Method, information processing device, and program

Abstract

To provide multiple types of attribute information of a building more quickly.SOLUTION: A method causes a computer to: acquire map information in a first range from map information; and acquire, based on the map information in the first range, multiple types of attribute information of a first building included in the first range that is not explicitly indicated in the map information. The map information includes a first map image in which at least regions of buildings and other regions are colored differently, and an aerial photograph image. The computer acquires the area of the first building as one piece of the attribute information of the first building based on the first map image in the first range, and acquires a roof type of the first building as one piece of the attribute information of the first building based on the aerial photograph image in the first range.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a technique for searching for real estate.

Background Art

[0002] Techniques for analyzing aerial photo images and extracting the contour shape of the roof of a specified building are disclosed (for example, Patent Document 1). Also, techniques for interpreting roof shapes using DSM (Digital Surface Model) data are disclosed (for example, Non-Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Non-Patent Documents

[0004]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, obtaining other attribute information along with the shape of the roof of a building is not disclosed, and when it is desired to obtain other attribute information of a building, it is necessary to refer to another database. As a result, it may take time to present the attribute information of the building. The attribute information of a building includes, for example, roof type, building area, corporation that owns or uses it, and location information. However, the attribute information of a building is not limited to these.

[0006] An object of the present invention is to provide an information processing method, an information processing apparatus, and a program capable of more quickly providing a plurality of types of attribute information of a building.

Means for Solving the Problem

[0007] One aspect of the present invention is a method in which a computer acquires map information within a first range from map information, and acquires a plurality of types of attribute information of a first building included within the first range that is not explicitly shown in the map information, based on the map information within the first range. The map information is information regarding the division of areas such as buildings, roads, rivers, and railways on the ground. The map information may be, for example, publicly available on the Internet or privately held without being publicly available. According to one aspect of the present invention, it is possible to acquire a plurality of types of attribute information of the first building from the map information without referring to a plurality of databases, and it is possible to more quickly provide a plurality of types of attribute information of the first building.

[0008] In one aspect of the present invention, the map information may include a first map image in which at least the area of the building and other areas are colored separately. Further, the computer may acquire the area of the first building as one of the attribute information of the first building based on the first map image within the first range. The area of the building is the area obtained by the outer perimeter when looking directly down at the building, that is, the so-called floor area. In one aspect of the present invention, the map information may include an aerial photograph. Further, the computer may acquire the roof type of the first building as one of the attribute information of the first building based on the aerial photograph image within the first range. According to one aspect of the present disclosure, it is possible to quickly provide the roof type or area of the first building from the map information without referring to a plurality of databases.

[0009] Further, the computer may obtain an aerial photo image of the first building cut out from the aerial photo image within the first range based on the first map image within the first range, and determine the roof type of the first building by performing image analysis on the aerial photo image of the first building. The aerial photo image of the building can also be said to be an image of the roof of the building. The image analysis may be performed using a machine learning model that has learned the relationship between the aerial photo image of the building and the roof type, for example. When using convformer as the machine learning model for performing image analysis on the aerial photo image of the first building, it can be learned with less training data, and the roof type can be determined faster and more accurately. However, the machine learning model for performing image analysis on the aerial photo image of the first building is not limited to convformer.

[0010] Further, the computer may further obtain the first map image and the aerial photo image within the second range. The second range is a range with a different zoom level from the first range and includes the first building. In this case, the computer may obtain the area of the first building based on either the first map image within the first range or the map image within the second range, and obtain the roof type of the first building based on either the aerial photo image within the first range or the aerial photo image within the second range.

[0011] For example, in the aerial photo image within the first range, the zoom level is low, the aerial photo image of the first building is not clear, and it may affect the determination accuracy of the roof type. Or, for example, conversely, in the aerial photo image within the first range, the zoom level is too high, and a part of the first building is missing, and it may not be possible to accurately obtain the area of the first building. According to one aspect of the present invention, since the first map image and the aerial photo image within the second range are also obtained, by using either one, the roof type and the area of the first building can be obtained more accurately.

[0012] One of the other aspects of the present invention can also be specified as an information processing apparatus that executes the above method. Specifically, the information processing apparatus acquires map information within a first range from map information, and based on the map information within the first range, acquires a plurality of types of attribute information of a first building included within the first range that is not explicitly shown in the map information, and includes a control unit that executes the above.

[0013] Also, one of the other aspects of the present invention can also be specified as a program for causing a computer to execute the above method. Specifically, the program causes the computer to acquire map information within a first range including a first building from map information, and based on the map information within the first range, acquire a plurality of types of attribute information of the first building included within the first range that is not explicitly shown in the map information. Note that one of the other aspects of the present invention can also be regarded as a non-transitory storage medium storing the above program.

Advantages of the Invention

[0014] According to the present invention, it is possible to more quickly provide a plurality of types of attribute information of a building.

Brief Description of the Drawings

[0015]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Mode for Carrying Out the Invention

[0016] Hereinafter, embodiments of the present invention will be described with reference to the drawings. The configurations of the following embodiments are examples, and the present invention is not limited to the configurations of the embodiments.

[0017] <First Embodiment> FIG. 1 is an example of the system configuration of a building attribute information providing system 100 according to the first embodiment. The building attribute information providing system 100 includes a server 1, a terminal 2, and a map information providing server 3. The server 1, the terminal 2, and the map information providing server 3 are connected to a network N1 and can communicate through the network N1. The communication network N1 may be a public network such as the Internet, or may be a private network such as an in-house network.

[0018] The map information providing server 3 is, for example, a server managed by a map producer, and is a server that publishes a map information database (DB) 31 on the Internet. The map information DB 31 holds map information. The map information DB 31 is published, for example, free of charge or for a fee. The map information held by the map information DB 31 includes multiple types of map image data. The map images held in the map information DB 31 include, for example, standard map image data in which buildings, roads, rivers, railways, etc. and other areas are colored and displayed separately, and aerial photo image data, etc. Also, in each type of map image data, longitude and latitude position coordinate information is assigned to each building, etc. Hereinafter, when simply referred to as map image data and map image, they shall indicate the image data of the standard map and the image of the standard map, respectively.

[0019] In the first embodiment, the server 1 analyzes the map image data and aerial photo image data held in the map information DB 31 to obtain attribute information that is not explicitly stated in the map information held in the map information DB 31, such as the size and shape of actual buildings, and constructs a digital twin. A "digital twin" is constructed by determining the size, color, and shape of actual buildings, facilities, land, etc. from digitized map information, aerial photos, satellite images, etc., and reconstructing them as digital data. For the digital twin, for example, a "digital twin of the roof of a building", a "digital twin of a parking lot", a "digital twin of unused land", etc. are assumed. More specifically, in the first embodiment, the server 1 analyzes the map image data held in the map information DB 31 to obtain the area of the building. In the first embodiment, when referred to as the "area of the building", it shall indicate the area obtained by the outer perimeter when looking directly down at the building, that is, the so-called building area. Also, the server 1 analyzes the aerial photo image held in the map information DB 31 to determine the roof type of the building.

[0020] ​Server 1 creates a database of the attribute information of buildings that is not explicitly shown in the map information obtained by analyzing the map information held in the map information DB 31, and provides the attribute information of the target building to terminal 2 in response to a request from terminal 2. Alternatively, Server 1 analyzes the map information held in the map information DB 31 in response to a request from terminal 2 to obtain the attribute information of the target building that is not explicitly shown in the map information, and provides it to terminal 2. In either case, Server 1 can quickly provide a plurality of attribute information of the building that is not explicitly shown in the map information of the building, such as the building area and roof type, by referring to the database of the attribute information or analyzing the map information held in the map information DB 31.

[0021] FIG. 2 is a diagram showing an example of the hardware configuration of Server 1. Server 1 is, for example, a dedicated computer or a general-purpose computer such as a PC (Personal Computer). Server 1 includes, for example, as hardware components, a CPU (Central Processing Unit ) 101, a memory 102, an auxiliary storage device 103, and a communication unit 104, and is an information processing device in which these are connected to each other by a bus. Server 1 is an example of an "information processing device".

[0022] The communication unit 104 is connected to, for example, a wired network or a wireless network. The communication unit 104 is, for example, a NIC (Network Interface Card), a wireless LAN (Local Area Network) card, or a wireless circuit for connecting to a mobile phone network. Data received by the communication unit 104 is output to the CPU 101.

[0023] The memory 102 is a storage device that provides a storage area and a work area for the CPU 101 to load the program stored in the auxiliary storage device 103, or is used as a buffer. The memory 102 is, for example, a semiconductor memory such as a RAM (Random Access Memory).

[0024] The auxiliary storage device 103 stores various programs and data used by the CPU 101 when executing each program. The auxiliary storage device 103 is, for example, an EPROM (Erasable Programmable ROM), a hard drive disc, or an SSD (Solid State Drive). The auxiliary storage device 103 stores, for example, an operating system (OS ), a building attribute information acquisition program, an API for using the map information DB 3, an image analysis program, a machine learning model, a building attribute information providing program, and various other application programs. The building attribute information acquisition program is a program that analyzes the map information DB 31 to acquire the attribute information of a building. The image analysis program is a program for performing predetermined image analysis on an image. The building attribute information providing program is a program for providing the attribute information of a building. The auxiliary storage device 103 is an example of a "storage unit".

[0025] The CPU 101 executes various processes by loading the OS and various application programs held in the auxiliary storage device 103 into the memory 102 and executing them. The CPU 101 may be one or a plurality. The CPU 101 is an example of a "control unit".

[0026] Note that the hardware configuration of the server 1 shown in FIG. 2 is an example and is not limited to the above. Depending on the embodiment, components can be omitted, replaced, or added as appropriate. For example, the server 1 may include an input device such as a keyboard and a mouse, and an output device such as a display. For example, the server 1 drives a portable recording medium and reads the data recorded on the portable recording medium It may be provided with a portable recording medium driving device for extraction. The portable recording medium is, for example, a USB (Universal Serial Bus) memory, a disc recording medium such as a CD (Compact Disc), a DVD (Digital Versatile Disc), or a Blu-ray (registered trademark) disc, or a recording medium such as a flash memory card.

[0027] The terminal 2 is, for example, a PC, a smartphone, a tablet terminal, or the like. As a hardware configuration, the terminal 2 includes, for example, a CPU, a memory, an auxiliary storage device, a communication unit, a touch panel display, a microphone, a speaker, and a camera. The communication unit included in the terminal 2 may be connected to either a wired network or a wireless network. The terminal 2 may, for example, connect from a browser to the website of the service of the building attribute information providing system 100 and transmit a request to the server 1 from the website, or install a program for the client of the building attribute information providing system 100 and transmit a request to the server 1 through the execution of the program.

[0028] Also, for example, the server 1 and the terminal 2 may be managed by the same administrator or by different administrators. For example, when Company B uses the building attribute providing service provided by Company A, the server 1 may belong to Company A and the terminal 2 may belong to Company B. Also, the terminal 2 is not limited to a terminal belonging to an organization such as a company and may be a terminal owned by an individual user.

[0029] FIG. 3 is a diagram showing an example of the functional configuration of the server 1. The server 1 includes, as functional components, an attribute information acquisition unit 11, a map information acquisition unit 12, an image analysis unit 13, a roof type determination model 14, a presentation control unit 15, and a building attribute information DB 16. These functional components are functional components achieved by the CPU 101 of the server 1 executing a predetermined program held in the auxiliary storage device 103.

[0030] The attribute information acquisition unit 11 controls the acquisition process of the building's attribute information. Details of the acquisition process of the building's attribute information will be described later. The attribute information acquisition unit 11 stores the acquired building attribute information in the building attribute information DB 16. Hereinafter, the building's attribute information is also referred to as building attribute information.

[0031] The map information acquisition unit 12 is, for example, an API for using the map information DB 31. The map information acquisition unit 12 accesses the map information providing server 3 according to an instruction from the attribute information acquisition unit 11, and acquires from the map information providing server 3 the map image data and aerial image data of a predetermined range held in the map information DB 31, or acquires the building's attribute information from the position coordinate information. The building's attribute information acquired from the position coordinate information by the map information acquisition unit 12 includes, for example, the address, the name of the corporation that owns or uses the building, etc. The map information acquisition unit 12 outputs the data acquired from the map information DB 31 to the attribute information acquisition unit 11.

[0032] The image analysis unit 13 performs a predetermined image analysis process on the map image data according to an instruction from the attribute information acquisition unit 11. In the first embodiment, the image analysis unit 13 is the open-source image analysis software OpenCV. In the first embodiment, the image analysis unit 13 determines the color information of the map image, leaves the color information in the building area, and converts other areas to white. The image analysis unit 13 outputs the map image data after the conversion of the color information to the attribute information acquisition unit 11.

[0033] The roof type determination model 14 is a machine learning model used for determining the roof type of a building. The roof type determination model 14 has learned the relationship between the aerial photo image of the building and the roof type, and outputs an estimated result of the roof type for the input of the aerial photo image of the building.

[0034] When the prompt control unit 15 receives a request for building attribute information from the terminal 2, it transmits the specified building attribute information to the terminal 2. In the first embodiment, when the building attribute information specified by the terminal 2 is stored in the building attribute information DB 16, the prompt control unit 15 reads and acquires it from the building attribute information DB 16. When the building attribute information specified by the terminal 2 is not stored in the building attribute information DB 16, the prompt control unit 15 requests the attribute information acquisition unit 11 to acquire the attribute information of the specified building and acquires it. Further, when the prompt control unit 15 receives not only the attribute information of the specified building but also a predetermined condition regarding the attribute information from the terminal 2, it searches for and presents buildings with attribute information that satisfies the condition, or presents the number of buildings with attribute information that satisfies the condition within a predetermined range, or the area of the building, or the ratio of the total value of the number of buildings with attribute information that satisfies the condition within the predetermined range or the total area of the building to the total number of buildings or the total area of all buildings in the predetermined range. A specific example of the method for presenting attribute information by the prompt control unit 15 will be described later.

[0035] The building attribute information DB 16 stores the building attribute information acquired by the attribute information acquisition unit 11. The building attribute information DB 16 is created in the storage area of the auxiliary storage device 103. Note that the functional configuration of the server 1 shown in FIG. 3 is an example and is not limited thereto.

[0036] FIG. 4 is a diagram showing an example of the information stored in the building attribute information DB 16 of the server 1. The building attribute information DB 16 stores the building attribute information. In the example shown in FIG. 4, one record of the building attribute information DB 16 includes fields for building ID, position coordinate information, address, building name, corporate information, roof type, area, and aerial photo image.

[0037] In the field of building ID, the identification information of the building is stored. The identification information of the building may be uniquely assigned to each building by the attribute information acquisition unit 11, for example. In the field of position coordinate information, latitude and longitude are stored as the position coordinate information of the building. For the position coordinate information of the building, for example, the latitude and longitude of the representative point of the building are used. The representative point of the building is, for example, the center point of the building, or the midpoint of the frontage, etc.

[0038] In the field of address, the address of the building is stored. In the field of building name, the name, common name, or pet name, etc. of the building are stored. In the field of corporate information, when the building is used or owned by a corporation, information about the corporation is stored. The information about the corporation includes, for example, the corporate name, business type, and business format, etc. However, the information about the corporation is not limited to these. When the building is used or owned by an individual, the field of corporate information is empty.

[0039] In the field of roof type, the roof type of the building is stored. In the field of area, the area of the building is stored. In the field of aerial photo image, the aerial photo image of the building or the address of its storage location is stored.

[0040] Among the fields included in the record of the building attribute information DB 16 in the example shown in FIG. 4, the position coordinate information, address, building name, corporate information, roof type, and area are the attribute information of the building, respectively. The details of the acquisition of each attribute information will be described later. In the example shown in FIG. 4, the records of the information in the building attribute information DB 16 are shown in tabular rows. However, the building attribute information DB 16 is not limited to a tabular database. For example, the building attribute information DB 16 may be constructed by a hierarchical language including keywords and values, such as HyperText Markup Language (HTML), Extensible Markup Language (XML ) etc.

[0041] (Details of the acquisition process of building attribute information) FIG. 5 is an example of a flowchart of building attribute information acquisition processing. The example shown in FIG. 5 is started when a predetermined event occurs. Events that serve as start triggers for building attribute information acquisition processing include, for example, when an acquisition instruction for the attribute information of a specified building is input from the administrator of server 1, and when the attribute information of the building specified by a request for building attribute information from terminal 2 is not held in building attribute information DB 16. The execution entity of the processing shown in FIG. 5 is CPU 101 of server 1, but for the sake of convenience, the description will be based on functional components.

[0042] In OP10, the attribute information acquisition unit 11 acquires the location information of the target building. The location information of the target building is acquired by input from the administrator of server 1 or by receiving it together with a request for building attribute information from terminal 2. The location information of the target building is, for example, latitude and longitude when the specified method of the target building is selection on a map. The location information of the target building is, for example, the address when the specified method of the target building is by address. For example, when the specified method of the target building is by the name of the building, it is the name of the building. Note that the name of the building may be not only the official name but also a common name and a pet name.

[0043] In OP20, the attribute information acquisition unit 11 executes a roof image acquisition process for acquiring an aerial photograph image of the target building, which is obtained by cutting out the aerial photograph image at the outer periphery of the target building and is an input to the roof type determination model 14. The aerial photograph image of the building cut out at the outer periphery of the building can also be said to be an image of the roof of the building. Therefore, hereinafter, the aerial photograph image of the building cut out at the outer periphery of the building will be referred to as a roof image. Details of the roof image acquisition process will be described later. By executing the roof image acquisition process, the attribute information acquisition unit 11 also acquires the area of the target building together with the roof image of the target building.

[0044] In OP30, the attribute information acquisition unit 11 inputs the roof image of the target building into the roof type determination model 14, and determines the roof type of the target building based on the output of the roof type determination model 14 in that case. Details of the determination process of the roof type using the roof type determination model 14 will be described later.

[0045] In OP40, the attribute information acquisition unit 11 acquires information about the corporation that owns or uses the target building as one of the attribute information of the target building. The attribute information acquisition unit 11 may acquire information about the corporation that owns or uses the target building from, for example, the information attached to the building in the map information DB 31, or may acquire it by searching the Internet from the address of the building. Further, the information about the corporation that owns or uses the target building may be obtained by input from the administrator of the server 1, or may be received together with a request for acquisition of building attribute information from the terminal 2.

[0046] In OP50, the attribute information acquisition unit 11 stores the attribute information of the target building acquired by the processes from OP10 to OP40 in the building attribute information DB 16. Then, the process shown in FIG. 5 ends. When an acquisition instruction for the attribute information of a specified building is input from the administrator of the server 1 and the process shown in FIG. 5 is started, then, the attribute information acquisition unit 11 may output, for example, a completion notice of the process. When the process shown in FIG. 5 is started because the attribute information of the building specified by the request for the building attribute information from the terminal 2 is not held in the building attribute information DB 16, then, the attribute information acquisition unit 11 transmits the attribute information of the target building to the terminal 2.

[0047] Next, with reference to FIGS. 6 to 10, the roof image acquisition process executed in OP20 of FIG. 5 will be described. FIG. 6 is an example of a flowchart of the roof image acquisition process. FIGS. 7 to 10 are diagrams showing an example of the processing result at each stage of the processing included in the roof image acquisition process executed for the target building when a certain building is the target building.

[0048] In OP21 of FIG. 6, the attribute information acquisition unit 11 instructs the map information acquisition unit 12 to obtain, from the map information providing server 3, two map image data at different zoom levels centered on the target building. The map information acquisition unit 12 transmits, to the map information providing server 3, for example, the position information of the target building and two zoom levels together with a request for acquisition of a map image. The position information of the building is the information acquired in OP10 of FIG. 5. The map information providing server 3 returns, as a response, a map image at the specified zoom level centered on the specified position information in response to the request for acquisition of a map image. The size of the map image provided by the map information providing server 3 is a predetermined size regardless of the zoom level.

[0049] One zoom level is the lowest zoom level at which the shape of each individual building can be determined in the map information providing server 3. The other zoom level is a zoom level that is a predetermined number higher than the lowest zoom level at which the shape of each individual building can be determined. The zoom level determines the range of the world displayed on the map image. The zoom level can also be said to be a scale. Generally, it is shown that the smaller the value of the zoom level, the lower it is, and the larger the value, the higher it is. When the zoom level becomes lower, the range of the world shown on the map image becomes wider and moves away from the target building. When the zoom level becomes higher, the range of the world shown on the map image becomes narrower and approaches the target building. The value of the lowest zoom level at which the shape of each individual building can be determined varies depending on each map creator and can be specified from publicly available information.

[0050] Also, the map information providing server 3 provides a service called reverse geocoding that converts latitude and longitude, address, and building name into each other. When acquiring map image data centered on the target building, the map information acquisition unit 12 uses the reverse geocoding of the map information providing server 3 to acquire the missing attribute information among latitude and longitude, building name, and address from the position information of the target building.

[0051] FIG. 7 is an example of an image of a standard map including a target building at two zoom levels obtained from the map information DB 31. From FIG. 7 to FIG. 10, it is assumed that Google Map (registered trademark) is used as the map providing service provided by the map information providing server 3. That is, from FIG. 7 to FIG. 10, the map information acquisition unit 12 is the Goole Map API. In Google Map (registered trademark), the lowest zoom level at which the shape of each building can be discriminated is zoom level 17. In FIG. 7, images of the standard map centered on the target building at zoom level 17 and zoom level 20 are shown.

[0052] In OP22 of FIG. 6, the attribute information acquisition unit 11 extracts buildings for each of the map image data at the two zoom levels acquired in OP21. In the first embodiment, first, the attribute information acquisition unit 11 instructs the image analysis unit 13 to acquire an image in which the area of the building has been extracted from the map image. The image analysis unit 13 performs color information determination on the map image using OpenCV, which is open source image analysis software. The color information of the areas painted for buildings in the standard map is left, and the color information of other areas is converted to white. As a result, the color used in the standard map remains in the area of the building, and other areas become white. The color used for buildings in the standard map has been acquired in advance. Next, the attribute information acquisition unit 11 converts the image in which the area of the building has been extracted from the map image so that the areas with remaining color information are white and the areas with white color information are black, and binarizes it into black and white. As a result, an image is obtained in which the area of the building is white and other areas are black.

[0053] FIG. 8 is a diagram showing an example of an image in which buildings are extracted from the map image shown in FIG. 7. In each of the images at zoom level 17 and zoom level 20, the buildings are displayed in white.

[0054] In OP23 of FIG. 6, the attribute information acquisition unit 11 acquires the buildings from the map image acquired in OP22 In the extracted image, leave the building including the center point and make the other areas black to generate a mask image of the target building, and obtain the area of the building from the mask image.

[0055] Figure 9 is an example of a mask image of the target building. In the mask image, the area of the target building is white, and the other areas are black. The area of the target building can be obtained, for example, from the number of dots included in the white area in the mask image. Since the map image data has scale data, the area per dot of the mask image can be obtained. The value obtained by multiplying the area per dot of the mask image by the number of dots having white color information in the mask image is the area of the target building.

[0056] Here, depending on the size of the target building, in the map image at zoom level 20, the entire target building may not fit within the range of the map image. In this case, the area of the building cannot be accurately obtained from the mask image at zoom level 20. On the other hand, zoom level 17 is the lowest zoom level at which the shape of each individual building can be distinguished in Google Map (registered trademark), so the entire target building fits in the mask image at zoom level 17. Therefore, in OP23 of FIG. 6, the attribute information acquisition unit 11 obtains the area of the target building using the mask image at the lowest zoom level at which the shape of each individual building can be distinguished.

[0057] On the other hand, in the mask image at zoom level 17, the target building may be too small, and the roof image cut out from the aerial photo image may be an unclear image. When an unclear roof image is input to the roof type determination model 14, the accuracy of the determination of the roof type may not be sufficiently obtained. That is, in order to obtain a higher determination accuracy of the roof type, it is desirable that the roof image is clearer. Therefore, in the first embodiment, the roof image is obtained using the mask image with a higher zoom level.

[0058] In OP24 of FIG. 6, the attribute information acquisition unit 11 instructs the map information acquisition unit 12 to acquire aerial photo image data centered on the target building at the higher of the two zoom levels. The map information acquisition unit 12 transmits, to the map information providing server 3, a request for acquiring the aerial photo image data together with, for example, the position information of the target building and the higher zoom level. The position information of the target building is the same information as that used in OP21. As a result, aerial photo image data in the same range as the map image data at the higher zoom level acquired in OP21 is acquired.

[0059] In OP25, the attribute information acquisition unit 11 overlays the higher-zoom mask image acquired in OP23 on the aerial photo image acquired in OP24, cuts out the aerial photo image corresponding to the white area in the mask image, and acquires the roof image of the target building. After that, the process shown in FIG. 6 ends, the process proceeds to OP30 in FIG. 5, the roof image is input to the roof type determination model 14, and the roof type of the target building is determined.

[0060] FIG. 10 is a diagram showing an example of acquiring a roof image from a mask image and an aerial photo image. In FIG. 10, the mask image and the aerial photo image at zoom level 20 are shown. When the mask image and the aerial photo image are overlaid, the black area in the mask image is masked in the aerial photo image. By overlaying the mask image and the aerial photo image and cutting out the unmasked area, an aerial photo image of only the area corresponding to the target building in the aerial photo image, that is, the roof image of the target building is acquired.

[0061] Even if the entire target building does not fit within the higher-zoom mask image and the roof image is an image with a part of the roof missing, the influence on the determination accuracy by the roof type determination model 14 is small, and it is considered that the accuracy drops less than when using the roof image acquired using the lower-zoom mask image. image is used.

[0062] As shown in the examples illustrated in FIGS. 6 to 10, by acquiring map images at two zoom levels and performing parallel processing, it is possible to accurately acquire each of the building area and the roof type in a single process.

[0063] Note that, without acquiring map images at two zoom levels, one map image at a predetermined zoom level may be acquired from the map information providing server 3, and the building area and the roof type may be acquired as described above using the one map image at the predetermined zoom level. The predetermined zoom level may be the lowest zoom level at which the shape of each individual building can be discriminated, or may be a zoom level of +α or -α from the lowest zoom level at which the shape of each individual building can be discriminated.

[0064] In this case, for example, the attribute information acquisition unit 11 may use a threshold value based on the ratio of the area (or the number of dots) of the target building to the area (or the number of dots) of the entire mask image created from the map image at the predetermined zoom level to determine whether a roof image with sufficient sharpness to obtain sufficient determination accuracy of the roof type can be obtained, or whether the entire building is included in the mask image without any gaps.

[0065] When the ratio is less than the first threshold value, the attribute information acquisition unit 11 determines that although the entire target building is included in the mask image, a roof image with sufficient sharpness to obtain sufficient determination accuracy of the roof type cannot be obtained. In this case, the attribute information acquisition unit 11 acquires, through the map information acquisition unit 12, a map image at a zoom level higher than the predetermined zoom level from the map information providing server 3, and acquires the roof image of the target building from the acquired map image at the higher zoom level. Note that the area of the target building is acquired from the map image at the predetermined zoom level.

[0066] When the ratio is equal to or greater than a second threshold value that is greater than the first threshold value, although a roof image with sufficient clarity for obtaining the determination accuracy of the roof type can be obtained, the attribute information acquisition unit 11 determines that the entire target building is not included in the mask image. In this case, the attribute information acquisition unit 11 acquires a map image at a zoom level lower than a predetermined zoom level from the map information providing server 3 through the map information acquisition unit 12, and acquires the area of the target building from the acquired map image at a low zoom level. Note that the roof image of the target building is acquired from a map image at a predetermined zoom level.

[0067] When the ratio is equal to or greater than the first threshold value and less than the second threshold value, the attribute information acquisition unit 11 determines that a roof image with sufficient clarity for obtaining the determination accuracy of the roof type can be obtained from a map image at a predetermined zoom level, and that the entire target building is included in the mask image. In this case, the attribute information acquisition unit 11 does not acquire a map image at a zoom level different from the predetermined zoom level.

[0068] Next, the roof type determination process in OP30 of FIG. 5 will be described. FIG. 11 is a diagram showing an example of the input and output of the roof type determination model 14. The roof type determination model 14 is a machine learning model that learns the relationship between a roof image and a roof type using a roof image of a building with a labeled roof type as teacher data, with a predetermined number of teacher data. The learning of the roof type determination model 14 may be performed by the attribute information acquisition unit 11 controlling at the server 1, or may be performed by a device other than the server 1.

[0069] The input of the roof type determination model 14 is a roof image. When a roof image is input to the roof type determination model 14, scores for each roof type are output from the roof type determination model 14. The attribute information acquisition unit 11 determines the roof type of the roof image for which the highest score is input as the roof type. For example, when the scores of all roof types are less than a predetermined threshold value, the attribute information acquisition unit 11 may determine that the roof type cannot be determined. For example, when the scores of all roof types are less than a predetermined threshold value, the attribute information acquisition unit 11 may determine that the roof type cannot be determined.

[0070] In the first embodiment, the roof types are "flat roof", "parking lot", "half roof", "tile roof", and "solar panels installed". The roof type "flat roof" is a roof that is horizontal or has almost no gradient. The roof type "parking lot" indicates a building where the rooftop is used as a parking lot. The roof type "half roof" is a roof made by processing a metal plate. The roof type "tile roof" is a roof covered with tiles. The roof type "solar panels installed" is a roof with solar panels already installed. Note that the roof type can be arbitrarily set by the administrator of the building attribute information providing system 100.

[0071] The roof type determination model 14 can be, for example, any of CNN (Convolutional Neural Network), Vision Transformer, convfo rmer, and other machine learning models capable of determining image features. CNN is often used for classification tasks based on image features. CNN is a method that captures local features of an image using a filter with a small target range to capture the features of the image and learns the feature patterns. When using CNN, it becomes easier to capture local features, but when there are features over a wide range of the image, the accuracy may not easily increase.

[0072] The Transformer is a model developed for natural language analysis. The Transformer is a method that decomposes a sentence into words, vectorizes them, and interprets the sentence based on the order and relevance between words. The Vision Transformer is an application of the Transformer to images. The Vision Transformer performs image analysis based on the individual vectors of finely classified image data and their order and positional relationships. In the Vision Transformer, because it can capture positional relationships that are a little distant, it is possible to improve the accuracy even when there are features over a wide range of the image. Also, the Transformer can be learned with relatively little training data by mathematically discovering patterns between elements. However, it has the characteristic that it is difficult to capture local spatial features like CNN. Also, since it does not use convolutional processing in the filter, the number of parameters tends to be large and the amount of calculation becomes enormous.

[0073] Convformer is a model that hybridizes CNN and Transformer. Convformer combines a model that grasps the relevance with features at distant positions of the Transformer and a model that extracts local feature amounts with an image filter of CNN, and is a model that can achieve high performance with few parameters (less computational cost). In the first embodiment, it has been confirmed that when Convformer is adopted as the roof type determination model 14, the roof type can be determined more accurately, faster, and with less training data.

[0074] For example, when aerial photo images are acquired at 1280 pixels × 1280 pixels and the roof image is 384 pixels × 384 pixels, the processing time by Convformer is 0.218 seconds and the accuracy score is 0.9274. Under the same conditions, the processing time by the Transformer is 0.553 seconds and the accuracy score is 0.8548. Under the same conditions, the processing time by CNN is 0.558 seconds and the accuracy score is 0.621.

[0075] In addition, since the roof image, which is the input of the roof type determination model 14, is obtained as an image without a background image using a mask image from an aerial photo image, the area for performing the relevance analysis can be limited, thereby reducing the calculation cost. Therefore, even with CPU-based operations without using a GPU or the like, it takes about 0.5 seconds per roof image, and it can be determined with high accuracy.

[0076] By using convformer as the roof type determination model 14 and using the roof image as described above as the input, as a cloud service, when a request for the attribute information of a specified building from the terminal 2 is received, the attribute information of the specified building can be obtained and the attribute information of the target building can be returned without making the user feel a delay.

[0077] In addition, in FIGS. 5 to 11, the acquisition process of the building attribute information is described assuming the case of acquiring the attribute information for a building with specified position information. However, the attribute information can be acquired in the same manner even when no specific individual building is clearly specified as the target building. The case where no specific individual building is clearly specified as the target building is, for example, when the range for acquiring the attribute information of buildings is specified, such as in the unit of a municipality or a user-specified range, rather than a specific individual building.

[0078] When a range is specified, the attribute information acquisition unit 11 instructs the map information acquisition unit 12 to transmit the predetermined position coordinate information within the specified range to the map information providing server 3 together with a request, and acquires the attribute information about the building located at the center in the map image centered on the predetermined position coordinate information from the map information providing server 3, and this may be repeatedly performed while moving the predetermined position coordinate information within the specified range. The initial value of the predetermined position coordinate information within the specified range may be, for example, any of the center point, the end point of the specified range, or a randomly selected point. In this case, although the attribute information acquisition unit 11 grasps the position coordinate information of the target building, it does not grasp the address, the building name, etc. Therefore, the attribute information acquisition unit 11 may instruct the map information acquisition unit 12 to use the reverse geocoding of the map information providing server 3 to acquire the address of the target building and the attribute information given in the map information DB 31 from the position coordinate information. Further, the attribute information acquisition unit 11 may search the Internet or a predetermined database based on those information to acquire the name of the corporation that owns or uses the target building, etc.

[0079] Depending on the map information providing server 3, when a range is specified, it may select a predetermined position from within the specified range and provide a map image centered on the selected position. In such a case, the attribute information acquisition unit 11 may extract all the buildings in the map image acquired from the map information providing server 3 in the same manner as OP22 in FIG. 6, generate a mask image for each of the extracted multiple buildings, acquire the area and the roof type, and repeatedly execute this while shifting the range of the map image so as to cover the specified range.

[0080] In this case, when the attribute information acquisition unit 11 generates a mask image of a certain building, although the position (coordinates) in the coordinate system of the mask image of the building can be obtained, the position coordinate information (latitude and longitude) cannot be obtained. Note that the map image is accompanied by information on latitude and longitude indicating the range of the map image. Therefore, the attribute information acquisition unit 11 converts the position in the coordinate system of the mask image of the target building into latitude and longitude using the latitude and longitude information attached to the map image. The latitude and longitude of the target building may be stored in the building attribute information DB 16 as one of the attribute information of the target building. Also, when the latitude and longitude of the target building are determined, since reverse geocoding of the map information providing server 3 can be used as described above, the attribute information acquisition unit 11 may acquire attribute information such as an address by reverse geocoding from the latitude and longitude of the target building from the map information providing server 3. Note that aerial photo image data for acquiring the roof image of the target building may be acquired as having the same position coordinate information as the map image.

[0081] (Example of providing building attribute information) FIG. 12 is an example of an input screen for search conditions for building attribute information in the terminal 2. In FIG. 12 The screen example shown is an example of an input screen for search conditions for building attribute information when the building attribute information providing system 100 provides a service for providing attribute information of a specified building and a service for searching for buildings with attribute information that satisfy specified conditions within a specified range.

[0082] The screen example shown in FIG. 12 includes an input field for "building specification" and an input field for "range specification". When the user wants to use the service for providing attribute information of a specified building, the user selects the radio button for "building specification". When the user wants to use the service for searching for buildings with attribute information that satisfy specified conditions within a specified range, the user selects the radio button for "range specification".

[0083] In the screen example shown in FIG. 12, the "Building Specification" column includes an input field for the address, an input field for the building name, and a button for screen transition to the map for selection from the map. As methods for specifying the target building, it corresponds to specification by address, specification by building name, and specification on the map. When the address of the target building is entered in the address input field, the address of the target building is transmitted as location information from the terminal 2 to the server 1 together with a request for building attribute information. When the name of the target building is entered in the building name input field, the name of the target building is transmitted as location information from the terminal 2 to the server 1 together with a request for building attribute information. When the location of the target building is selected on the map, the selected latitude and longitude (location coordinate information) are transmitted as location information from the terminal 2 to the server 1 together with a request for building attribute information.

[0084] In the screen example shown in FIG. 12, the "Range Specification" column includes an input field for the address and a button for screen transition to the map for specifying the search range. As methods for specifying the search range, it corresponds to specification by address and specification on the map. In the address input field, a pull-down menu for selecting the prefecture and city / town / village is arranged, and the search range can be set in units of city / town / village. When the address of the search range is entered in the address input field, the name of the specified city / town / village is transmitted as the search range from the terminal 2 to the server 1 together with a search request. When the search range is selected on the map, the information on the latitude and longitude of the specified range (location coordinate information) is transmitted as the search range from the terminal 2 to the server 1 together with a search request.

[0085] In the "Range Specify" column, there are further radio buttons for "Roof Type", "Area", and "Solar Panel Installable" to specify search conditions for attribute information. As search conditions for attribute information, the roof type, the area, and whether a solar panel can be installed can be specified. Note that when the roof type is defined in FIG. 11, the condition that a solar panel can be installed means that the roof type is any one of "Gable Roof", "Half-Hipped Roof", and "Tile Roof". Along with the search request, the search conditions for the selected and specified attribute information are also sent from the terminal 2 to the server 1.

[0086] When the search button is selected, a request for building attribute information or a search request is sent from the terminal 2 to the server 1 according to the selection of the radio button. Note that the input screen for search conditions shown in FIG. 12 is an example, and the configuration of the input screen for search conditions can be arbitrarily set by the administrator of the building attribute information providing system 100.

[0087] FIG. 13 is a diagram showing an example of a presentation screen of building attribute information on the terminal 2. The presentation screen of building attribute information shown in FIG. 13 is a screen for presenting the attribute information of the specified building. Therefore, the presentation screen of building attribute information shown in FIG. 13 includes a map showing the location of the specified building and a display column for the attribute information of the specified building.

[0088] The presentation screen of building attribute information shown in FIG. 13 is, for example, displayed as a response to the request when the radio button of "Building Specify" is selected on the input screen of the building attribute information shown in FIG. 12. When the presentation control unit 15 of the server 1 receives a request for building attribute information and the location information of the target building from the terminal 2, the building attribute information DB ​Search for 16 by location information, read the attribute information of the corresponding building from the building attribute information DB 16, and transmit it to the terminal 2. When there is no attribute information in the building attribute information DB 16 that matches the location information of the target building, the presentation control unit 15 may request the attribute information acquisition unit 11 to acquire the attribute information of the target building, and transmit the acquired attribute information of the target building to the terminal 2.

[0089] The method of presenting building attribute information as shown in FIG. 13 is effective, for example, when grasping in advance the installable area of solar panels and the type of solar panel installation work according to the roof type of a specified building.

[0090] FIG. 14 is a diagram showing an example of a presentation screen of building attribute information on the terminal 2. The presentation screen of building attribute information shown in FIG. 14 is a screen that presents buildings having attribute information that satisfies the specified search conditions within the specified search range. Therefore, the presentation screen of building attribute information shown in FIG. 14 includes a map of the search range in which the location of the building that satisfies the specified conditions is highlighted, and a display column for the search conditions and the attribute information of the building that satisfies the search conditions as the search result. In the example shown in FIG. 14, in the map, it is highlighted by pinning the building that satisfies the specified conditions.

[0091] The building attribute information presentation screen shown in FIG. 14 is displayed as a response to the request when, for example, the radio button of "range designation" is selected on the building attribute information input screen shown in FIG. 12. When the presentation control unit 15 of the server 1 receives a search request, a search range, and search conditions for attribute information from the terminal 2, it searches the building attribute information DB 16 for buildings located within the search range and whose attribute information satisfies the search conditions, reads out the attribute information of the corresponding one or more buildings from the building attribute information DB 16, and transmits it to the terminal 2. If the building attribute information DB 16 does not contain the attribute information of the buildings located within the search range, the presentation control unit 15 may request the attribute information acquisition unit 11 to acquire the attribute information of the buildings within the search range, extract the buildings that satisfy the search conditions from the acquired attribute information of the target buildings, and transmit the attribute information of the buildings that satisfy the search conditions within the search range to the terminal 2.

[0092] The method for presenting building attribute information as shown in FIG. 14 is effective, for example, when learning a roof image with an aging label in the roof type determination model 14 by including the roof type of "aging" in the types of roof types, specifying that the roof type in the search conditions is "aging", extracting the buildings with the roof type of "aging" as candidate buildings for redevelopment, and identifying the corporations that use or own the buildings. Also, for example, it is effective when extracting and presenting buildings with a desired store size from the building area for a company that newly wishes to open a store.

[0093] FIG. 15 is a diagram showing an example of a presentation screen of building attribute information on the terminal 2. The presentation screen of the building attribute information shown in FIG. 15 is a screen that presents the number or total floor area of buildings having attribute information that satisfies the specified search conditions within the specified search range. Therefore, on the presentation screen of the building attribute information shown in FIG. 15, there is a map of the search range in which buildings that satisfy the specified conditions are color-coded and displayed, the number and total floor area of buildings that satisfy the search conditions within the search range, and the number and total floor area of all buildings within the search range. Note that the ratio of the number or total floor area of buildings that satisfy the search conditions to the number of buildings or the total floor area of all buildings in the entire search range may be shown.

[0094] The presentation screen of the building attribute information shown in FIG. 15 may be displayed, for example, as one of the responses to the request when the radio button of "range designation" is selected on the input screen of the building attribute information shown in FIG. 12. When the presentation control unit 15 of the server 1 receives a search request, a search range, and search conditions for attribute information from the terminal 2, it acquires from the building attribute information DB 16 the number of buildings located within the search range and the total value of the areas of all buildings, searches for buildings located within the search range and whose attribute information satisfies the search conditions, acquires the number of the corresponding one or more buildings and the total of the areas of the one or more buildings, and transmits them to the terminal 2. Note that when the building attribute information DB 16 does not include the attribute information of buildings located within the search range, the presentation control unit 15 may request the attribute information acquisition unit 11 to acquire the attribute information of the buildings within the search range, update the building attribute information DB 16 with the acquired attribute information of the target buildings, and then perform the above processing to transmit the search results to the terminal 2.

[0095] <Operation and Effect of the First Embodiment> In the first embodiment, the server 1 analyzes map information published on the Internet and acquires the building area and roof type, which are attribute information not explicitly shown in the map information. In this case, since no other databases are referred to, for example, there is no need to execute processing related to verification of consistency between different databases, and it becomes possible to quickly present the attribute information of the target building. Also, since the attribute information of each building can be quickly acquired, a digital twin can be quickly constructed based on the attribute information of the building.

[0096] Also, in order to acquire the area of the target building, the server 1 performs simple processing such as color determination, color conversion, and binarization of the map image to generate a mask image of the target building. From the target mask image, the area of the target building is acquired through a process of acquiring the number of white bits corresponding to the target building and acquiring the area per bit according to the zoom level of the map image. Also, the roof type is determined based on the output obtained as a result of inputting the roof image obtained from the aerial photo image using the mask image into the roof type determination model 14. Since neither the generation of the mask image nor the acquisition of the roof image is complex processing, it is possible to quickly present multiple types of attribute information such as the building area and roof type.

[0097] <Recording Medium> A program for causing a computer or other machine or device (hereinafter referred to as a computer or the like) to realize any of the above functions can be recorded on a computer-readable recording medium. By causing the computer or the like to read and execute the program of this recording medium, the function can be provided.

[0098] Here, a computer-readable recording medium refers to a non-transitory recording medium that accumulates information such as data and programs by means of electrical, magnetic, optical, mechanical, or chemical actions and can be read by a computer or the like. Among such recording media, removable ones from a computer or the like include, for example, flexible disks, magneto-optical disks, CD-ROMs, CD-R / Ws, DVDs, Blu-ray disks, DATs, 8mm tapes, memory cards such as flash memories, and the like. Also, as recording media fixed to a computer or the like, there are hard disks, ROMs (read-only memories), and the like. Furthermore, an SSD (Solid State Drive) can be used as both a removable recording medium from a computer or the like and a recording medium fixed to a computer or the like. It can also be used as a recording medium fixed to a computer or the like.

Explanation of Signs

[0099] 1 ··· Server 2 ··· Terminal 3 ··· Map Information Providing Server 11 ··· Attribute Information Acquisition Unit 12 ··· Map Information Acquisition Unit 13 ··· Image Analysis Unit 14 ··· Roof Type Judgment Model 15 ··· Presentation Control Unit 16 ··· Building Attribute Information DB 31 ··· Map Information DB 100 ··· Building Attribute Information Providing System 101 ··· CPU 102 ··· Memory 103 ··· Auxiliary Storage Device 104 ··· Communication Unit

Claims

1. A computer, obtains map information within a first range from map information, and obtains, based on the map information within the first range, a plurality of types of attribute information of a first building included within the first range that is not explicitly shown in the map information. A method for performing the above.

2. The map information includes at least a first map image in which the areas of buildings and other areas are colored differently. The computer obtains the area of the first building as one of the attribute information of the first building based on the first map image within the first range. The method according to claim 1.

3. The computer classifies the first map image within the first range into the area of the first building and other areas and binarizes it, and obtains the area of the first building from the size of the area of the first building. The method according to claim 2.

4. The map information includes an aerial photo image. The computer obtains the roof type of the first building as one of the attribute information of the first building based on the aerial photo image within the first range. The method according to claim 1.

5. The map information further includes an aerial photo image. The computer obtains the roof type of the first building as one of the attribute information of the first building based on the aerial photo image within the first range. The method according to claim 2.

6. The computer obtains a first aerial photo image of the first building cut out from the aerial photo image within the first range based on the first map image within the first range, and determines the roof type of the first building by performing image analysis on the first aerial photo image of the first building. The method according to claim 5.

7. The computer determines the roof type of the first building based on the output result obtained by inputting the first aerial photo image of the first building into a machine learning model that has learned the relationship between the aerial photo image of the building and the roof type. The method according to claim 6.

8. The machine learning model is a convformer. The method according to claim 7.

9. The computer classifies the first map image within the first range into the area of the first building and other areas and generates a binarized mask image. Superimpose the mask image on the aerial photo image within the first range, cut out the area of the aerial photo within the first range corresponding to the area of the first building in the mask image, and obtain the aerial photo image of the first building. The method according to claim 6.

10. The computer further obtains a first map image and an aerial photo image within a second range including the first building with a different zoom level from the first range; obtains the area of the first building based on either the first map image within the first range or the map image within the second range; obtains the roof type of the first building based on either the aerial photo image within the first range or the aerial photo image within the second range. The method according to claim 4.

11. The second range is a range when the zoom level is higher than that of the first range. The computer obtains the area of the building based on the first map image within the first range; when the ratio of the area of the building to the area of the first range is less than a predetermined value, obtains the roof type of the building based on the aerial photo image within the second range. The method according to claim 10.

12. The computer further obtains, as one of the attribute information of the first building, corporate information using the first building. The method according to claim 1.

13. The computer further executes associating the first building with the obtained plurality of types of attribute information and storing them in a storage unit. The method according to claim 1.

14. The computer receives a request for attribute information of the first building from a terminal; when the request is received, obtains the attribute information of the first building from the storage unit; transmits the attribute information of the first building to the terminal. and further executes. The method according to claim 13.

15. The computer receives a request for attribute information of the first building from a terminal; when the request is received, executes obtaining the map information within the first range from the map information and obtaining a plurality of types of attribute information of the first building based on the map information within the first range; transmits the attribute information of the first building to the terminal. and further executes. The method according to claim 1.

16. The computer Receiving, from a terminal, a specification of a third range and a first condition including a condition related to at least one of a roof type or an area Identifying, among a plurality of first buildings included in the third range, a first building that satisfies the first condition Transmitting, to the terminal, information regarding the first building that satisfies the first condition Further executing The method according to claim 1.

17. The computer Receiving, from a terminal, a specification of a third range and a first condition including a condition related to at least one of a roof type or an area Obtaining a total value of the number or area of the first buildings that satisfy the first condition among a plurality of first buildings included in the third range Transmitting, to the terminal, the total value of the number or area of the first buildings that satisfy the first condition Further executing The method according to claim 1.

18. The map information is publicly available on the Internet The method according to claim 1.

19. Obtaining map information within a first range from map information Obtaining a plurality of types of attribute information not explicitly shown in the map information of the first buildings included within the first range based on the map information within the first range A control unit that executes An information processing apparatus comprising

20. Causing a computer To obtain map information within a first range from map information To obtain a plurality of types of attribute information not explicitly shown in the map information of the first buildings included within the first range based on the map information within the first range A program for causing execution

Images