WO2021238359A1

WO2021238359A1 - Method for constructing multi-level of detail model of object, and computer device

Info

Publication number: WO2021238359A1
Application number: PCT/CN2021/081695
Authority: WO
Inventors: 刘子健; 张彦峰; 康一飞
Original assignee: 华为技术有限公司
Priority date: 2020-05-28
Filing date: 2021-03-19
Publication date: 2021-12-02
Also published as: CN113743155A

Abstract

A method for constructing a multi-level of detail (LOD) model of an object, and a computer device, applicable in the construction of a high-precision map. The method comprises: nesting a digital surface model (DSM) corresponding to a satellite image and outline vectors of buildings in the satellite image; then performing sampling on the basis of the nested DSM and outline vectors of the buildings to obtain discrete 3D point clouds of the buildings; and constructing the buildings using the 3D point clouds, wherein the 3D point clouds are obtained by performing sampling on the basis of the outlines of the buildings, and thus can highly precisely restore top surface information of the buildings, such that a high-precision multi-LOD model can be obtained. The method does not require individual one-by-one acquisition of height information of each building, can directly use an automatic construction process to reduce workload and a manufacturing period, thus effectively reducing production costs of an LOD model; in addition, the method ensures the precision of a complicated building model, and implements quick production of a low-cost city-scale high-precision map.

Description

Method and computer equipment for constructing multi-level of detail model of object

This application claims the priority of the Chinese patent application filed with the Chinese Patent Office on May 28, 2020, the application number is 202010466779.1, and the application title is "a method and computer equipment for constructing a multi-level-of-detail model of an object", and its entire contents Incorporated in this application by reference.

Technical field

This application relates to the field of map construction, and in particular to a method and computer equipment for constructing a multi-level-of-detail model (digital surface model, DSM) of an object.

Background technique

With the advent of the fifth-generation mobile communication technology (5th-generation, 5G) era, high-precision maps, as a bridge connecting the virtual world and the real world, will play an important role in various fields and have broad application prospects. Human driving, positioning and navigation, augmented reality (AR), virtual reality (VR), 5G simulation, etc. With the increasing demand for high-precision maps, various production solutions for high-precision maps have been proposed. Among them, solutions based on drone images and laser data are limited by cost and cannot be promoted on a large scale; production solutions based on satellite images (also known as satellite images) are still the current mainstream solutions.

In the production plan based on satellite imagery, the level of detail (LOD) model of the building is the basis for constructing high-precision maps, that is, maps based on satellite imagery are composed of multiple associated satellite images. The LOD model of each building is obtained after merging. Therefore, if a good LOD model of the building can be constructed, it means that a good satellite map can be constructed in the end. At present, the LOD model of buildings based on satellite images is generally: first perform a regional network adjustment on the satellite images to obtain the aligned satellite images, and then import the aligned satellite images into the stereo mapping software, and pass the front intersection The method measures the height of each building in the satellite image one by one (that is, one building corresponds to a height), and then based on the obtained height information and contour vector of each building, use related software (or algorithm) along the building The contours "pull up" the buildings according to their respective heights, thereby obtaining the LOD model of each building in a certain satellite image.

However, since each building in the above method corresponds to a height value, the buildings under the LOD model are flat-topped after being "pulled up", and most of the current buildings are not completely flat-topped, such as " For “herringbone” roofs, “spires” buildings, etc., the LOD model obtained by the above method will lose the top profile information of the building, resulting in loss of accuracy.

Summary of the invention

The embodiments of the application provide a method and computer equipment for constructing a multi-level-of-detail model of an object, which are used to construct an LOD model through the acquired 3-dimension (3-dimension, 3D) point cloud of a building, because the 3D point cloud is based on The contour of the building is sampled, and the top surface information of the building can be restored with high precision, and a high-precision LOD model can be obtained based on this.

Based on this, the embodiments of the present application provide the following technical solutions:

In the first aspect, the embodiments of the present application first provide a method for constructing a multi-level-of-detail model of an object, which can be used in high-precision map construction. Meter satellite image) corresponding to the DSM and the outline vector of each building in the satellite image, and the DSM and the outline vector of each building in the satellite image are combined to obtain the combined result. The result of the fit (ie, the overlapped DSM and the contour vector of each building) sample each building in the current satellite image to obtain the 3D point cloud of each building in the satellite image. Finally, the computer equipment can be based on the obtained various buildings. The 3D point cloud of the building constructs each building in the current satellite image to obtain the LOD model of each building in the satellite image. Similarly, the computer device can obtain the LOD model of each building in each satellite image in the satellite image collection (including the collection of multiple satellite images) in the above-mentioned manner.

In the above-mentioned embodiment of this application, the DSM corresponding to the satellite image and the contour vector of each building in the satellite image are combined for the first time, and then the 3D point of each building is obtained based on the combined DSM and building contour vector Cloud, that is to say, the fit result obtained according to the DSM and the silhouette vector of the building defines the sampling area of the 3D point cloud (ie, the sampling range is delineated), and the sampling of the 3D point cloud is performed in this sampling area, which can be high The top surface information of the building is restored with precision, and a high-precision LOD model can be obtained based on this. In addition, the foregoing embodiments of the present application do not need to obtain the height information of each building individually, which saves costs and reduces the amount of data at the same time.

In combination with the first aspect of the embodiments of the present application, in the first implementation of the first aspect of the embodiments of the present application, each building in the satellite image is constructed according to the 3D point cloud, and the LOD model of each building in the satellite image is obtained. The method can be through, but not limited to, the following methods: first reconstruct the grid model of each building in the satellite image from the 3D point cloud according to the first preset algorithm, and then simplify the grid model according to the second preset algorithm to obtain the satellite image Simplified LOD model of each building in.

In the above-mentioned embodiment of the present application, a method of how to obtain the LOD model is specifically explained, that is, the mesh model is first obtained according to the first preset algorithm, and then the mesh model is simplified to obtain the LOD model. This method is simple, easy to operate, and flexible.

In combination with the first implementation manner of the first aspect of the embodiments of the present application, in the second implementation manner of the first aspect of the embodiments of the present application, the second preset algorithm may have multiple specific manifestations, for example, vertex deletion The method can also be a method in which the side becomes a point (that is, the side collapses), or a method in which the surface becomes a point (that is, the area shrinks), which is not specifically limited here. For ease of understanding, taking the second preset algorithm as the edge collapse method (ie, edge folding algorithm) as an example, how to simplify the mesh model is introduced: First, the mesh model of each building is obtained according to the edge folding algorithm. Iterative simplification until the LOD model of each building in the simplified satellite image meets the preset conditions, then the simplified LOD model obtained at this time is the final output LOD model.

In the above-mentioned embodiment of this application, a method of how to simplify the mesh model is specifically explained, that is, the obtained mesh model is simplified by the edge folding algorithm, and a self-checking mechanism (ie, setting preset conditions) is added to prevent LOD. The model is oversimplified, while taking into account the simplification ratio and model accuracy.

Combining the first aspect of the embodiments of the present application, the first implementation to the second implementation of the first aspect, in the third implementation of the first aspect of the embodiments of the present application, the current satellite image Sampling of each building to obtain the 3D point cloud of each building in the satellite image can be specifically but not limited to the following methods: each building in the current satellite image is followed by a certain step on the corresponding DSM along its outline. (step) Sampling 3D points, so that each building can discretize the corresponding 3D point cloud, where step = a*R, a is a preset coefficient, which can be set according to the actual situation, if you want to get For dense 3D point clouds, a can be set smaller, such as 0.3; if you want to get a less dense 3D point cloud, you can set a larger, such as 0.9. The specifics are not limited here, R It is the image resolution of the satellite image.

In the above-mentioned embodiments of this application, it is specifically explained how to obtain discrete 3D point clouds according to the contours of buildings. The density of the 3D point clouds can be adjusted according to the step length, which can meet the different needs of users and is flexible.

In combination with the first aspect of the embodiments of the present application and the first to third implementation manners of the first aspect, in the fourth implementation manner of the first aspect of the embodiments of the present application, each building in the multiple satellite images The LOD model is partitioned and merged to obtain the target map.

In the above-mentioned embodiment of this application, after obtaining the LOD model of each building in each satellite image in the satellite image collection, the LOD model of each building in the satellite image collection can be partitioned and merged to obtain a high-precision target map. The map can be further sent to end-side devices (such as mobile phones, tablet computers, etc.) or edge devices for use, which is practical.

In combination with the first aspect of the embodiments of the present application and the first implementation to the fourth implementation of the first aspect, in the fifth implementation of the first aspect of the embodiments of the present application, the contour vector of each building in the satellite image is Save it as a shp file. Specifically, it includes but is not limited to the following methods: 1) The outline of each building obtained based on the satellite image can be stored as a shp file, and a satellite image corresponds to a shp file. For example, there are 4 buildings in a satellite image, then The obtained contour vectors of the 4 buildings are all stored in the same shp file; 2) Each building corresponds to a shp file. For example, if there are 4 buildings in a certain satellite image, then these 4 buildings are obtained The contour vectors of are saved as 4 different shp files. The specific implementation of storing the outline vector of each building in each satellite image as a shp file is not limited here.

In the above-mentioned embodiments of this application, several ways of saving the building as a shp file are described, which are optional.

With reference to the first aspect of the embodiments of the present application and the first to fifth implementation manners of the first aspect, in the sixth implementation manner of the first aspect of the embodiments of the present application, the building includes at least the following objects At least one of: houses, bridges, electrical towers, tunnels, iron towers, water towers, iconic sculptures, dams, communication base stations, etc. Further, as long as it is a target determined by people in satellite images, and extract the outline vector of the target , Then the target can be called the building described in the embodiment of this application.

In the foregoing embodiments of the present application, it is explained which objects the buildings described in the present application can be, so that when the outline of the building is acquired, it is convenient to know how to determine the required building type in the satellite image.

With reference to the first implementation to the sixth implementation of the first aspect of the embodiments of the present application, in the seventh implementation of the first aspect of the embodiments of the present application, the first preset algorithm includes: a Poisson surface reconstruction algorithm.

In the foregoing implementation manners of the present application, a specific form of the first preset algorithm is given, which is optional.

A second aspect of the embodiments of the present application provides a computer device that has the function of implementing the foregoing first aspect or any one of the possible implementation methods of the first aspect. This function can be realized by hardware, or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above-mentioned functions.

The third aspect of the embodiments of the present application provides another computer device, which may include a memory, a processor, and a bus system, where the memory is used to store a program, and the processor is used to call the program stored in the memory to execute the first embodiment of the present application. Aspect or any one of the possible implementation methods of the first aspect.

The fourth aspect of the present application provides a computer-readable storage medium that stores instructions in the computer-readable storage medium. When the instructions are run on a computer, the computer can execute the first aspect or any one of the first aspects. Way of realization.

The fifth aspect of the embodiments of the present application provides a computer program or computer program product. When the computer program or computer program product is run on a computer, the computer executes the first aspect or any one of the possible implementation manners of the first aspect. method.

The sixth aspect of the embodiments of the present application provides a chip. The chip includes at least one processor and at least one interface circuit, the interface circuit is coupled to the processor, and the at least one interface circuit is used to perform a transceiver function and send instructions to At least one processor, at least one processor is used to run a computer program or instruction, which has the function of realizing the method of the above-mentioned first aspect or any one of the possible implementation manners of the first aspect. This function can be realized by hardware or software Realization can also be realized by a combination of hardware and software. The hardware or software includes one or more modules corresponding to the above-mentioned functions. In addition, the interface circuit is used to communicate with other modules outside the chip. For example, the interface circuit can send the LOD model of the building obtained by the on-chip processor to end-side devices (such as mobile phones, personal computers, smart phones). Watches, etc.) or cloud-side devices (such as cloud servers, clusters, etc.).

Description of the drawings

FIG. 1 is a schematic diagram of three mesh models of the same target 3D object with different accuracy provided by an embodiment of the application;

2 is a schematic diagram of a mesh model in which each triangle is divided into 4 new triangles according to an embodiment of the application;

FIG. 3 is a schematic diagram of the working principle of the LOD model provided by an embodiment of the application;

4 is a schematic diagram of a folding process of the edge folding algorithm provided by an embodiment of the application;

Figure 5 is a schematic diagram of a DSM provided by an embodiment of the application;

Figure 6 is a schematic flow diagram of a mainstream solution for building a LOD model of a building based on satellite images;

Fig. 7 is a schematic diagram of the initial input satellite image;

Figure 8 is a schematic diagram of measuring the height of a certain building in a satellite image by using stereo mapping software;

Fig. 9 is a schematic diagram of the outline of the building corresponding to the satellite image;

Figure 10 is a schematic diagram of saving a building outline vector as a shp file;

Figure 11 is a schematic diagram of a city-level high-precision map obtained by segmenting and merging the LOD models of various buildings;

FIG. 12 is a schematic flowchart of a method for constructing an object's LOD model provided by an embodiment of the application;

Figure 13 is a schematic diagram of a semi-globally optimized DSM extraction method;

Fig. 14 is a schematic diagram of a building contour recognition method based on artificial intelligence (AI);

15 is a schematic diagram of another flow chart of the method for constructing an object's LOD model provided by an embodiment of the application;

FIG. 16 is a schematic diagram of a vertex deletion method provided by an embodiment of the application to achieve the purpose of simplifying the mesh model;

FIG. 17 is a schematic diagram of a method of area shrinkage provided by an embodiment of the application to achieve the purpose of simplifying the mesh model;

FIG. 18 is a schematic diagram of the LOD model at different degrees of simplification provided by the embodiments of the application; FIG.

FIG. 19 is a schematic diagram of a flow chart of implementing mesh model simplification through an edge folding algorithm provided by an embodiment of the application; FIG.

FIG. 20 is a schematic diagram of a method for constructing a LOD model of an object provided by an embodiment of the application;

21 is a schematic diagram of a high-precision map obtained by merging the LOD models of various buildings according to an embodiment of the application;

FIG. 22 is a schematic diagram of comparison between the LOD model obtained by the prior art and the LOD model obtained by the solution of the present application;

FIG. 23 is a schematic diagram of an application scenario provided by an embodiment of the application;

FIG. 24 is a schematic diagram of a computer device provided by an embodiment of the application;

FIG. 25 is another schematic diagram of a computer device provided by an embodiment of this application.

Detailed ways

The embodiments of the present application provide a method and computer equipment for constructing a multi-level-of-detail model of an object, which are used to construct an LOD model from the acquired 3D point cloud of a building. Since the 3D point cloud is sampled based on the outline of the building, It can restore the top surface information of the building with high precision, and obtain a high-precision LOD model based on this.

The terms "first", "second", etc. in the description and claims of the application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific sequence or sequence. It should be understood that the terms used in this way can be interchanged under appropriate circumstances, and this is only used to describe the method of distinguishing objects with the same attribute in the description of the embodiments of the present application. In addition, the terms "include" and "have" and any variations of them are intended to cover non-exclusive inclusion, so that a process, method, system, product or device containing a series of units is not necessarily limited to those units, but may include Listed or inherent to these processes, methods, products, or equipment.

The embodiments of the present application involve a lot of relevant knowledge about the processing of satellite images. In order to better understand the solutions of the embodiments of the present application, the following first introduces related terms and concepts that may be involved in the embodiments of the present application.

Satellite image: It can also be called satellite image, satellite image, satellite image, etc. It refers to the map-like photos of the earth or other planets taken by the photography equipment mounted on the artificial satellite. At present, there are generally two ways to acquire the original satellite image. The first is to use photographic equipment to shoot the film, place the photographic equipment with the film on the satellite, and send the satellite into the designed orbit to take pictures of the ground. After the photography is completed, the satellite is recovered, and the negative film is obtained after a series of photographic processing, which can be digitized by image scanning to obtain digital satellite images; the second is "digital imaging", and the imaging principle is similar to that of a digital camera.

Regional network adjustment: Adjustment refers to the reasonable allocation of accidental errors of observations, pre-correction of systematic errors of observations, and the use of certain observation principles and manual methods to control the gross errors of observations. . In the general adjustment algorithm, the control network is fixed on the known data based on the known starting data. The regional network adjustment refers to dividing the area with the largest area not exceeding the number of image pairs, and performing the adjustment according to the divided area as a unit.

Stereo mapping: generally refers to the simulation method of stereo mapping, that is, using two optical or mechanical projectors, or optical-mechanical projectors, the transparent aerial photograph is installed in the projector, illuminated with light, and simulated shooting overshoot. Reconstruct a reduced three-dimensional model similar to the actual field. The measurement on this three-dimensional model is equivalent to the measurement of the original object. The results obtained can be directly drawn on the drawing table by mechanical or gear transmission to various topographic maps or topics The basic principle of maps and stereo mapping is the geometric reversal of the photographic process.

Spatial front intersection: It can also be called the spatial front intersection of the stereo pair. It refers to the method of restoring the beam of the stereo pair during photography and establishing the geometric model, and then using the intersection of the light rays of the same name to determine the spatial position of the model point. Specifically, the stereo The method of determining the object space coordinates of the point (the coordinates in a tentative three-dimensional coordinate system or the coordinates of the ground measurement coordinate system) by the internal and external azimuth elements of the left and right images and the image coordinate measurement values of the image point with the same name .

Nesting: Perform geometric matching of the related data of two or more images of the same target, so that the details or pixel data of each image can be overlapped together. The images or image data to be nested can be photos or data of each spectrum captured by a multi-spectral camera or scanner at the same time period or different time periods, or data collected by different sensors.

Mesh model: It can also be called Mesh model, or, 3D mesh model. Mesh model refers to the unified expression form of a 3D object. Each 3D object is presented in the form of a mesh model. Mesh model It is spliced by polygons (commonly used quadrilaterals and triangles), and a complex polygon is actually spliced by multiple triangular faces. In the embodiments of the present application, the mesh model is a mesh model composed of triangles as an example for illustration. As shown in Figure 1, there are three mesh models with different precisions of the same target 3D object "dog". The so-called different precision refers to the same 3D object model, and the high precision refers to more faces (such as , The number of triangles) is constructed into the 3D object. On the contrary, the low precision means that the 3D object is constructed from fewer faces. As shown in Figure 2, a mesh model is divided into four by each original triangle. With the new triangle, the total number of faces has been increased by 4 times, and the accuracy has also been improved.

Multi-level of detail model: LOD model, a real-time three-dimensional computer graphics technology, its working principle is: when the viewpoint is close to the object, the model can be observed with rich details; when the viewpoint is far away from the model, the observed details are gradually blurred. Refer to the example diagram shown in Figure 3. The system drawing program selects the corresponding details to display according to certain judgment conditions, thus avoiding the time waste caused by drawing those details with relatively little meaning, and effectively harmonizing the relationship between the continuity of the picture and the resolution of the model.

Poisson surface reconstruction algorithm: It is an implicit function surface reconstruction method. For example, an object can be represented by an indicator function whose outside is 1 and 0 inside the object, and the surface is obtained by solving this function and then performing isosurface extraction. The process of solving this function is the process of constructing a Poisson equation and solving the Poisson equation. This algorithm is usually used to reconstruct a discrete 3D point cloud to reconstruct a watertight triangular mesh model.

Edge folding algorithm: It can also be called an edge collapse algorithm or an edge collapse simplification algorithm. It is a type of geometric element deletion method. Its essence is vertex deletion, also called edge collapse. As shown in Figure 4, Figure 4 is a schematic diagram of the folding process. Each time it is simplified, a directed edge x (the dotted arrow in Figure 1 indicates the direction) and two related points (p1, p2) are selected through the algorithm. ), "fold" one of the points p1 to p2, then modify the topological relationship, map the edges related to p1 to p2, and finally complete the simplified operation. One simplification can reduce 1 edge and 2 faces of the source model. In Figure 4, the edge folding algorithm folds the edge x and the point p1, the triangle ①② disappears from the original mesh model after folding, and the triangle ③④⑤⑥The original point with p1 as the vertex is changed to p2 as the vertex. . The advantage of the edge folding algorithm is that it can generate continuous levels of detail and has a corresponding method of processing texture information.

Digital surface model (DSM): refers to a ground elevation model that includes the heights of buildings, bridges, and trees on the ground. Compared with the digital elevation model (DEM), DEM only includes the elevation of the terrain The information does not contain other surface information. DSM is based on DEM and further covers the elevation of other surface information besides the ground. As shown in Figure 5, a schematic diagram of DSM is illustrated. It has been widely used in some areas where building height is required. DSM represents the most true expression of ground ups and downs, and can be widely used in all walks of life. For example, in forest areas, it can be used to detect the growth of forests; in urban areas, DSM can be used to check the development of cities; especially the well-known cruise missiles require not only digital ground models, but also digital surface models. Only in this way is it possible to make it possible for the cruise missile to surrender to the mountains at every mountain and forest to the forest at every forest during its low-altitude flight.

In addition, before introducing the embodiments of the present application, a brief introduction to the current common ways of constructing the LOD model of each building in the satellite image will be given to facilitate the understanding of the embodiments of the present application later.

Please refer to Figure 6 for details. Figure 6 is a mainstream solution for building LOD models based on satellite images. The process is as follows: the initial input data is satellite images (as shown in Figure 7), and a regional network is performed for each group of satellite images. Adjust to obtain the aligned satellite image (the aligned satellite image optimizes the camera parameters of the camera equipment, and the satellite image itself has no influence); then, import the aligned satellite image into the relevant stereo mapping software and pass the front The rendezvous method can measure the height of each building (measured one by one) in each satellite image (as shown in Figure 8); in addition, it is also necessary to obtain the building outline corresponding to the input satellite image (as shown in Figure 9). As shown), the building outline is usually drawn manually based on the corresponding satellite image, that is, it can be constructed according to each building in each satellite image. Now it is generally purchased directly from a third party. The purchased satellite image and the satellite image The building outlines in is already marked. The building outlines obtained based on each satellite image can be stored as a shp file. A satellite image corresponds to a shp file. For example, there are 3 buildings in a satellite image. The outline vectors of these three buildings will be obtained and stored in the shp file; after that, the height of each building in each satellite image obtained above is also stored in the shp file (as shown in Figure 10); finally based on The contour and height information of each building in the shp file corresponding to each satellite image. Use related software (or algorithm) to "pull up" all the contoured buildings in a certain satellite image at one time along the contour of the building. Obtain the LOD model corresponding to each building (for example, as shown in the right part of Figure 11, showing two buildings "pull up" according to their respective heights), each building outline in each satellite image performs the above operations , And then merge into blocks to finally get a city-level high-precision map, as shown in the left part of Figure 11.

However, the above-mentioned mainstream solution for building LOD models based on satellite images has at least the following defects: 1) The cost is high. This solution needs to use relevant software or algorithms to measure each satellite image one by one in the entire production process. The height of the building increases the time cost and labor cost of data production; 2) Loss of accuracy, as can be seen from Figure 11, due to program restrictions, these models are flat-topped. For some complex buildings, such as " "Herringbone" roofs, "spire" buildings, etc., the LOD model produced by the above solution will lose the top profile information of the building, resulting in loss of accuracy.

In order to solve the above-mentioned problem, the embodiment of the present application first provides a method of constructing an object's LOD model, which is used to construct the LOD model from the acquired 3D point cloud of the building, because the 3D point cloud is completely based on the building The contour is sampled, and the top surface information of the building can be restored with high precision, and a high-precision LOD model can be obtained based on this.

The embodiments of the present application will be described below in conjunction with the drawings. A person of ordinary skill in the art knows that with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are equally applicable to similar technical problems. Please refer to FIG. 12 for details. FIG. 12 is a schematic flowchart of a method for constructing a LOD model of an object according to an embodiment of the application, which specifically includes:

1201. Acquire a DSM corresponding to a satellite image and a contour vector of each building in the satellite image.

First, the computer equipment will obtain the DSM corresponding to a certain target satellite image (for example, a satellite image with a resolution of 0.5 meters) and the contour vector of each building in the satellite image. In the implementation of this application, how the computer equipment obtains the DSM corresponding to the satellite image and the outline vector of each building in the satellite image are all mature solutions in the industry. The details are not repeated here, including but not limited to: computer The device can obtain the DSM corresponding to the satellite image through a semi-globally optimized DSM extraction method as shown in Figure 13, and then obtain the satellite through an AI-based building contour recognition method as shown in Figure 14. The outline vector of each building in the image. It should be noted here that the satellite image mentioned in this application refers to the satellite image that has been adjusted and aligned through the regional network.

It should be noted that in some embodiments of this application, the outline vector of each building in each satellite image can be stored as a shp file, including but not limited to the following methods: 1) Each building obtained based on the satellite image The outline can be stored as a shp file, and a satellite image corresponds to a shp file. For example, if there are 4 buildings in a satellite image, the obtained outline vectors of these 4 buildings are all stored in the same shp file; 2) Each building corresponds to a shp file. For example, if there are 4 buildings in a certain satellite image, the obtained outline vectors of the 4 buildings are stored as 4 different shp files. The specific implementation of storing the outline vector of each building in each satellite image as a shp file is not limited here. It should be noted that, in some embodiments of the present application, in addition to storing the outline vector of each building in the satellite image as a shp file, it can also be stored as other files, and the details are not limited here.

It should also be noted that the buildings described in the embodiments of the present application may include at least one of the following objects: houses, bridges, electrical towers, tunnels, iron towers, water towers, iconic sculptures, dams, communication base stations, etc., and further , As long as it is a target determined by people in the satellite image and the contour vector of the target is extracted, then the target can be called the building described in the embodiment of the present application.

1202. The DSM is combined with the contour vector of each building in the satellite image to obtain a combined result.

After the computer device obtains the DSM of the current satellite image and the contour vector of each building in the satellite image, it will fit the DSM with the contour vector of each building in the satellite image to obtain the fit result. Since the contour vectors of DSM and each building are extracted based on the satellite image after the same adjustment, the two can naturally fit together strictly, that is, based on the sum of the contour vectors of each building in the same satellite image. The corresponding DSM is overlapped, and the overlapped DSM and the contour vector of each building are the result of the integration.

1203. Sampling each building in the satellite image according to the result of the nesting, to obtain a 3D point cloud of each building in the satellite image.

After the computer equipment obtains the set result, it can sample each building in the current satellite image according to the set result (ie the overlapped DSM and the contour vector of each building) to obtain the 3D of each building in the satellite image Point cloud.

Specifically, in some embodiments of the present application, sampling each building in the current satellite image according to the result of the set-up to obtain the 3D point cloud of each building in the satellite image can be specifically, but not limited to, the following methods: Each building in the image is sampled with a certain step size (step) along its contour on the corresponding DSM 3D points, so that each building can be discretized out of the corresponding 3D point cloud, where step = a *R and a are preset coefficients, which can be set according to the actual situation. If you want to get a dense 3D point cloud, you can set a smaller, such as 0.3; if you want to get a less dense 3D point cloud , Then a can be set larger, such as 0.9, which is not specifically limited here, and R is the image resolution of the satellite image.

1204. Construct a LOD model of each building in the satellite image according to the 3D point cloud.

Finally, the computer equipment can construct each building in the current satellite image according to the obtained 3D point cloud of each building, so as to obtain the LOD model of each building in the satellite image. Similarly, the computer device can obtain the LOD model of each building in each satellite image in the satellite image collection (including the collection of multiple satellite images) in the above-mentioned manner.

It should be noted that, in some embodiments of this application, after the satellite image collection obtained by the computer equipment obtains the LOD model of each building in each satellite image, the LOD model of each building in the satellite image collection can be partitioned and merged to obtain A high-precision target map, which can be further sent to end-side devices (such as mobile phones, tablet computers, etc.) or edge devices for use.

It should also be noted that the computer device described in this application can be a cloud-side device (eg, cloud server, cluster, etc.), or an end-side device (eg, mobile phone, personal computer, etc.), as long as it can execute this application The device corresponding to each step in the embodiment in FIG. 12 can be referred to as a computer device, and the specific expression form of the computer device is not specifically limited herein.

For ease of understanding, the following takes Fig. 15 as an example to illustrate the method of constructing the LOD model of the object described in the above embodiment: First, select a target satellite image (may be called a satellite image S), based on the satellite image S Extract the outline vector of each building in the DSM and the satellite image S (for example, the outline vector of each building can be saved as a shp file), and then the DSM and the outline vector of each building (for example, the corresponding shp File) to fit, get the fit result shown on the right side of Figure 15, and then based on the fit result, sample the discrete 3D point cloud according to the contour of each building, Figure 15 selects a building in DSM as the target The building (may be called building z) is shown to obtain the discrete 3D point cloud of building z, and finally the building z is constructed according to the 3D point cloud of building z to obtain the LOD model of building z. For example, the LOD model can be reconstructed from the 3D point cloud through the Poisson surface reconstruction algorithm, and the reconstructed model has good water tightness and geometric surface characteristics. Similarly, for other buildings in the satellite image S, the LOD model corresponding to each building in the satellite image S is obtained according to similar processing to the building z.

It should be noted that, in the above-mentioned embodiment of the present application, each building in the satellite image is constructed according to the 3D point cloud, and the LOD model of each building in the satellite image is obtained, which is essentially a high-precision grid model. Because all the 3D point clouds obtained for each building are used to construct the LOD model of the building, although the LOD model of the building obtained in this way has high accuracy, it also has the problem of large amount of data. Eventually, the amount of synthesized satellite map data may be too large, causing freezes during use and affecting user experience. Therefore, in some embodiments of the present application, each building in the satellite image is constructed according to the 3D point cloud, and the way to obtain the LOD model of each building in the satellite image can be through, but not limited to, the following methods: first according to the first preset The algorithm reconstructs the grid model of each building in the satellite image from the 3D point cloud (that is, the above-mentioned unsimplified LOD model), and then simplifies the grid model according to the second preset algorithm to obtain a simplified model of each building in the satellite image LOD model.

It should also be noted that in some embodiments of the present application, the first preset algorithm may include, but is not limited to: Poisson surface reconstruction algorithm.

The following specifically introduces how to simplify the mesh model according to the second preset algorithm. It needs to be explained here that the second preset algorithm can have a variety of specific manifestations, for example, the vertex deletion shown in Figure 16 The method can also be a method in which the side becomes a point (that is, the side collapses) as shown in FIG. 4, or a method in which the surface becomes a point (that is, the area shrinks) as shown in FIG. 17, which is not specifically limited here. . However, for ease of understanding, take the method of edge collapse (ie edge folding algorithm) shown in Figure 4 as an example for the second preset algorithm to introduce how to simplify the mesh model: The mesh model of the object is iteratively simplified until the LOD model of each building in the simplified satellite image meets the preset conditions. By setting the preset conditions, both simplification ratio and model accuracy are taken into consideration. The core technical idea is: the simplification of the mesh model is achieved through the edge folding algorithm (as shown in Figure 4), and a self-checking mechanism (ie, setting preset conditions) is added to prevent excessive simplification of the LOD model, as shown in Figure 18, that is Because the LOD model is "pulled" due to over-simplification, the building in this case has been severely distorted and cannot be used.

Specifically, the steps to realize the simplification of the grid model through the edge folding algorithm can be shown in Figure 19: First, calculate the collapse cost function of the obtained grid model of each building in the current satellite image, and the collapse cost function can be expressed as Is cost=|S ₀ -S ₁ |, where S ₀ is the local area enclosed by the triangles associated with the current side x (that is, the area enclosed by the 9 triangles on the left in Figure 4), S ₁ To assume that after the edge x collapses to vertex p2, the local area enclosed by the triangles associated with vertex p2 (that is, the area enclosed by the 7 triangles on the right in Figure 4), cost is the local area before and after the collapse The amount of change; similarly, for each edge in the grid model, the value of the cost corresponding to each edge can be calculated according to the above collapse cost function, and then the edge corresponding to one cost is selected for collapse according to the preset rules (That is, one edge collapses into a vertex). For example, the edge with the smallest cost value can be selected for collapse, or the edge with the cost value less than a certain value can be selected for collapse. Make a limit; in a similar way to the above, after each iteration a certain step length (for example, 10-20 times, each cycle continues to collapse 10-20 times in the above way, that is, 10-20 edges are collapsed), the model is The self-checking calculation is used to determine whether the model is "pulled". Specifically, the self-checking calculation can be carried out by, but not limited to, the following methods: Calculate the 3D point cloud that constitutes the initial mesh model (that is, the 3D point cloud that is not performed on the left side of Figure 4 above) The distance d from each vertex at the time of collapse) to the current simplified LOD model (that is, the area formed by the 7 triangle rows on the right side of Figure 4, which can be approximated as a plane). The specific calculation formula can be d=μ+2*σ, Among them, μ=∑ _i=1～n d _i /n, which is the average value _{of the distance d i} from each 3D point to the current simplified LOD model,

_{It is the standard deviation of the distance d i} from each 3D point to the current simplified LOD model; after that, it is judged whether the distance between the current simplified LOD model and the point cloud exceeds the threshold value t of "drawing", t=ε*step, where, ε is the scale factor, which can be set according to the actual situation. According to the experience of the experiment, the value can be any value from 2 to 5. The step is the sampling step of the 3D point cloud. The sampling step can also be set by yourself, for example, The possible step size is step=0.5*0.5m; if the calculated distance d does not exceed the threshold t, then return to the above steps to continue the iterative simplification of the next cycle. If the calculated distance d exceeds the threshold t, stop the simplification. And output the LOD model obtained from the last simplification cycle iteration as the final simplified LOD model.

For ease of understanding, the following takes Figure 20 as an example to illustrate the LOD model finally simplified from the initial satellite image: input the initial satellite image (for example, a satellite image with a resolution of 0.5 meters) into a computer device, and then , Using a preset method (such as the semi-global optimization method shown in Figure 13) to extract DSM from the satellite image, while using a preset method (such as the AI recognition method shown in Figure 14) to extract the satellite image The outline of each building and the shp file is generated, and then the DSM and the shp file of the building are combined, and then each building is set along its outline on the corresponding DSM according to a certain step (eg, step= 0.5*0.5 meters, other steps are also possible) sample 3D points, discretize the 3D point cloud, and finally reconstruct a high-precision 3D mesh model from the 3D point cloud through the Poisson surface reconstruction algorithm (other algorithms are also available); then , For the grid model corresponding to each building, get the threshold t according to the distance of 1 meter (5*step) (or other distances), and then perform the simplified steps shown in Figure 20 to generate the final simplified LOD model . Finally, the obtained LOD models of each building can be merged into zones to obtain a city-level high-precision map. 21 It can be seen that it restores the top information of each building very well.

In the above-mentioned embodiments of this application, the automated construction process can reduce labor and production cycles, effectively reduce the production cost of LOD models, and at the same time ensure the accuracy of complex building models, and achieve high-precision, low-cost city-level high-precision maps Rapid production.

In order to have a more intuitive understanding of the beneficial effects brought by the embodiments of the present application, please refer to FIG. 22. In FIG. 22, the LOD model of the building obtained from the same satellite image is the building obtained by the embodiment of the present application. The accuracy of the LOD model is much higher than the accuracy of the LOD model of the building obtained by the current existing technology.

Since the high-precision map constructed based on the LOD model of the building in the embodiment of this application can be used in the fields of smart security, smart city, smart terminal, etc., for example, the LOD model of the building obtained based on the method of this application can be used to construct high-precision maps. Accuracy map. This high-precision map can be applied to various scenarios, such as some common tasks: unmanned obstacle avoidance, intelligent recognition, positioning navigation, further construction of AR/VR maps, etc. Introduction of multiple application scenarios.

(1) AR map

AR map (also known as River Map) has the ability to integrate 3D high-precision maps, spatial computing capabilities, strong environment understanding capabilities, and realistic virtual and real fusion rendering capabilities. Under the 5G architecture of terminal, tube and cloud fusion, it will provide an earth-level virtual and real world. Construction and service capabilities. The high-precision map, as a bridge between the virtual world and the real world, is an important foundation for AR maps. The technical solution proposed in this application has been verified, and it has been proved that the solution can quickly produce LOD models at a lower cost and realize the city Level of high-precision map coverage, while the accuracy of the model meets the application requirements, as shown in FIG. 23, the application effect of the high-precision map provided in this embodiment of the application in the AR map.

(2) Unmanned obstacle avoidance

The high-precision map obtained based on the LOD model constructed in the embodiments of this application can also be applied to the field of unmanned driving, such as unmanned intelligent flight equipment (such as unmanned aerial vehicles, aerial fire extinguishing equipment, etc.). The smart devices of can be equipped with high-precision maps provided by this application, so that these smart flying devices can accurately avoid various buildings during low-altitude flights.

It should be understood that the AR map and the obstacle avoidance for unmanned driving introduced above are only two specific scenarios for the application of the high-precision map constructed based on the method of the embodiment of the present application, and the application of the embodiment of the present application is not limited to the foregoing scenarios. , It can be applied to any scene where a map needs to be used.

On the basis of the embodiments corresponding to FIG. 12, FIG. 15 and FIG. 20, in order to better implement the above solutions of the embodiments of the present application, a computer device for implementing the above solutions is also provided below. For details, refer to FIG. 24. FIG. 24 is a schematic diagram of a computer device provided by an embodiment of the application. The computer device may specifically include: an acquisition module 2401, a nesting module 2402, a sampling module 2403, and a construction module 2404, wherein the acquisition module 2401 , Used to obtain the DSM corresponding to the satellite image and the contour vector of each building in the satellite image; the matching module 2402, used to merge the DSM with the contour vector of each building in the satellite image, Obtain the nested result; the sampling module 2403 is used to sample each building in the satellite image according to the nested result to obtain the 3D point cloud of each building in the satellite image; the building module 2404 is used to The 3D point cloud constructs each building in the satellite image to obtain a LOD model of each building in the satellite image.

In the above-mentioned embodiment of the present application, first, the acquisition module 2401 will acquire the DSM corresponding to the satellite image and the contour vector of each building in the satellite image, and use the matching module 2402 to convert the DSM and satellite image corresponding to the satellite image. The contour vectors of each building are combined, and then based on the combined DSM and building contour vectors, the 3D point cloud of the discrete buildings is sampled by the sampling module 2403, and finally the 3D point cloud is used for each building through the building module 2404. Since the 3D point cloud is sampled based on the outline of the building, it can restore the top surface information of the building with high precision, and thus a high-precision LOD model can be obtained. In addition, the foregoing embodiments of the present application do not need to obtain the height information of each building individually, which saves costs and reduces the amount of data at the same time.

In a possible design, the construction module 2404 is specifically configured to: reconstruct a grid model of each building in the satellite image from the 3D point cloud according to a first preset algorithm, and then, according to the first preset algorithm The second preset algorithm simplifies the grid model to obtain the LOD model of each building in the satellite image.

In a possible design, the construction module 2404 is specifically used to: iteratively simplify the grid model according to the edge folding algorithm until the simplified LOD model of each building in the satellite image satisfies With the preset conditions, the simplified LOD model obtained at this time is the final output LOD model.

In a possible design, the sampling module 2403 is specifically configured to: based on the outline of each building in the satellite image, perform sampling on the DSM at a preset step to obtain each 3D point cloud of buildings. In the above-mentioned embodiments of this application, it is specifically explained how to obtain discrete 3D point clouds according to the contours of buildings. The density of the 3D point clouds can be adjusted according to the step length, which can meet the different needs of users and is flexible.

In a possible design, the construction module 2404 is also used to: merge the LOD models of various buildings in multiple satellite images to obtain a target map.

In a possible design, the outline vector of each building in the satellite image is saved as a shp file. Specifically, it includes but is not limited to the following methods: 1) The outline of each building obtained based on the satellite image can be stored as a shp file, and a satellite image corresponds to a shp file. For example, there are 4 buildings in a satellite image, then The obtained contour vectors of the 4 buildings are all stored in the same shp file; 2) Each building corresponds to a shp file. For example, if there are 4 buildings in a certain satellite image, then these 4 buildings are obtained The contour vectors of are saved as 4 different shp files. The specific implementation of storing the outline vector of each building in each satellite image as a shp file is not limited here.

In the above-mentioned embodiments of this application, several ways of saving the building as a shp file are described, which are optional. In a possible design, the building includes at least one of the following objects: houses, bridges, electrical towers, tunnels, iron towers, water towers, iconic sculptures, dams, communication base stations, etc., further, as long as they are People determine the target in the satellite image and extract the contour vector of the target, then the target can be called the building described in the embodiment of the present application.

In a possible design, the first preset algorithm includes: a Poisson surface reconstruction algorithm.

It should be noted that FIG. 24 corresponds to the information interaction and execution process between the various modules/units in the computer equipment described in the embodiment, and is based on the same method embodiment as that of FIG. 12, FIG. 15 and FIG. 20 in this application. For the concept, the specific content can be referred to the description in the method embodiment shown in the foregoing application, which will not be repeated here.

Next, another computer device provided by an embodiment of this application will be introduced. Please refer to FIG. 25. FIG. 25 is a schematic structural diagram of a computer device provided by an embodiment of this application. The described modules are used to implement the functions of the computer device in the embodiment corresponding to FIG. 24. Specifically, the computer device 2500 is implemented by one or more servers. The computer device 2500 may have relatively large differences due to different configurations or performances. Including one or more central processing units (CPU) 2522 (e.g., one or more central processing units) and memory 2532, and one or more storage media 2530 (e.g., one or more central processing units) storing application programs 2542 or data 2544 A storage device in Shanghai). Among them, the memory 2532 and the storage medium 2530 may be short-term storage or permanent storage. The program stored in the storage medium 2530 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations on the computer device 2500. Furthermore, the central processing unit 2522 may be configured to communicate with the storage medium 2530, and execute a series of instruction operations in the storage medium 2530 on the computer device 2500.

The computer device 2500 may also include one or more power supplies 2526, one or more wired or wireless network interfaces 2550, one or more input and output interfaces 2558, and/or one or more operating systems 2541, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.

In this embodiment of the application, the central processing unit 2522 is configured to execute the method in the embodiment corresponding to FIG. 12, FIG. 15 or FIG. 20. For example, the central processing unit 2522 can be used to: obtain a DSM corresponding to a certain target satellite image (for example, a satellite image with a resolution of 0.5 meters) and the contour vector of each building in the satellite image, and combine the DSM with The contour vectors of the various buildings in the satellite image are combined to obtain the combined result. After that, the combined result (ie, the overlapped DSM and the contour vector of each building) can be used for each building in the current satellite image. Sampling is performed to obtain the 3D point cloud of each building in the satellite image. Finally, the computer equipment can construct each building in the current satellite image according to the obtained 3D point cloud of each building, so as to obtain each of the satellite images. The LOD model of the building. Similarly, the computer device can obtain the LOD model of each building in each satellite image in the satellite image collection (including the collection of multiple satellite images) in the above-mentioned manner.

It should be noted that the central processing unit 2522 may also be used to execute any step in the method embodiment corresponding to FIG. 12, FIG. 15 or FIG. 20 in this application. For details, please refer to the method embodiment shown in the foregoing application. The narrative, I won’t repeat it here.

The embodiment of the present application also provides a computer-readable storage medium. The computer-readable storage medium stores a program for signal processing. When it runs on a computer, the computer executes the above-mentioned embodiment description. The steps performed by the computer equipment.

In addition, it should be noted that the device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physically separate. The physical unit can be located in one place or distributed across multiple network units. Some or all of the modules can be selected according to actual needs to achieve the objectives of the solutions of the embodiments. In addition, in the drawings of the device embodiments provided in the present application, the connection relationship between the modules indicates that they have a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines.

Through the description of the above embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general hardware. Of course, it can also be implemented by dedicated hardware including dedicated integrated circuits, dedicated CPUs, dedicated memory, Dedicated components and so on to achieve. Under normal circumstances, all functions completed by computer programs can be easily implemented with corresponding hardware, and the specific hardware structure used to achieve the same function can also be diverse, such as analog circuits, digital circuits or special purpose circuits. Circuit etc. However, for this application, software program implementation is a better implementation in more cases. Based on this understanding, the technical solution of this application essentially or the part that contributes to the prior art can be embodied in the form of a software product. The computer software product is stored in a readable storage medium, such as a computer floppy disk. , U disk, mobile hard disk, read only memory (read only memory, ROM), random access memory (random access memory, RAM), magnetic disk or optical disk, etc., including several instructions to make a computer device (which can be a personal Computers, training devices, or network devices, etc.) execute the methods described in the various embodiments of the present application.

In the above-mentioned embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented by software, it can be implemented in the form of a computer program product in whole or in part.

The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website, computer, training device, or data. The center transmits to another website site, computer, training equipment, or data center through wired (such as coaxial cable, optical fiber, digital subscriber line) or wireless (such as infrared, wireless, microwave, etc.). The computer-readable storage medium may be any available medium that can be stored by a computer or a data storage device such as a training device or a data center integrated with one or more available media. The usable medium may be a magnetic medium (for example, a floppy disk, a hard disk, and a magnetic tape), an optical medium (for example, a high-density digital video disc (digital video disc, DVD)), or a semiconductor medium (for example, a solid state disk (solid state disk)). , SSD)) etc.

Claims

A method for constructing a multi-level-of-detail model of an object, which is characterized in that it includes:

Acquiring a digital surface model DSM corresponding to the satellite image and the contour vector of each building in the satellite image;

Merging the DSM with the contour vectors of each building in the satellite image to obtain a merging result;

Sampling each building in the satellite image according to the combination result to obtain a 3D point cloud of each building in the satellite image;

Constructing a multi-level-of-detail LOD model of each building in the satellite image according to the 3D point cloud.
The method according to claim 1, wherein the constructing the LOD model of each building in the satellite image according to the 3D point cloud comprises:

To obtain a grid model of each building in the satellite image by reconstructing from the 3D point cloud according to a first preset algorithm;

The grid model is simplified according to the second preset algorithm to obtain the LOD model of each building in the satellite image.
The method according to claim 2, wherein the simplifying the mesh model according to a second preset algorithm comprises:

The grid model is iteratively simplified according to the edge folding algorithm until the simplified LOD model of each building in the satellite image meets a preset condition.
The method according to any one of claims 1 to 3, wherein the sampling of each building in the satellite image is performed according to the result of the integration to obtain a 3D image of each building in the satellite image The point cloud includes:

Based on the outline of each building in the satellite image, sampling is performed on the DSM at a preset step length to obtain a 3D point cloud of each building in the satellite image.
The method according to any one of claims 1-4, wherein the method further comprises:

The LOD model of each building in the multiple satellite images is partitioned and merged to obtain the target map.
The method according to any one of claims 1 to 5, wherein the outline vector of each building in the satellite image is saved as a shp file.
The method according to any one of claims 1-6, wherein the building at least includes at least one of the following objects:

Houses, bridges, electrical towers, tunnels, iron towers, water towers, iconic sculptures, dams, communication base stations.
The method according to any one of claims 2-7, wherein the first preset algorithm comprises:

Poisson surface reconstruction algorithm.
A computer device, characterized in that it comprises:

An acquisition module, which is used to acquire the DSM corresponding to the satellite image and the contour vector of each building in the satellite image;

The nesting module is used to nest the DSM with the contour vectors of each building in the satellite image to obtain a nesting result;

A sampling module, configured to sample each building in the satellite image according to the set result to obtain a 3D point cloud of each building in the satellite image;

The construction module is used to construct the LOD model of each building in the satellite image according to the 3D point cloud.
The device according to claim 9, wherein the building module is specifically used for:

To obtain a grid model of each building in the satellite image by reconstructing from the 3D point cloud according to a first preset algorithm;

The grid model is simplified according to the second preset algorithm to obtain the LOD model of each building in the satellite image.
The device according to claim 10, wherein the building module is specifically further used for:

The grid model is iteratively simplified according to the edge folding algorithm until the simplified LOD model of each building in the satellite image meets a preset condition.
The device according to any one of claims 9-11, wherein the sampling module is specifically configured to:

Based on the outline of each building in the satellite image, sampling is performed on the DSM at a preset step length to obtain a 3D point cloud of each building in the satellite image.
The device according to any one of claims 9-12, wherein the building module is further used for:

The LOD model of each building in the multiple satellite images is partitioned and merged to obtain the target map.
The device according to any one of claims 9-13, wherein the outline vector of each building in the satellite image is saved as a shp file.
The device according to any one of claims 9-14, wherein the building includes at least one of the following objects:

Houses, bridges, electrical towers, tunnels, iron towers, water towers, iconic sculptures, dams, communication base stations.
The device according to any one of claims 10-15, wherein the first preset algorithm comprises:

Poisson surface reconstruction algorithm.
A computer device includes a memory and one or more processors, the one or more processors are coupled with the memory, and is characterized in that:

The memory is used to store programs;

The one or more processors are configured to execute the program in the memory, so that the computer device executes the method according to any one of claims 1-8.
A computer-readable storage medium, comprising computer-readable instructions, characterized in that, when the computer-readable instructions are run on a computer, the computer is caused to execute the method according to any one of claims 1-8.
A computer program product comprising computer-readable instructions, wherein when the computer-readable instructions are run on a computer, the computer is caused to execute the method according to any one of claims 1-8.
A chip, characterized in that, the chip includes a memory and one or more processors, and the chip is used to read a computer program stored in the memory, so that the one or more processors execute as claimed in claim 1- 8. The method of any one.