodiments of the present application have been described, additional changes and modifications to these embodiments may occur to those skilled in the art once the basic inventive concepts are known. Therefore, the appended claims are intended to be construed to include the preferred embodiment and all changes and modifications that fall within the scope of this application.

Finally, it should also be noted that in this document, relational terms such as first and second are used only to distinguish one entity or operation from another, and do not necessarily require or imply these entities or that there is any such actual relationship or sequence between operations. Moreover, the terms "comprising", "comprising" or any other variation thereof are intended to encompass non-exclusive inclusion, such that a process, method, article or terminal device comprising a list of elements includes not only those elements, but also a non-exclusive list of elements. other elements, or also include elements inherent to such a process, method, article or terminal equipment. Without further limitation, an element defined by the phrase "comprises a..." does not preclude the presence of additional identical elements in the process, method, article or terminal device comprising said element.

The application has been introduced in detail above, and specific examples are used to illustrate the principles and implementations of the application. The descriptions of the above embodiments are only used to help understand the methods and core ideas of the application; According to the idea of the present application, there will be changes in the specific embodiments and application scope for those skilled in the art. To sum up, the content of this specification should not be construed as a limitation on the present application.

Claims

A method for establishing a model, the method comprising:

Acquiring description accuracy information of the scene by the first grid display layer of the 3D model of the scene, wherein the first grid display layer includes a plurality of grid surfaces used to display the surface of the 3D model of the scene, and the description accuracy The information includes: the physical size information of the scene corresponding to the sampling point in the first grid display layer, and/or the actual projection point of the position point of the scene displayed by the sampling point in the first grid display layer deviation information from the sampling point;

The target geometric error of the second mesh display layer of the three-dimensional scene model is determined according to the description accuracy information, wherein the target geometric error is used to characterize the relative relationship between the mesh surfaces in the second mesh display layer and the The geometric error of the surface of the 3D model of the scene;

Perform grid simplification processing on the plurality of grid surfaces in the first grid display layer to obtain a new grid surface, wherein the grid simplification processing includes: folding the grid surfaces of the plurality of grid surfaces. The edge forms a new point, and the formed new point is used as the corner point of the new mesh surface, and the actual geometric error of the new mesh surface relative to the surface of the three-dimensional model of the scene is equal to the target geometric error. The difference is less than the preset threshold;

The new mesh surface formed is stored as the second mesh display layer of the three-dimensional model of the scene.
The method according to claim 1, wherein the determining the target geometric error of the second mesh display layer of the three-dimensional scene model according to the description error information comprises:

determining the target geometric error of the first mesh display layer according to the description accuracy information and the first coefficient;

A target geometric error of the second grid presentation layer is determined according to the target geometric error of the first grid presentation layer.
The method according to claim 2, wherein the determining the target geometric error of the second grid display layer according to the target geometric error of the first grid display layer comprises:

The target geometric error of the second grid display layer is determined according to the target geometric error of the first grid display layer and the second coefficient.
The method according to claim 3, wherein the second coefficient increases as the number of layers where the second mesh display layer is located in the three-dimensional model of the scene increases.
The method according to claim 2, wherein the first grid display layer comprises a plurality of first grid blocks, and each of the first grid blocks comprises a plurality of grid surfaces; the The target geometric error of the first grid display layer includes the geometric error of each of the first grid blocks, and the description accuracy information of the first grid display layer includes the description of each of the first grid blocks precision value;

The second grid display layer includes a plurality of second grid blocks, and each of the second grid blocks includes at least one grid surface; the target geometric error of the second grid display layer includes each The geometric error of the second grid block, and the description precision information of the second grid display layer includes a description precision value of each of the second grid blocks.
The method of claim 5, wherein the number of the second grid blocks is smaller than the number of the first grid blocks.
The method according to claim 5, wherein the performing grid simplification processing on the plurality of grid surfaces in the first grid display layer to obtain new grid surfaces, comprising:

Perform grid simplification processing on the grid surface of each first grid block in the first grid display layer to obtain a new grid surface in each of the first grid blocks.
The method according to claim 1, characterized in that, before storing the new mesh surface formed as the second mesh display layer of the three-dimensional model of the scene, the method further comprises:

Under the condition that the difference between the actual geometric error of the new mesh surface relative to the surface of the three-dimensional model of the scene and the target geometric error is greater than or equal to the preset threshold, all the new mesh surfaces As the first mesh display layer, and enter the step of performing mesh simplification processing on the plurality of mesh surfaces in the first mesh display layer to obtain a new mesh surface.
The method according to claim 1, wherein after storing the new mesh surface formed as the second mesh display layer of the three-dimensional model of the scene, the method further comprises:

Obtaining a three-dimensional scene model including at least two grid display layers when the number of all grid surfaces of the second grid display layer satisfies a preset condition;

The target mesh display layer in the three-dimensional model of the scene within the range of the visible window of the screen is rendered and displayed.
The method according to claim 9, wherein the rendering of the target grid display layer in the three-dimensional model of the scene within the scope of the visible window of the screen is performed, comprising:

Calculate the maximum pixel error of each target grid display layer on the visible window in the order from the smallest to the largest number of grid faces included, and when the maximum pixel error is less than or equal to a preset value When the error value is the target grid display layer, the target grid display layer whose maximum pixel error is less than or equal to the preset threshold is rendered and displayed.
The method according to claim 1, characterized in that, before acquiring the description accuracy information of the scene on the first grid display layer of the three-dimensional model of the scene, the method further comprises:

Obtain at least one image to be reconstructed;

The first grid display layer is constructed and obtained according to the to-be-reconstructed image.
The method according to claim 11, wherein the constructing and obtaining the first grid display layer according to the to-be-reconstructed image comprises:

Determine the camera pose and texture map corresponding to each of the to-be-reconstructed images;

According to the camera posture, point cloud matching is performed on two target images in the plurality of images to be reconstructed to obtain point cloud data, and there are overlapping parts in the frames of the two target images;

According to the point cloud data, construct a textureless mesh display layer;

The textureless mesh display layer is mapped to the texture map to determine a texture area corresponding to each mesh face in the textureless mesh display layer from the texture map, and the texture The regions are added to the corresponding mesh faces to obtain the first mesh display layer.
An apparatus for establishing a model, characterized in that the apparatus comprises: a memory and a processor;

The memory is used for acquiring description accuracy information of the scene by a first grid display layer of the scene 3D model, wherein the first grid display layer includes a plurality of grids used to display the surface of the scene 3D model On the other hand, the description accuracy information includes: physical size information of the scene corresponding to the sampling point in the first grid display layer, and/or, the position point of the scene displayed by the sampling point is displayed in the first grid deviation information between the actual projection point in the layer and the sampling point;

storing the formed new mesh surface as the second mesh display layer of the three-dimensional model of the scene;

The processor is used to:

The target geometric error of the second mesh display layer of the three-dimensional scene model is determined according to the description accuracy information, wherein the target geometric error is used to characterize the relative relationship between the mesh surfaces in the second mesh display layer and the The geometric error of the surface of the 3D model of the scene;

Perform grid simplification processing on the plurality of grid surfaces in the first grid display layer to obtain a new grid surface, wherein the grid simplification processing includes: folding the grid surfaces of the plurality of grid surfaces. The edge forms a new point, and the formed new point is used as the corner point of the new mesh surface, and the actual geometric error of the new mesh surface relative to the surface of the three-dimensional model of the scene is equal to the target geometric error. The difference is less than the preset threshold.
The apparatus according to claim 13, wherein the processor is specifically configured to:

determining the target geometric error of the first mesh display layer according to the description accuracy information and the first coefficient;

A target geometric error of the second grid presentation layer is determined according to the target geometric error of the first grid presentation layer.
The apparatus according to claim 14, wherein the processor is specifically configured to:

The target geometric error of the second grid display layer is determined according to the target geometric error of the first grid display layer and the second coefficient.
The apparatus according to claim 15, wherein the second coefficient increases as the number of layers where the second mesh display layer is located in the three-dimensional model of the scene increases.
The device according to claim 14, wherein the first grid display layer comprises a plurality of first grid blocks, and each of the first grid blocks comprises a plurality of grid surfaces; the The target geometric error of the first grid display layer includes the geometric error of each of the first grid blocks, and the description accuracy information of the first grid display layer includes the description of each of the first grid blocks precision value;

The second grid display layer includes a plurality of second grid blocks, and each of the second grid blocks includes at least one grid surface; the target geometric error of the second grid display layer includes each The geometric error of the second grid block, and the description precision information of the second grid display layer includes a description precision value of each of the second grid blocks.
18. The apparatus of claim 17, wherein the number of the second grid blocks is smaller than the number of the first grid blocks.
The apparatus according to claim 17, wherein the processor is specifically configured to:

Perform grid simplification processing on the grid surface of each first grid block in the first grid display layer to obtain a new grid surface in each of the first grid blocks.
The apparatus of claim 13, wherein the processor is further configured to:

Under the condition that the difference between the actual geometric error of the new mesh surface relative to the surface of the three-dimensional model of the scene and the target geometric error is greater than or equal to the preset threshold, all the new mesh surfaces As the first mesh display layer, and enter the step of performing mesh simplification processing on the plurality of mesh surfaces in the first mesh display layer to obtain a new mesh surface.
The apparatus of claim 13, wherein the processor is further configured to:

Obtaining a three-dimensional scene model including at least two grid display layers when the number of all grid surfaces of the second grid display layer satisfies a preset condition;

The target mesh display layer in the three-dimensional model of the scene within the range of the visible window of the screen is rendered and displayed.
The apparatus according to claim 21, wherein the processor is specifically configured to:

Calculate the maximum pixel error of each target grid display layer on the visible window in the order from the smallest to the largest number of grid faces included, and when the maximum pixel error is less than or equal to a preset value When the error value is the target grid display layer, the target grid display layer whose maximum pixel error is less than or equal to the preset threshold is rendered and displayed.
The apparatus according to claim 13, wherein the processor is further configured to: acquire at least one image to be reconstructed;

The first grid display layer is constructed and obtained according to the to-be-reconstructed image.
The apparatus according to claim 23, wherein the processor is specifically configured to:

Determine the camera pose and texture map corresponding to each of the to-be-reconstructed images;

According to the camera posture, point cloud matching is performed on two target images in the plurality of images to be reconstructed to obtain point cloud data, and there are overlapping parts in the frames of the two target images;

According to the point cloud data, construct a textureless mesh display layer;

The textureless mesh display layer is mapped to the texture map to determine a texture area corresponding to each mesh face in the textureless mesh display layer from the texture map, and the texture The regions are added to the corresponding mesh faces to obtain the first mesh display layer.
An electronic device, characterized by comprising a processor, a memory, and a computer program stored on the memory and running on the processor, the computer program being executed by the processor to achieve the method as claimed in claim 1 The method for building a model of any one of to 12.
A computer-readable storage medium, characterized by comprising instructions that, when executed on a computer, cause the computer to perform the method for building a model according to any one of claims 1 to 12.