A 3D model splicing method based on volume-volume spatial relationship

Abstract

The present invention discloses a three-dimensional model splicing method based on body-body spatial relationship, by selecting a plurality of geometric body models in a geometric body model set for splicing, calculating the geometric distance between each geometric body model, judging whether the geometric distance is less than a first threshold value, if the geometric distance is less than the first threshold value and greater than 0, then splicing each geometric body model according to the area of ​​two faces to be spliced ​​between each geometric body model to obtain a three-dimensional model, if the geometric distance is less than the first threshold value and less than 0, then splicing each geometric body model according to the volume of each geometric body model to obtain a three-dimensional model. The above method improves the splicing accuracy of the three-dimensional model by splicing the three-dimensional model according to the spatial relationship between the three-dimensional models.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional model processing, and in particular to a three-dimensional model splicing method and device based on body-body spatial relationship. Background Art

[0002] At present, in order to build power system equipment models (such as substation models) as realistically as possible to reflect the actual scene, the image set generated by the drone driving the camera to shoot the real scene measurement area from multiple perspectives is usually processed through the motion recovery structure (SFM) algorithm to generate a sparse 3D point cloud. After reconstructing the sparse 3D point cloud through the multi-view stereo vision (MVS) algorithm, the dense 3D point cloud is used to build a Mesh model, and the image set is used to reconstruct the texture of the Mesh model, finally forming a real scene 3D model with geometric and texture details. However, with the increasing business needs such as smart operation and maintenance and smart inspection, it is necessary to segment the overall 3D model into independent entities with business and attribute information, and it is indispensable to provide a monomer 3D model construction technology in the substation scenario.

[0003] Existing 3D model stitching algorithms usually only stitch at least two 3D models with a specific spatial relationship. The aforementioned specific spatial relationship refers to the spatial relationship between two 3D models as body-to-body, and any two separated 3D models cannot be stitched together. For some scenes with many facilities and complex deployment (such as substations or production plants), due to the limitations of data collection and 3D reconstruction algorithms, it is difficult to balance modeling computing power and modeling accuracy, and it is easy to cause geometric errors in the 3D scene model. Even if a 3D model stitching algorithm is used, some errors and omissions will occur. For example, different parts that should belong to the same device need to be spliced ​​together, but they cannot be spliced ​​together because they are separated in a 3D spatial relationship. This separation relationship may be caused by various factors such as modeling errors. Summary of the invention

[0004] In order to solve the above technical problems, an embodiment of the present invention provides a three-dimensional model stitching method based on body-body spatial relationship, which improves the three-dimensional model stitching accuracy by stitching the three-dimensional models according to the spatial relationship between the three-dimensional models.

[0005] A first aspect of an embodiment of the present invention provides a three-dimensional model splicing method based on a volume-volume spatial relationship, the method comprising:

[0006] Select several geometric models from the geometric model set for splicing;

[0007] Calculate the geometric distance between each geometric model, and determine whether the geometric distance is less than a first threshold value. If the geometric distance is less than the first threshold value and greater than 0, splice each geometric model according to the area of ​​two faces to be spliced ​​between each geometric model to obtain a three-dimensional model.

[0008] If the geometric distance is less than the first threshold and less than 0, the geometric models are spliced ​​according to the volumes of the geometric models to obtain a three-dimensional model.

[0009] In a possible implementation of the first aspect, the geometric models are spliced ​​according to the areas of two surfaces that need to be spliced ​​between the geometric models to obtain a three-dimensional model, including:

[0010] Calculate the areas of two faces that need to be spliced ​​between several geometric models respectively to obtain the area ratio;

[0011] If the area ratio is greater than the second threshold, the first splicing mode is used to splice the geometric models;

[0012] If the area ratio is less than or equal to the second threshold, the second splicing mode is used to splice the geometric models.

[0013] In a possible implementation manner of the first aspect, if the area ratio is greater than the second threshold, the first splicing mode is used to splice the geometric models, including:

[0014] If the area ratio is greater than the second threshold, the boundaries of two faces that need to be spliced ​​between the plurality of geometric body models are connected to obtain a truncated cone;

[0015] Several geometric models are spliced ​​together according to the frustum.

[0016] In a possible implementation manner of the first aspect, if the area ratio is less than or equal to the second threshold, the second splicing mode is used to splice the geometric models, including:

[0017] If the area ratio is less than or equal to the second threshold, the geometric model corresponding to the surface with a smaller area is extended toward the geometric model corresponding to the surface with a larger area to obtain an intersection area;

[0018] Delete the intersecting area to complete the splicing.

[0019] In a possible implementation of the first aspect, the geometric models are spliced ​​according to the volumes of the geometric models to obtain a three-dimensional model, including:

[0020] Calculate the volume of each geometric model respectively to obtain the volume ratio between the geometric models;

[0021] Determine whether the volume ratio between the geometric models is greater than a third threshold value, and if it is less than or equal to the third threshold value, splice the geometric models using the third splicing mode or the fourth splicing mode according to the number of non-overlapping regions;

[0022] If it is greater than the third threshold, the geometric model with a smaller volume is extended, and the overlapping area is deleted to complete the splicing to obtain a three-dimensional model.

[0023] In a possible implementation manner of the first aspect, the non-overlapping area is obtained according to the extended overlapping area, including:

[0024] Determine the overlapping area between the geometric models, and expand the overlapping area along the axis direction of each geometric model to obtain an expanded overlapping area;

[0025] According to the extended overlapping area, the corresponding non-overlapping area is obtained.

[0026] In a possible implementation manner of the first aspect, the geometric models are spliced ​​using a third splicing mode according to the number of non-overlapping areas, including:

[0027] If the number of non-overlapping regions is greater than or equal to two, the volume of each non-overlapping region is calculated to obtain a volume ratio between the non-overlapping regions; if the volume ratio between the non-overlapping regions is less than a fourth threshold, the non-overlapping region with a small volume is deleted to complete the stitching;

[0028] If the volume ratio between the non-overlapping regions is greater than a fourth threshold, the overlapping region is deleted.

[0029] In a possible implementation manner of the first aspect, the geometric models are spliced ​​using a fourth splicing mode according to the number of non-overlapping regions, including:

[0030] If the number of non-overlapping regions is less than two, the overlapping regions are deleted.

[0031] A second aspect of an embodiment of the present invention provides a three-dimensional model splicing method for a device, the method comprising:

[0032] Collect point cloud data of the equipment scene to be modeled;

[0033] Perform scene instance segmentation on the point cloud data of the scene to be modeled to obtain multiple scene instances, and analyze the geometric semantic information of each scene instance to obtain the shape information of each scene instance;

[0034] Based on the shape information, local fitting is performed using the point cloud data to obtain a set of geometric models of each scene instance, wherein the shape information includes multiple geometric bodies, and the set of geometric models includes multiple geometric body models;

[0035] The geometric body models in the geometric body model set are spliced ​​to obtain a three-dimensional model of the device scene to be modeled, wherein the three-dimensional model is obtained by the three-dimensional model splicing method based on the body-body spatial relationship.

[0036] In a possible implementation manner of the first aspect, after obtaining the three-dimensional model of the device scene to be modeled, the method further includes:

[0037] A horizontal plane of a preset height is used as the ground to determine whether the scene instances corresponding to each three-dimensional model in the three-dimensional model of the device scene to be modeled are of the grounding category. If so, the three-dimensional model is extended or cut; if not, it remains unchanged.

[0038] The three-dimensional model splicing method based on the body-body spatial relationship provided by the embodiment of the present invention selects several geometric models from the geometric model set for splicing, calculates the geometric distance between each geometric model, and determines whether the geometric distance is less than a first threshold value. If the geometric distance is less than the first threshold value and greater than 0, the geometric models are spliced ​​according to the areas of the two faces to be spliced ​​between the geometric models to obtain a three-dimensional model. If the geometric distance is less than the first threshold value and less than 0, the geometric models are spliced ​​according to the volumes of the geometric models to obtain a three-dimensional model. The above method improves the splicing accuracy of the three-dimensional model by splicing the three-dimensional model according to the spatial relationship between the three-dimensional models. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 : A flow chart of an embodiment of a three-dimensional model splicing method based on body-body spatial relationship provided by the present invention;

[0040] Figure 2 : A schematic diagram of shape information and geometric semantics of an embodiment of a three-dimensional model splicing method based on body-body spatial relationship provided by the present invention;

[0041] Figure 3 : A schematic diagram of geometric body penetration in a three-dimensional model splicing method based on body-body spatial relationship according to an embodiment of the present invention;

[0042] Figure 4 : A schematic diagram of a flow chart of a method for assembling a monomerized three-dimensional model according to an embodiment of a method for assembling a three-dimensional model based on a volume-volume spatial relationship provided by the present invention;

[0043] Figure 5 : A schematic diagram of the separation of monomer A and monomer B according to an embodiment of the three-dimensional model splicing method based on body-body spatial relationship provided by the present invention;

[0044] Figure 6: A schematic diagram of a cone according to an embodiment of a three-dimensional model splicing method based on body-body spatial relationship provided by the present invention;

[0045] Figure 7 : A schematic diagram of the extension of monomer A and monomer B in one embodiment of the three-dimensional model splicing method based on body-body spatial relationship provided by the present invention;

[0046] Figure 8 : A schematic diagram of an overlapping area expansion of an embodiment of a three-dimensional model splicing method based on body-body spatial relationship provided by the present invention;

[0047] Fig. 9 : A schematic diagram of geometric segmentation of an embodiment of a three-dimensional model splicing method based on body-body spatial relationship provided by the present invention;

[0048] Fig.10 : A schematic diagram of a splicing method flow chart of an embodiment of a three-dimensional model splicing method based on body-body spatial relationship provided by the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] Please refer to Figure 1 , which is a flow chart of an embodiment of a three-dimensional model splicing method based on body-body spatial relationship provided in an embodiment of the present invention, including steps S101 to S103.

[0051] S101, selecting a plurality of geometric models from a geometric model set for splicing.

[0052] In this embodiment, the shape information of the power grid equipment is obtained, such as Figure 2 As shown, the shape information includes multiple different geometric bodies. After analyzing the shape information in the scene instance, the RANSAC algorithm is used to locally fit the device point cloud with simple basic geometric bodies such as a cuboid and a cylinder to obtain a multi-geometric body model belonging to at least one device.

[0053] It should be noted that the geometric model is a geometric model obtained by fitting the point cloud data of the power grid equipment.

[0054] S102, calculating the geometric distance between each geometric model, and determining whether the geometric distance is less than a first threshold value; if the geometric distance is less than the first threshold value and greater than 0, splicing each geometric model according to the areas of two faces to be spliced ​​between each geometric model to obtain a three-dimensional model.

[0055] In this embodiment, due to modeling errors, there are abnormal conditions such as separation and penetration between the geometric models generated by the power grid equipment, such as Figure 3 Therefore, assuming that the two geometric models are monomer A and monomer B, the three-dimensional spatial relationship between A and B is used to achieve refined splicing of monomer three-dimensional models to prevent missing splicing. The specific flow chart of the monomer three-dimensional model splicing method is as follows: Figure 4 shown.

[0056] If monomer A and monomer B are far apart in three-dimensional space, if the distance between the two exceeds a preset threshold M, it means that monomer A is unrelated to monomer B, and in this case there is no need to splice the two. For example, there are input terminals and output terminals on the transformer, and any two terminals should not be directly connected, or the knife switch has two columns, and the two columns should not be directly connected. Therefore, the corresponding monomer A and monomer B are also unrelated, and there is no need to splice monomer A and monomer B.

[0057] If the distance between cell A and cell B does not exceed M and is greater than 0, it means that the two are separated but need to be connected, for example. In the equipment in the substation scenario, this situation occurs at the bent wire (hereinafter referred to as: bent separation situation) or the inclined component connection (hereinafter referred to as: inclined separation situation). The threshold M can be set according to actual needs.

[0058] like Figure 5 As shown, in the case of bending separation, monomer A and monomer B are usually of the same shape and similar size, generally in the form of a cylinder or a cuboid; in the case of tilting separation, monomer A and monomer B are usually different in shape and have a large size difference, and there is a large size difference between the two adjacent faces. For simplicity, the characteristics of the two adjacent faces are used as indicators to distinguish by comparing the closest face Sa of monomer A to monomer B and the closest face Sb of monomer B to monomer A.

[0059] It should be noted that the first threshold, ie, the threshold M, is the threshold of the distance between each geometric model in space, and can be set according to actual conditions, such as 20 cm or 30 cm.

[0060] In some embodiments, in step S102, "joining the geometric models according to the areas of the two faces to be joined between the geometric models to obtain a three-dimensional model" includes but is not limited to the following steps:

[0061] Calculate the areas of two faces that need to be spliced ​​between several geometric models respectively to obtain the area ratio;

[0062] If the area ratio is less than or equal to the second threshold, the first splicing mode is used to splice the geometric models;

[0063] If the area ratio is less than the second threshold, the second splicing mode is used to splice the geometric models.

[0064] In this embodiment, the area of ​​the surface Sa of monomer A closest to monomer B and the area of ​​the surface Sb of monomer B closest to monomer A are calculated, and the sizes and area ratios of the comparison areas are calculated. If Sa is a smaller surface and Sb is a larger surface, and the area ratio of the smaller surface Sa to the larger surface Sb is less than or equal to a preset threshold value N, which belongs to a tilted separation situation, the second stitching mode is used to stitch the various geometric models.

[0065] If the area ratio of the smaller surface Sa to the larger surface Sb is equal to or higher than the preset threshold value N, it belongs to the curved separation situation. In this case, both monomers are cylindrical, and the surface Sa and the surface Sb are connected at the boundary to form a truncated cone that allows monomer A and monomer B to be seamlessly connected, that is, the first splicing mode is used to splice the various geometric models.

[0066] It should be noted that the second threshold also refers to a preset threshold N, which is a threshold of the ratio of the area of ​​the first geometric model to the area of ​​the second geometric model. The preset threshold N should be closer to 1, such as 0.9 or 0.8.

[0067] In some embodiments, in step S102, "if the area ratio is less than or equal to the second threshold, then using the first stitching mode to stitch the geometric models together" includes but is not limited to the following steps:

[0068] If the area ratio is less than or equal to the second threshold, the boundaries of two faces that need to be spliced ​​between the plurality of geometric body models are connected to obtain a truncated cone;

[0069] Several geometric models are spliced ​​together according to the frustum.

[0070] In this embodiment, if the area ratio of the smaller surface Sa to the larger surface Sb is equal to or greater than the preset threshold value N, it belongs to the bending separation situation. In this case, both monomers are cylindrical, and the surface Sa and the surface Sb are connected at the boundary to form a truncated cone that allows the monomer A and the monomer B to be seamlessly connected, such as Figure 6 shown.

[0071] In some embodiments, in step S102, "if the area ratio is less than or equal to the second threshold, then using the second stitching mode to stitch the geometric models together" includes but is not limited to the following steps:

[0072] If the area ratio is less than the second threshold, the geometric model corresponding to the surface with a smaller area is extended toward the geometric model corresponding to the surface with a larger area to obtain an intersection area;

[0073] Delete the intersecting area to complete the splicing.

[0074] In this embodiment, if the area ratio of the smaller surface Sa to the larger surface Sb is less than or equal to the preset threshold value N, it is a case of tilted separation. In this case, the monomer A with a smaller adjacent surface Sa is attached to the monomer B with a larger adjacent surface Sb, and is extended from the end of the monomer A close to the monomer B along the length direction of the monomer A to the inside of the monomer B, forming A', and C=A'∩B is calculated, and C is subtracted from the monomer A, so that the monomer A and the monomer B can be seamlessly connected. Extending the monomer A means extending the monomer along the length direction, such as Figure 7 As shown, as monomer A expands, the overlapping volume between monomer A and monomer B will also increase. When the volume corresponding to the extended part of monomer A overlapping with monomer B reaches the maximum, that is, the junction of segment a and segment b, the monomer extension is completed to obtain the intersection area, and then the intersection area is deleted to complete the splicing.

[0075] S103: If the geometric distance is less than the first threshold and less than 0, then the geometric models are spliced ​​according to the volumes of the geometric models to obtain a three-dimensional model.

[0076] In this embodiment, if the geometric distance between monomer A and monomer B does not exceed 0, it means that there is an overlapping part between the two (non-separable relationship). Furthermore, by comparing the volume Va of monomer A with the volume Vb of monomer B, the size relationship between the two can be identified, and then the splicing can be judged based on the volume ratio of the two to obtain a three-dimensional model.

[0077] In some embodiments, in step S103, "joining the geometric models according to the volumes of the geometric models to obtain a three-dimensional model" includes but is not limited to the following steps:

[0078] Calculate the volume of each geometric model respectively to obtain the volume ratio between the geometric models;

[0079] Determine whether the volume ratio between the geometric models is greater than a third threshold value, and if it is less than or equal to the third threshold value, splice the geometric models using the third splicing mode or the fourth splicing mode according to the number of non-overlapping regions;

[0080] If it is greater than the third threshold, the geometric model with a smaller volume is extended, and the overlapping area is deleted to complete the splicing to obtain a three-dimensional model.

[0081] In this embodiment, the volume Va of monomer A and the volume Vb of monomer B are calculated and compared to determine the size relationship between the two, and the volume ratio is calculated to determine whether the volume ratio of Va to Vb is close to 1. If the ratio of Va to Vb is close to 1, it means that monomer A and monomer B are two parts of similar size and molded together. The overlapping parts are expanded along the axial direction of each monomer and then spliced ​​after processing.

[0082] It should be noted that the third threshold is a value close to 1.

[0083] In some embodiments, in step S103, “the non-overlapping area is obtained by expanding the overlapping area” includes but is not limited to the following steps:

[0084] Determine the overlapping area between the geometric models, and expand the overlapping area along the axis direction of each geometric model to obtain an expanded overlapping area;

[0085] According to the extended overlapping area, the corresponding non-overlapping area is obtained.

[0086] In this embodiment, if the ratio of Va to Vb is close to 1, it means that monomer A and monomer B are two parts of similar size and through-molded, and the overlapping part is expanded along the axis direction of each monomer. Thereafter, when comparing monomer A or monomer B, in addition to the corresponding expanded overlapping area, there are one or two non-overlapping areas complementary to the expanded overlapping area, such as Figure 8 As shown, the dotted line represents the overlapping part before expansion, and the red line represents the expanded overlapping area.

[0087] In some embodiments, step S103 of "joining the geometric models using a third joining mode according to the number of non-overlapping regions" includes but is not limited to the following steps:

[0088] If the number of non-overlapping regions is greater than or equal to two, the volume of each non-overlapping region is calculated to obtain a volume ratio between the non-overlapping regions; if the volume ratio between the non-overlapping regions is less than a fourth threshold, the non-overlapping region with a small volume is deleted to complete the stitching;

[0089] If the volume ratio between the non-overlapping regions is greater than a fourth threshold, the overlapping region is deleted.

[0090] In this embodiment, if the number of non-overlapping areas is greater than or equal to two, for monomer A, the smaller volume of the two non-overlapping areas is recorded as Vc and the larger volume is recorded as Vd. If Vc / Vd is less than or equal to the threshold value P2, the smaller non-overlapping area and the overlapping part of any monomer component are deleted respectively, so that monomer A and monomer B can be seamlessly connected.

[0091] Since a non-overlapping region with a smaller volume is usually caused by an error and its volume is much smaller than another non-overlapping region, the preset threshold P2 should be smaller than 0.5, such as 0.1 or 0.2.

[0092] In some embodiments, step S103 of "joining the geometric models in a fourth joining mode according to the number of non-overlapping regions" includes but is not limited to the following steps:

[0093] If the number of non-overlapping regions is less than two, the overlapping regions are deleted.

[0094] In this embodiment, if the ratio of Va to Vb is not close to 1, that is, less than or equal to the threshold value P1, it means that monomer A and monomer B are larger and smaller components of different sizes and penetrate the mold. The smaller component is extended and the area of ​​the smaller component that overlaps with the larger component is deleted so that monomer A and monomer B can be seamlessly spliced.

[0095] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple steps or multiple stages. These steps or stages are not necessarily performed at the same time, but can be performed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed in turn or alternately with other steps or at least part of the steps or stages in other steps.

[0096] In one embodiment, a flowchart of a three-dimensional model splicing method for a device provided in an embodiment of the present application is shown, including:

[0097] Collect point cloud data of the equipment scene to be modeled;

[0098] Perform scene instance segmentation on the point cloud data of the scene to be modeled to obtain multiple scene instances, and analyze the geometric semantic information of each scene instance to obtain the shape information of each scene instance;

[0099] Based on the shape information, local fitting is performed using the point cloud data to obtain a set of geometric models of each scene instance, wherein the shape information includes multiple geometric bodies, and the set of geometric models includes multiple geometric body models;

[0100] The geometric body models in the geometric body model set are spliced ​​to obtain a three-dimensional model of the device scene to be modeled, wherein the three-dimensional model is obtained by the three-dimensional model splicing method based on the body-body spatial relationship.

[0101] In this embodiment, in order to obtain point cloud data for power grid equipment, such as substation equipment, a dense three-dimensional point cloud is reconstructed from multi-view images through the motion recovery structure (SFM) algorithm and the multi-view stereo vision (MVS) algorithm, and then a laser three-dimensional point cloud is obtained by a lidar scanning instrument. The dense three-dimensional point cloud and the laser three-dimensional point cloud are fused using the existing algorithm to obtain the substation scene point cloud data.

[0102] The pre-trained deep learning algorithm is used to perform instance segmentation on the substation scene point cloud data to obtain multiple scene instances, such as Fig. 9 As shown, in Fig. 9 In Figure b, the algorithm divides the device into multiple different instance objects.

[0103] Then the geometric semantic information of each scene instance is analyzed to obtain the shape information of each scene instance, such as Figure 2 As shown, the shape information includes a plurality of different geometries.

[0104] It should be noted that the deep learning algorithm may be PointNet++, RandLA-Net, 3D-BoNet and other algorithms used in the field, and the algorithm for instance segmentation is a commonly used technology in the field, so it will not be repeated here.

[0105] After analyzing the shape information in the scene instance, the RANSAC algorithm is used to perform local fitting on the device point cloud with simple basic geometric bodies such as cuboids and cylinders to obtain a multi-geometric model belonging to at least one device. The specific steps for performing local fitting using point cloud data to obtain a set of geometric models for each scene instance are as follows:

[0106] For a point cloud With a set of simple geometric shapes . Find a point cloud with s points , and a certain geometric body in Ψ (denoted as ) has a score of c, such that for any point set with more than s points ∈P, The fit score with any basic shape in Ψ is greater than c. , Geometry , and the remaining point cloud (P- ), and then continue to iterate and search according to the above method , … ,Finally the device point cloud is divided into multiple geometric body models.

[0107] like Fig.10 As shown, after obtaining the set of geometric models of each scene instance, the spatial relationship between each geometric model is used to splice each geometric model to obtain a monomer model. Then, according to the spatial relationship between each monomer model, each monomer model is connected to obtain a three-dimensional model of the device scene to be modeled.

[0108] In some embodiments, after obtaining the three-dimensional model of the device scene to be modeled, the method further includes:

[0109] A horizontal plane of a preset height is used as the ground to determine whether the scene instances corresponding to each three-dimensional model in the three-dimensional model of the device scene to be modeled are of the grounding category. If so, the three-dimensional model is extended or cut; if not, it remains unchanged.

[0110] In this embodiment, for the substation scenario, the device model splicing mainly involves connecting multiple components belonging to different device models. After the device models are spliced, a horizontal plane with a fixed height is used as the ground to determine whether the category of each device model corresponding to the scene point cloud instance segmentation result is a grounding category. If so, the bottom of the device model is extended or cut to make the device model fit the ground. Otherwise, the corresponding device model is maintained unchanged.

[0111] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0112] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A three-dimensional model splicing method based on body-body spatial relationship, characterized in that: include: Select several geometric models from the geometric model set for splicing; Calculating the geometric distance between each of the geometric models, determining whether the geometric distance is less than a first threshold, and if the geometric distance is less than the first threshold and the geometric models are separated, splicing each of the geometric models according to the areas of two faces to be spliced ​​between each of the geometric models to obtain a three-dimensional model; If the geometric distance is less than the first threshold and there is an overlapping part between the geometric models, which is a non-separable relationship, then according to the volume of each geometric model, each geometric model is spliced ​​to obtain a three-dimensional model; The method of splicing the geometric models according to the areas of the two surfaces to be spliced ​​between the geometric models to obtain a three-dimensional model includes: Calculating the areas of two faces that need to be spliced ​​between a plurality of the geometric body models respectively to obtain an area ratio; If the area ratio is greater than a second threshold, the first splicing mode is used to splice the geometric models; If the area ratio is less than or equal to the second threshold, the second splicing mode is used to splice the geometric models; The step of splicing the geometric models according to the volumes of the geometric models to obtain a three-dimensional model includes: Calculating the volume of each of the geometric models respectively to obtain a volume ratio between the geometric models; Determine whether the volume ratio between the geometric models is greater than a third threshold value, and if it is less than or equal to the third threshold value, splice the geometric models using the third splicing mode or the fourth splicing mode according to the number of non-overlapping areas; If it is greater than the third threshold, the geometric model with a smaller volume is extended, and the overlapping area is deleted to complete the splicing to obtain a three-dimensional model.

2. The three-dimensional model splicing method based on body-body spatial relationship according to claim 1, characterized in that: If the area ratio is greater than a second threshold, each of the geometric models is spliced ​​using a first splicing mode, including: If the area ratio is greater than a second threshold, connecting the boundaries of two faces that need to be spliced ​​between the plurality of geometric body models to obtain a truncated cone; Several geometric models are spliced ​​together according to the frustum.

3. The three-dimensional model splicing method based on body-body spatial relationship according to claim 1, characterized in that: If the area ratio is less than or equal to the second threshold, the second splicing mode is used to splice the geometric models, including: If the area ratio is less than or equal to the second threshold, the geometric model corresponding to the surface with a smaller area is extended toward the geometric model corresponding to the surface with a larger area to obtain an intersection area; The intersecting area is deleted to complete the splicing.

4. The three-dimensional model splicing method based on body-body spatial relationship according to claim 1, characterized in that: The non-overlapping area is obtained according to the extended overlapping area, including: Determine the overlapping area between each of the geometric body models, and expand the overlapping area along the axis direction of each of the geometric body models to obtain an expanded overlapping area; According to the extended overlapping area, a corresponding non-overlapping area is obtained.

5. The three-dimensional model splicing method based on body-body spatial relationship according to claim 1, characterized in that: The step of splicing the geometric models using a third splicing mode according to the number of non-overlapping areas includes: If the number of the non-overlapping regions is greater than or equal to two, the volumes of the non-overlapping regions are calculated to obtain a volume ratio between the non-overlapping regions; if the volume ratio between the non-overlapping regions is less than a fourth threshold, the non-overlapping regions with smaller volumes are deleted to complete the stitching; If the volume ratio between the non-overlapping areas is greater than a fourth threshold, the overlapping areas are deleted.

6. The three-dimensional model splicing method based on body-body spatial relationship according to claim 1, characterized in that: The step of splicing the geometric models using a fourth splicing mode according to the number of non-overlapping areas includes: If the number of the non-overlapping regions is less than two, the overlapping regions are deleted.

7. A three-dimensional model splicing method for equipment, characterized in that: include: Collect point cloud data of the equipment scene to be modeled; Performing scene instance segmentation on the point cloud data of the scene to be modeled to obtain a plurality of scene instances, and performing analysis based on the geometric semantic information of each of the scene instances to obtain shape information of each of the scene instances; Based on the shape information, the point cloud data is used for local fitting to obtain a set of geometric models of each of the scene instances, wherein the shape information includes a plurality of geometric bodies, and the set of geometric models includes a plurality of geometric models; The geometric body models in the geometric body model set are spliced ​​to obtain a three-dimensional model of the device scene to be modeled, wherein the three-dimensional model is obtained by the three-dimensional model splicing method based on body-body spatial relationship as described in any one of claims 1 to 6.

8. The three-dimensional model splicing method for equipment according to claim 7, characterized in that: After obtaining the three-dimensional model of the device scene to be modeled, the method further includes: A horizontal plane at a preset height is used as the ground, and it is determined whether the scene instance corresponding to each three-dimensional model in the three-dimensional model of the device scene to be modeled is a grounding category. If so, the three-dimensional model is extended or cut, and if not, it remains unchanged.

Images