3D model generation apparatus, 3D model generation method, and computer program

Abstract

This invention provides a 3D model generation device, a 3D model generation method, and a computer program that can improve the accuracy of information about an object. [Solution] The 3D model generation device 100 includes a data acquisition unit 11 that acquires 3D point cloud data representing a 3D point cloud including buildings and structures, a 3D data storage unit 3 that has a data file 21 containing multiple 3D CAD data of structures, and an information processing unit 12 that compares the 3D point cloud data of structures with the 3D CAD data contained in the data file using a learning model 22 and identifies the 3D CAD data that matches the 3D point cloud data from among the multiple 3D CAD data. The information processing unit 12 generates a 3D model of the target object by combining the identified 3D CAD data of structures with the 3D point cloud data of buildings.

Description

Technical Field

[0001] The present invention relates to a three-dimensional model generation device, a three-dimensional model generation method, and a computer program.

Background Art

[0002] In order to represent the shape of an object three-dimensionally, a three-dimensional model is used. For example, a three-dimensional model of a building is used in BIM (Building Information Modeling), CIM (Construction Information Modeling), etc. A three-dimensional model can be generated, for example, using three-dimensional point cloud data. The three-dimensional point cloud data can be obtained by analyzing images taken of an object from multiple locations (see, for example, Patent Document 1). Patent Document 1 discloses a device for generating a three-dimensional model of a building.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, the three-dimensional model obtained by the device described in Patent Document 1 had room for improvement in terms of the accuracy of information regarding the position, shape, etc. of the structures installed inside the building. Therefore, it has been demanded to improve the accuracy of information regarding the object.

[0005] One aspect of the present invention aims to provide a three-dimensional model generation device, a three-dimensional model generation method, and a computer program that can improve the accuracy of information regarding an object.

Means for Solving the Problems

[0006] A three-dimensional model generation apparatus according to embodiment 1 of the present invention is a three-dimensional model generation apparatus that generates a three-dimensional model of an object having a building and a structure installed inside the building, comprising: a data acquisition unit that acquires three-dimensional point cloud data representing a three-dimensional point cloud including the building and the structure; a three-dimensional data storage unit having a data file containing a plurality of three-dimensional CAD data of the structure; and an information processing unit that compares the three-dimensional point cloud data of the structure and the three-dimensional CAD data contained in the data file using a learning model and identifies the three-dimensional CAD data that matches the three-dimensional point cloud data from among the plurality of three-dimensional CAD data, wherein the information processing unit generates a three-dimensional model of the object by combining the identified three-dimensional CAD data of the structure and the three-dimensional point cloud data of the building.

[0007] In the 3D model generation apparatus according to embodiment 2 of the present invention, the information processing unit, when comparing the 3D point cloud data and the 3D CAD data, superimposes the 3D point cloud data and the 3D CAD data, and determines that the 3D point cloud data and the 3D CAD data match when the degree of agreement between the surface shape of the 3D point cloud data and the surface shape of the 3D CAD data is greater than or equal to a preset value.

[0008] A 3D model generation apparatus according to embodiment 3 of the present invention is a 3D model generation apparatus according to embodiment 1 or embodiment 2, wherein the information processing unit generates corrected 3D point cloud data by interpolating the gaps between the point clouds constituting the 3D point cloud data of the structure as surfaces, and compares the corrected 3D point cloud data with the 3D CAD data.

[0009] A three-dimensional model generation apparatus according to aspect 4 of the present invention is a three-dimensional model generation apparatus according to any one of aspects 1 to 3, wherein the information processing unit stores attribute information and position information of the structure in association.

[0010] A 3D model generation apparatus according to aspect 5 of the present invention is a 3D model generation apparatus according to any one of aspects 1 to 4, wherein the structure having the 3D CAD data is at least one of a pump, turbine, speed reducer, refrigerator, and control panel.

[0011] A 3D model generation method according to embodiment 6 of the present invention is a 3D model generation method for generating a 3D model of an object having a building and a structure installed inside the building, comprising the steps of: acquiring 3D point cloud data representing a 3D point cloud including the building and the structure; comparing the 3D point cloud data of the structure and the 3D CAD data included in a data file containing multiple 3D CAD data of the structure using a learning model, and identifying the 3D CAD data that matches the 3D point cloud data from among the multiple 3D CAD data; and generating a 3D model of the object by combining the identified 3D CAD data of the structure and the 3D point cloud data of the building.

[0012] A computer program according to aspect 7 of the present invention is a computer program for generating a three-dimensional model of an object having a building and a structure installed inside the building, and is a computer program for realizing by computer the following functions: acquiring three-dimensional point cloud data representing a three-dimensional point cloud including the building and the structure; comparing the three-dimensional point cloud data of the structure with the three-dimensional CAD data included in a data file containing multiple three-dimensional CAD data of the structure using a learning model, and identifying the three-dimensional CAD data that matches the three-dimensional point cloud data from among the multiple three-dimensional CAD data; and generating a three-dimensional model of the object by combining the identified three-dimensional CAD data of the structure and the three-dimensional point cloud data of the building. [Effects of the Invention]

[0013] One aspect of the present invention provides a 3D model generation apparatus, a 3D model generation method, and a computer program that can improve the accuracy of information about an object. [Brief explanation of the drawing]

[0014] [Figure 1] This figure shows the configuration of a 3D model generation apparatus according to the first embodiment. [Figure 2] This figure shows the computer configuration of the 3D model generation apparatus according to the first embodiment. [Figure 3] This is a flowchart of the 3D model generation method according to the first embodiment. [Figure 4] This is an explanatory diagram of the 3D model generation method according to the first embodiment. [Figure 5] This is a flowchart of the 3D model generation method according to the second embodiment. [Modes for carrying out the invention]

[0015] Hereinafter, a 3D model generation apparatus, a 3D model generation method, and a computer program according to embodiments of the present invention will be described with reference to the drawings.

[0016] [3D Model Generation Device] (First Embodiment) Figure 1 is a block diagram showing the configuration of a 3D model generation apparatus 100 according to the first embodiment. In this embodiment, the objects to be modeled include buildings and structures installed inside buildings. Examples of buildings include drainage pumping stations, waste incineration plants, factories, water treatment plants, port facilities, hospitals, schools, airports, power plants, stadiums, train stations, and the like.

[0017] As shown in Figure 1, the 3D model generation device 100 has an information processing device 101. The information processing device 101 comprises a control unit 1, a communication unit 2, a 3D data storage unit 3, and a trained model storage unit 4.

[0018] The control unit 1 includes a data acquisition unit 11 and an information processing unit 12. The data acquisition unit 11 acquires three-dimensional point cloud data representing the three-dimensional point cloud of the object. The data acquisition unit 11 may acquire the three-dimensional point cloud data from the imaging device 102. The data acquisition unit 11 may also acquire the three-dimensional point cloud data via the communication unit 2. The data acquisition unit 11 stores the three-dimensional point cloud data of the object.

[0019] The three-dimensional point cloud has a plurality of points to which three-dimensional coordinate values and normal vectors are assigned. The normal vector is the normal vector of the plane where the plurality of points are estimated to exist. The three-dimensional point cloud can be acquired by, for example, an imaging device using a laser-based technology. The three-dimensional point cloud can be generated based on, for example, a plurality of images captured by a plurality of imaging devices with different installation positions. The three-dimensional point cloud can be generated by, for example, using a known SfM process. The three-dimensional point cloud can be generated by, for example, extracting common features within the overlap region between a plurality of images and associating these features.

[0020] The three-dimensional point cloud includes points representing the structural surfaces (such as wall surfaces, floor surfaces, ceiling surfaces, etc.) of the building and points representing the structures installed inside the building. Examples of the structures include pumps, turbines, speed reducers, refrigerators, control panels, etc. At least one of these can be selected as the structure.

[0021] The communication unit 2 is connected to an external device via a network. The communication unit 2 functions as a communication interface for transmitting and receiving data.

[0022] The 3D data storage unit 3 has a data file 21 that stores 3D CAD data of structures (pumps, turbines, speed reducers, refrigerators, control panels, etc.). The data file 21 contains 3D CAD data of multiple structures. The 3D CAD data includes attribute information of the structure. For example, if the structure is a pump, the 3D CAD data includes information that it has the attribute of being a pump. Examples of attribute information included in the 3D CAD data include the specifications of the structure (e.g., shape, size, thickness, structure, material, function, etc.), manufacturer, model number, etc.

[0023] The trained model storage unit 4 contains the trained model 22. The trained model 22 is obtained by learning the correlation between 3D point cloud data and 3D CAD data of structures using machine learning. The trained model storage unit 4 stores the trained model 22, which has learned the correlation between 3D point cloud data and 3D CAD data.

[0024] Figure 2 is a configuration diagram showing a computer 200 that executes the 3D model generation method according to the embodiment. As shown in Figure 2, the computer 200 includes a processor 201, memory 202, storage 203, input device 204, output device 205, display device 206, and media input / output unit 207.

[0025] The processor 201 has one or more arithmetic processing units (such as a CPU (Central Processing Unit), MPU (Micro Processing Unit), and GPU (Graphics Processing Unit)). The processor 201 controls the overall operation of the computer 200. The memory 202 stores various data and programs. The memory 202 consists of volatile memory, non-volatile memory, flash memory, etc. The storage 203 consists of, for example, an HDD or SSD.

[0026] The input device 204 is, for example, a keyboard, mouse, numeric keypad, etc. The output device 205 is, for example, an audio output device, a vibration device, etc. The display device 206 is, for example, a liquid crystal display, an organic EL display, etc. The media input / output unit 207 is, for example, a drive device such as a DVD drive or a CD drive. The media input / output unit 207 can read and write data to media such as DVDs and CDs (non-temporary storage media).

[0027] The control unit 1 shown in Figure 1 is realized, for example, by the processor 201 executing a computer program (software) stored in the storage 203. Some or all of these functional units may be realized by hardware (including circuitry) such as LSI (Large Scale Integration), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), and GPU (Graphics Processing Unit), or by the cooperation of software and hardware.

[0028] [Method for generating a 3D model] (First embodiment)

[0029] (Machine learning methods) Prior to generating the 3D model, the correlation between 3D point cloud data and 3D CAD data of the structure is trained in a machine learning model 22.

[0030] Figure 3 is a flowchart of the 3D model generation method according to the first embodiment.

[0031] (Acquisition of 3D point cloud data) As shown in Figure 3, in step S1, the data acquisition unit 11 acquires 3D point cloud data representing the 3D point cloud of the object (see Figure 1). The data acquisition unit 11 may acquire 3D point cloud data from the imaging device 102 or via the communication unit 2. The 3D point cloud data includes points representing structural surfaces of the building (walls, floors, ceilings, etc.) and points representing structures installed inside the building (pumps, turbines, speed reducers, refrigerators, control panels, etc.). The 3D point cloud data includes, for example, information on 3D coordinates.

[0032] (Identification of 3D CAD data) In steps S3 and S4, the information processing unit 12 compares the 3D point cloud data of the structure (pump, etc.) obtained by the data acquisition unit 11 with the 3D CAD data of the structure (pump, etc.) contained in the data file 21 using the learning model 22.

[0033] More specifically, in step S3, the information processing unit 12 overlays the 3D point cloud data and the 3D CAD data. In step S4, the information processing unit 12 compares the surface shape of the 3D point cloud data with the surface shape of the 3D CAD data and determines the degree of agreement between the two surfaces. If the degree of agreement between the surface shape of the 3D point cloud data and the surface shape of the 3D CAD data is greater than or equal to a preset value, the information processing unit 12 can determine that the 3D point cloud data and the 3D CAD data are a match.

[0034] In steps S3 and S4, the information processing unit 12 identifies 3D CAD data of structures that match the 3D point cloud data from among the 3D CAD data related to the structures. In step S5, the information processing unit 12 stores the 3D CAD data that has been determined to match.

[0035] Since 3D CAD data includes attribute information, by identifying the 3D CAD data of a structure that matches the 3D point cloud data, both location information and attribute information can be associated with the structure. The information processing unit 12 stores the attribute information and location information of the structure in association.

[0036] (Generation of 3D models) In step S6, a 3D model is generated by combining the 3D CAD data of the structures (such as pumps) that have been determined to match with the 3D point cloud data of the building (for example, 3D point cloud data of structural surfaces (walls, floors, ceilings, etc.)). The 3D model can be used for BIM (Building Information Modeling), CIM (Construction Information Modeling), etc.

[0037] In step S4, if the degree of agreement between the surface shape of the 3D point cloud data and the surface shape of the 3D CAD data does not reach a predetermined value, the process returns to step S3, and the surface shape of the 3D point cloud data is compared with the surface shape of other 3D CAD data. In other words, the surface shapes are compared again using other 3D CAD data.

[0038] Identifying 3D CAD data (steps S3-S6) can be performed for multiple structures. For example, 3D CAD data can be identified for each of the following: pumps, turbines, gearboxes, refrigerators, control panels, etc.

[0039] Figure 4 is an explanatory diagram of the 3D model generation method. More specifically, Figure 4 is an explanatory diagram that schematically shows the steps (see Figure 3) up to the identification of 3D CAD data.

[0040] As shown in Figure 4, the information processing unit 12 compares the 3D point cloud data of the structure obtained by the data acquisition unit 11 with the 3D CAD data of the structure contained in the data file 21 using the learning model 22. This yields structure data 300 that contains both positional information and attribute information. Therefore, a 3D model containing the information of data 300 can be generated. Positional information includes, for example, 3D coordinate information.

[0041] [Effects of the 3D model generation apparatus and 3D model generation method according to the embodiment] In the 3D model generation device 100 according to this embodiment, the information processing unit 12 compares the 3D point cloud data and 3D CAD data of a structure using a learning model 22, and identifies the 3D CAD data that matches the 3D point cloud data from among multiple 3D CAD data. As a result, structure data 300 that has both positional information and attribute information is obtained (see Figure 4). Thus, it is possible to generate a 3D model with high accuracy regarding information about the structure (pump, etc.).

[0042] The 3D model generation device 100 directly compares the 3D point cloud data and 3D CAD data of the structure, enabling it to select the correct 3D CAD data. Therefore, it can generate a 3D model with high accuracy regarding information about the structure.

[0043] The 3D model generation device 100 can obtain structural data 300 that includes both positional information and attribute information, and therefore can efficiently generate a 3D model that includes both positional information and attribute information.

[0044] When comparing 3D point cloud data and 3D CAD data, the information processing unit 12 overlays the 3D point cloud data and the 3D CAD data. If the degree of agreement between the surface shape of the 3D point cloud data and the surface shape of the 3D CAD data is greater than or equal to a preset value, it determines that the 3D point cloud data and the 3D CAD data are a match (see step S4 in Figure 3). Therefore, the correct 3D CAD data can be selected. Thus, a 3D model with high accuracy regarding structural information can be generated.

[0045] The information processing unit 12 stores attribute information and location information of structures (such as pumps) in association, thereby enabling the generation of highly accurate 3D models of information related to structures.

[0046] The 3D model generation method according to this embodiment compares 3D point cloud data and 3D CAD data of a structure using a learning model 22, and identifies 3D CAD data that matches the 3D point cloud data from among multiple 3D CAD data. As a result, structure data 300 that has both positional information and attribute information is obtained (see Figure 4). Thus, it is possible to generate a 3D model with high accuracy regarding information about the structure (pump, etc.).

[0047] The computer program according to this embodiment is a computer program for implementing the following functions using a computer. (i) A function to acquire 3D point cloud data including buildings and structures. (ii) A function that compares 3D point cloud data of a structure with 3D CAD data contained in a data file containing multiple 3D CAD data of the structure using a learning model, and identifies the 3D CAD data that matches the 3D point cloud data from among the multiple 3D CAD data. (iii) A function to generate a 3D model of an object by combining the 3D CAD data of the identified structure with the 3D point cloud data of the building.

[0048] [3D Model Generation Device] (Second Embodiment) Figure 5 is a flowchart of a 3D model generation method using a 3D model generation apparatus according to the second embodiment.

[0049] The 3D model generation method in this example differs from the 3D model generation method of the first embodiment (see Figure 3) in that it includes a step S2 for correcting 3D point cloud data between step S1 and step S3.

[0050] In step S2, the information processing unit 12 can generate corrected 3D point cloud data by interpolating the gaps between the point clouds that make up the 3D point cloud data of the structure (pump, etc.) as surfaces. This correction results in 3D point cloud data with fewer defects. Therefore, it becomes easier to compare the 3D point cloud data with 3D CAD data.

[0051] Next, a 3D model can be generated by steps S3 to S6, similar to the 3D model generation method of the first embodiment.

[0052] In this example of a 3D model generation device, the information processing unit 12 generates corrected 3D point cloud data by interpolating the gaps between the point clouds that make up the 3D point cloud data of the structure as surfaces, and this corrected 3D point cloud data can be compared with 3D CAD data. Because the correction results in 3D point cloud data with fewer defects, it becomes easier to compare the 3D point cloud data with the 3D CAD data. Therefore, it is possible to generate a 3D model with high accuracy regarding information about the structure (pump, etc.).

[0053] Although the present invention has been described above based on the best mode, the present invention is not limited to the best mode described above, and various modifications are possible without departing from the spirit of the invention. [Explanation of symbols]

[0054] 3...3D data storage unit, 11...Data acquisition unit, 12...Information processing unit, 22...Learning model, 100...3D model generation device

Claims

1. A three-dimensional model generation apparatus for generating a three-dimensional model of an object having a building and a structure installed inside the building, A data acquisition unit that acquires three-dimensional point cloud data representing a three-dimensional point cloud including the aforementioned building and structure, A 3D data storage unit having a data file containing multiple 3D CAD data of the aforementioned structure, An information processing unit compares the three-dimensional point cloud data of the structure with the three-dimensional CAD data contained in the data file using a learning model, and identifies the three-dimensional CAD data that matches the three-dimensional point cloud data from among a plurality of three-dimensional CAD data; Equipped with, The information processing unit generates a three-dimensional model of the object by combining the three-dimensional CAD data of the identified structure and the three-dimensional point cloud data of the building. A 3D model generation device.

2. When comparing the three-dimensional point cloud data and the three-dimensional CAD data, the information processing unit superimposes the three-dimensional point cloud data and the three-dimensional CAD data, and determines that the three-dimensional point cloud data and the three-dimensional CAD data match if the degree of agreement between the surface shape of the three-dimensional point cloud data and the surface shape of the three-dimensional CAD data is greater than or equal to a preset value. A three-dimensional model generation apparatus according to claim 1.

3. The information processing unit generates corrected 3D point cloud data by interpolating the gaps between the point clouds constituting the 3D point cloud data of the structure as surfaces, and compares the corrected 3D point cloud data with the 3D CAD data. A three-dimensional model generation apparatus according to claim 1.

4. The aforementioned information processing unit stores the attribute information and location information of the structure in association with each other. A three-dimensional model generation apparatus according to claim 1.

5. The structure having the three-dimensional CAD data is at least one of the following: a pump, a turbine, a speed reducer, a refrigerator, and a control panel. A three-dimensional model generation apparatus according to claim 1.

6. A method for generating a three-dimensional model of an object having a building and a structure installed inside the building, The steps include: acquiring three-dimensional point cloud data representing a three-dimensional point cloud including the aforementioned building and structure; The steps include comparing the three-dimensional point cloud data of the structure with the three-dimensional CAD data included in a data file containing multiple three-dimensional CAD data of the structure using a learning model, and identifying the three-dimensional CAD data that matches the three-dimensional point cloud data from among the multiple three-dimensional CAD data, The method includes the step of generating a three-dimensional model of the object by combining the three-dimensional CAD data of the identified structure and the three-dimensional point cloud data of the building. Method for generating 3D models.

7. A computer program for generating a three-dimensional model of an object having a building and a structure installed inside the building, A function to acquire three-dimensional point cloud data representing a three-dimensional point cloud including the aforementioned building and structure, A function that compares the three-dimensional point cloud data of the structure with the three-dimensional CAD data included in a data file containing multiple three-dimensional CAD data of the structure using a learning model, and identifies the three-dimensional CAD data that matches the three-dimensional point cloud data from among the multiple three-dimensional CAD data, A computer program for realizing, by computer, the function of generating a three-dimensional model of an object by combining the three-dimensional CAD data of the identified structure and the three-dimensional point cloud data of the building.

Images