JP2024073172A - Accounting risk evaluation support system, accounting risk evaluation support method and accounting risk evaluation support program - Google Patents

Abstract

To provide an accounting risk evaluation support system enabled to evaluate the relationship between geographical information on a company and information associated with finance without visiting the site.SOLUTION: The present invention relates to an accounting risk evaluation support system 1 for evaluating the relationship between geographical information on a company and information associated with the finance. The accounting risk evaluation support system comprises: an image acquisition part 31 which acquires a photographic image including position information specified as the geographical information, an image extraction part 32 which extracts a corresponding object image from the photographic image based upon the position information, a real estate index estimation part 33 which recognizes the corresponding object image to estimate a real estate index associated with the geographical information, and a relationship output part 34 which outputs the relationship between the real estate index and a financial index as information associated with the finance.SELECTED DRAWING: Figure 1

Description

The present invention relates to an accounting risk assessment support system, an accounting risk assessment support method, and an accounting risk assessment support program for evaluating the relationship between a company's geographical information and financial information.

In accounting audit work such as internal audits and external audits, the contents of accounting documents such as financial statements are analyzed and checked to see if there are any fraudulent transactions, such as fictitious transactions (see Patent Documents 1 and 2, etc.).

On the other hand, the importance of the three actual things principle, which means actually going to the site, checking the actual things, and understanding the reality, is also emphasized in audit work. For example, the three actual things principle is used to check whether geographical information such as the location of a company's building listed on accounting documents is correct when compared to financial information such as sales.

Patent No. 7052135 JP 2019-179531 A

However, actually going to the site to check not only requires travel time and money, but also depends on the schedule of the company being audited and the person in charge, so it may not be easy to do. Also, in order to effectively implement the three actual places principle, it is advisable to get a feel for fraudulent transactions in advance.

The present invention aims to provide an accounting risk assessment support system, an accounting risk assessment support method, and an accounting risk assessment support program that enable the assessment of the relationship between a company's geographical information and financial information without the need to go to the site.

To achieve the above-mentioned object, the accounting risk assessment support system of the present invention is an accounting risk assessment support system for evaluating the relationship between a company's geographical information and financial information, and is characterized by comprising an image acquisition unit that acquires a photographic image including location information specified as the geographical information, an image extraction unit that extracts a relevant object image from the photographic image based on the location information, a real estate index estimation unit that estimates a real estate index related to the geographical information by performing image recognition on the relevant object image, and a relationship output unit that outputs the relationship between the real estate index and a financial index that is information related to the finances.

The invention of the accounting risk assessment support method is an accounting risk assessment support method for evaluating the relationship between a company's geographical information and financial information, and is characterized by comprising the steps of: acquiring a photographic image including location information designated as the geographical information; extracting a relevant object image from the photographic image based on the location information; estimating a real estate index related to the geographical information from the relevant object image; and outputting the relationship between the real estate index and a financial index that is information related to the finances.

Furthermore, the invention of the accounting risk assessment support program is an accounting risk assessment support program for evaluating the relationship between a company's geographical information and financial information, characterized in that the program causes a computer to execute the steps of acquiring a photographic image including location information designated as the geographical information, extracting a relevant object image from the photographic image based on the location information, estimating a real estate index related to the geographical information by performing image recognition on the relevant object image, and outputting the relationship between the real estate index and financial indexes that are information related to the finances.

Another accounting risk assessment support system invention is an accounting risk assessment support system for evaluating the relationship between a company's geographical information and financial information, and is characterized by comprising an image acquisition unit that acquires a map image including location information specified as the geographical information, an image extraction unit that extracts a relevant object image from the map image based on the location information, a real estate index estimation unit that calculates the building area of a building identified in the relevant object image and the number of buildings in the surrounding area of the relevant object image as real estate indexes, and a relationship output unit that outputs the relationship between the real estate indexes and financial indexes that are information related to the finances.

The accounting risk assessment support system, accounting risk assessment support method, and accounting risk assessment support program of the present invention thus configured acquire a photographic image containing location information specified as geographical information, and estimate real estate indicators, such as the type of building that exists there, from the relevant object image extracted from the photographic image. Then, the relationship between the real estate indicators and financial indicators such as sales is output.

This makes it possible to efficiently evaluate the appropriateness of a company's geographical information compared to sales, etc., without having to visit the company's location or holdings.

1 is a block diagram illustrating the configuration of an accounting risk assessment support system according to an embodiment of the present invention. 1 is a flowchart illustrating the process flow of an accounting risk assessment support method according to an embodiment of the present invention. 1A and 1B are diagrams illustrating a map image, in which FIG. 1A is an explanatory diagram of an example of an acquired map image, and FIG. 1B is an explanatory diagram of an example of a binarized map image. 13A and 13B are diagrams illustrating a process for identifying a relevant building from a map image, in which FIG. 13A is an explanatory diagram of a process for identifying a relevant building based on distance, and FIG. 13B is an explanatory diagram of an example of an identified relevant building. 1A and 1B are diagrams illustrating photographic images, in which FIG. 1A is an explanatory diagram of an example of an acquired aerial image, and FIG. 1B is an explanatory diagram of an example in which a corresponding object image has been identified. 11 is an explanatory diagram illustrating an example of classification results of image recognition and corresponding flags. FIG. 13 is an example of a method for calculating the small building degree, where (a) is an explanatory diagram of an example in which the small building degree is calculated based on the ratio of detected objects to non-detected objects, and (b) is an explanatory diagram of an example in which the small building degree is weighted. FIG. 13 is an explanatory diagram showing an example of an output of the relationship between sales and small building index. 11 is an explanatory diagram of an example of estimating a real estate index using the similarity rank of a corresponding object image. FIG. 1 is a flowchart illustrating the processing flow of the accounting risk assessment support system of Example 1. 11A and 11B are diagrams illustrating an image of a surrounding area, in which FIG. 11A is an explanatory diagram showing an example of calculation of the number of buildings, and FIG. 11B is an explanatory diagram showing an example of a corresponding building. 13 is an explanatory diagram illustrating an example of the relationship between the number of buildings in the surrounding area, the building area of the relevant building, and a flag. 13 is an explanatory diagram illustrating an example of a relationship matrix for determining whether a fraudulent transaction has occurred based on the relationship between the number of buildings in the surrounding area and the building area of the relevant building. FIG.

The following describes an embodiment of the present invention with reference to the drawings. FIG. 1 is a block diagram illustrating the configuration of an accounting risk assessment support system 1 according to this embodiment. FIG. 2 is a flowchart illustrating the process flow of the accounting risk assessment support method according to this embodiment.

The accounting risk assessment support system 1 of this embodiment is a system for evaluating the relationship between geographical information and financial information of a company that is the subject of an accounting audit, such as an internal audit or an external audit.

Geographical information here refers to information about the company's location, the type of buildings on its land, and things other than buildings, such as parking lots, vacant lots, and forests. Financial information, on the other hand, refers to financial indicators such as sales, total capital, and assets listed in the accounting documents of the company being audited.

In the accounting risk assessment support system 1 of this embodiment, as shown in FIG. 1, the calculation processing unit 3 processes data input from an input device 2 or the like, and the judgment results are displayed on a display device 5 or the like.

The calculation processing unit 3 of the accounting risk assessment support system 1 is mainly composed of an image acquisition unit 31 that acquires photographic images such as aerial photographs and map images, an image extraction unit 32 that extracts target object images from the photographic images, a real estate index estimation unit 33 that estimates the real estate index of the object image, and a relationship output unit 34 that outputs the relationship between the real estate index and the financial index.

The input device 2 for inputting various data is a data input means connected to or equipped on a personal computer (PC), a notebook computer, a tablet terminal, a wearable terminal, a smartphone, etc. Examples of the input device 2 include a keyboard, a mouse, a touch panel, a touch pad, a scanner, and a microphone for voice input. A camera can also be used as the input device 2.

On the other hand, the display device 5 can be a liquid crystal display, an organic EL (Electro-Luminescence) display, a printer, etc.

The storage unit 4 is a storage medium that records data generated by processing in the arithmetic processing unit 3 and data required for arithmetic processing, and includes a hard disk, a solid-state drive (SSD), a flash memory (such as an SD memory card), a magnetic disk, and an optical disk. In addition, an external online storage (cloud storage) such as a server connected via a network can also be used as the storage unit 4.

The hardware of the calculation processing unit 3 is composed of a CPU (Central Processing Unit), an MPU (Micro-processing unit), a GPU (Graphics Processing Unit), etc., and is equipped with memories such as RAM (Random Access Memory) and ROM (Read Only Memory).

The arithmetic processing unit 3 can execute various functions by applications installed on the computer. In addition, part or all of the arithmetic processing unit 3 can be executed by a server connected via a network such as the Internet. The network is composed of part or all of the Internet, a Wide Area Network (WAN), a wired Local Area Network (LAN), a wireless LAN (Wi-Fi), a provider device, a wireless base station, a dedicated line, etc.

The image acquisition unit 31 of the calculation processing unit 3 acquires an image including location information specified as geographical information. Here, "location information specified as geographical information" refers to addresses and latitude and longitude information such as the company's location and ownership. Addresses of company branches, sales offices, factories, subsidiaries, etc. can also be used as location information.

Images acquired based on location information include, for example, map information such as a map image. Fig. 3 is a diagram for explaining a map image acquired by the image acquisition unit 31, and Fig. 3(a) shows an example of an acquired map image. Fig. 3(a) uses an electronic topographical map obtained by inputting latitude and longitude information into the data source of the Digital National Land Base Map of the Geospatial Information Authority of Japan's "Geographical Survey Institute Tile (registered trademark), https://maps.gsi.go.jp/development/ichiran.html". When acquiring a map image, information regarding the time of creation, such as the creation date, is also acquired.

Figure 3(b) shows a map image obtained by binarizing the Digital Topographic Map of Figure 3(a). Binarization is a process performed to improve the accuracy of detecting buildings by distinguishing between buildings and non-buildings on the Digital Topographic Map. Here, binarization was performed to paint areas other than the gray color (RGB(205,205,205) - RGB(215,215,215)) that indicates buildings black.

The images acquired by the image acquisition unit 31 based on the location information also include photographic images such as aerial images (see FIG. 5(a)). Photographic images are images such as photographs taken from the sky of the location corresponding to the location information, and include aerial photographs, photographs taken by drones, and satellite images. Note that FIG. 5(a) uses aerial photographs obtained by inputting latitude and longitude information into the data source of the latest national photographs from the Geospatial Information Authority of Japan's Geospatial Information Authority of Japan tiles. When acquiring photographic images, information regarding the time of shooting, such as the shooting date, is also acquired.

The image extraction unit 32 extracts relevant object images from the photographic images acquired by the image acquisition unit 31 based on the location information. In other words, the photographic images acquired by the image acquisition unit 31 include not only the building at the specified location information but also the surrounding area, so it is necessary to extract only images within the range in which image recognition is to be performed.

For example, Geospatial Information Authority tiles use a concept called "zoom level" to classify the map display magnification, and you can obtain rectangular image data at various scales depending on the specified zoom level.

For this reason, the image extraction unit 32 performs a process of extracting only images of buildings related to the location information of the company being audited as relevant object images from the acquired photographic images. As will be described later, the extracted relevant object images will be used to determine the type of building present at the location specified by the location information, whether it is a vacant lot or other type of object, etc., but for ease of understanding in the following explanation, it will basically be assumed that the relevant object images show buildings.

The process of extracting relevant object images from photographic images can be performed by directly performing image recognition on the photographic image, but the image extraction unit 32 of this embodiment uses a map image to extract relevant object images.

As described above, the map image used is binarized to make buildings easier to distinguish (see Figure 3(b)). Data on the contours of the white (gray) or black parts in the binarized image can be obtained by an object detection device created using known programming (such as the findContours function in the OpenCV library). Then, from this binarized map image, the building associated with the location information is identified as the relevant building.

Figure 4 is a diagram explaining the process of identifying the relevant building MB from the binary image MA, which is a map image that has been binarized. Here, the relevant building MB is identified from the acquired binary image MA based on the Euclidean distance, etc.

FIG. 4A is a diagram for explaining the process of identifying a relevant building MB based on distance.
For the buildings present in the binary image MA, the centers of the buildings are found for all the detected buildings. In Fig. 4(a), for ease of viewing, only the building centers B1 and B2 of two buildings are shown with circles.

On the other hand, the rectangular binary image MA has a map center M1 (star mark). This map center M1 is a position that represents the input location information, and the building closest to this map center M1 can be said to be the building of the company that is being audited.

The straight-line distance (Euclidean distance) between the building centers (B1, B2, ...) of the buildings detected in the binary image MA and the map center M1 is calculated, and the building with the closest building center B1 is identified as the relevant building MB. In Figure 4 (b), the identified relevant building MB is shown surrounded by a rectangular area.

In this way, the area of the relevant building MB identified based on the map image is superimposed on the photographic image, thereby extracting a relevant object image corresponding to the relevant building MB from the photographic image. Figure 5 is an explanatory diagram illustrating an aerial image PA as an example of a photographic image used for explanation.

As with map images, by inputting location information such as latitude and longitude of the company to be audited, an aerial image PA including the relevant building MB, as shown in Figure 5(a), can be obtained.

When the map image and the aerial image PA are superimposed, the magnification is adjusted and distortion is corrected as necessary. Then, on the aerial image PA, a building that appears near the area of the relevant building MB identified in the binary image MA is identified as the relevant object image PB. Figure 5 (b) shows the state in which the relevant object image PB has been identified in the aerial image PA.

The size of the relevant object image PB does not need to be the same as the area of the relevant building MB identified in the binary image MA, and the area of the relevant object image PB is set to a size that can contain the entire building shown in the aerial image PA. Note that if the relevant object image PB is something other than a building, such as a vacant lot, it may remain the same size as the area of the relevant building MB.

The image of the building cut out by the relevant object image PB is used to estimate the type of building through image recognition. In other words, the real estate index estimation unit 33 estimates the type of building shown in the relevant object image PB.

The real estate index estimation unit 33 performs machine learning using training data that contains a set of buildings shown in a photographic image and a flag indicating their type, and can use this trained model to estimate the type of building shown in the relevant object image PB. For example, by extracting images of various types of buildings, forests, parking lots, etc. from an aerial image and then overlaying machine learning on them, a trained model with high recognition accuracy can be created.

It is also possible to use a known image recognition tool (such as the Places dataset (http://places2.csail.mit.edu/)) to estimate the type of building shown in the relevant object image PB.

Figure 6 is an explanatory diagram illustrating the classification results of image recognition and the corresponding flags. For example, names such as a house, forest, vacant lot, construction site, park, hospital, office building, city hall, airport, fire station, police station, research institute, factory, etc. can be assigned to the corresponding object image PB as the classification result.

When evaluating the relationship with financial information, there are things that should be detected and things that can be excluded from detection. For example, the type of building, such as whether it is a small single-family home, or information such as whether it is a vacant lot, can be important information in detecting fraudulent transactions such as fictitious transactions. In contrast, distinguishing whether a building is a hospital or a city hall is not necessary when estimating real estate indicators. Therefore, a flag such as "detection target" or "not detection target" is attached to the classification result name.

In this embodiment, the real estate index estimation unit 33 applies a real estate index as an index related to geographical information. A real estate index is an index that indicates the evaluation of buildings and the like that appear in the relevant object image PB. For example, the evaluation of a building is expressed as a building degree, and an index called "small building degree" is applied as an index for evaluating the smallness of a building.

Figure 7 is a diagram for explaining the method of calculating the small building degree. In the calculation method shown in Figure 7 (a), the corresponding object image PB is subjected to image recognition, and the top five classifications with the highest similarity ranks are listed in a table.

Among the categories that are listed as possibly corresponding to the relevant object image PB, only the single-family home has a flag of "detection target," and the probability of this being the case is 7.21%. Therefore, the ratio of the total probability of the flag being "detection target" to the total probability of the top five ranks is calculated as the small building probability (building probability).
7.21/(7.21+5.30+4.43+4.21+3.95)=28.73%
This small building degree can be said to be a numerical representation of the possibility that the corresponding object image PB is a small building.

7B shows an example of a calculation method for weighting the probability of occurrence. The weighting is performed based on the similarity rank, and the weighted probability is calculated by dividing each probability of occurrence by the similarity rank.
7.21/1=7.21, 5.30/2=2.65, 4.43/3=1.48, 4.21/4=1.05, 3.95/5=0.79
(7.21+1.48)/(7.21+2.65+1.48+1.05+0.79)=65.91

The small building index, which indicates the real estate index calculated in this manner, is output by the relationship output unit 34 so that the relationship with financial indicators such as sales can be understood. Figure 8 is an explanatory diagram showing an example of the relationship between sales and the small building index.

When the small building degree is high, indicating that the building is likely to be a small building such as a detached house, the sales revenue of the company being audited is generally low. However, if the small building degree value calculated from the relevant object image PB extracted based on the company's location information is high, that is, if the recorded sales revenue is high even though a detached house stands at the company's location, some kind of fraudulent transaction, such as a fictitious transaction, is suspected.

Therefore, as shown in Figure 8, by plotting the small building index calculated from the relevant object image PB on a graph showing the relationship between the two, with sales on the horizontal axis and the small building index on the vertical axis, it becomes possible to determine the possibility of fraudulent transactions.

For example, if the data is plotted within the area enclosed by the dashed line in Figure 8, it can be determined that there is a high possibility of fraudulent transactions. If the output indicates that there is a high possibility of fraudulent transactions, the current situation can be confirmed using the latest satellite images, etc.

Such fraudulent transactions can also be determined by other methods. Figure 9 is an explanatory diagram of an example of estimating a real estate index using the similarity rank of the relevant object image PB. As described above, when the relevant object image PB is subjected to image recognition, a similarity rank is assigned in order of the classification most likely to match.

If the top similarity rankings include even one classification that may be a fictitious transaction, the transaction must be checked separately. For example, if the company's location is shown as possibly being a "vacant lot," it can be determined that there is a possibility of a fictitious transaction.

In addition, if an apartment is recorded as a real estate asset but does not appear in the top similarity rankings (e.g., in the top 5) of the results of image recognition of the relevant object image PB, it is possible that something other than a real estate asset is being falsely represented, and therefore the transaction must be checked separately.

In this way, even if a quantified real estate index such as the small building index is not calculated, a real estate index called a similarity rank can be output to determine whether or not a transaction should be rechecked based on its relationship with financial indexes.

The process flow of the accounting risk assessment support method of this embodiment will be described below with reference to the flowchart shown in FIG.
In the following, an example will be described in which the accounting risk assessment support program constituting the above-mentioned accounting risk assessment support system 1 is installed on a computer.

In other words, the installed program (application) causes the computer to execute the steps of acquiring a photographic image containing specified location information, extracting a corresponding object image from the photographic image based on the location information, estimating a real estate index by performing image recognition on the corresponding object image, and outputting the relationship between the real estate index and financial index.

First, in step S1, the location information of the company to be audited is entered according to the input screen displayed on the display device 5 connected to the computer. The location information entered may include the address and latitude and longitude information.

When the location information is input, map images and photo images of the corresponding area can be obtained from the Geospatial Information Authority of Japan's "Geospatial Information Authority Tiles" (step S2). Then, in step S3, the obtained map image is subjected to a binarization process in which all colors except those indicating buildings (gray) are replaced with black (see Figure 3(b)).

Next, in step S4, buildings are detected in the binarized image MA (Figure 4(a)), which is the map image that has been binarized, and the centers of the detected buildings are calculated as building centers (B1, B2, ...). Then, the building with the building center B1 that is closest to the building center (B1, B2, ...) and the map center M1 of the binarized image MA is detected as the corresponding building MB.

The area of the detected building MB is overlaid on the aerial image PA, which is the photographic image acquired in step S2, and after adjustment is made so that the entire building shown in the aerial image PA is included at the position of the building MB, as shown in Figure 5 (b), it is extracted as the relevant object image PB (step S5).

In step S6, the extracted matching object image PB is subjected to image recognition, and the probability of matching the top five similarity ranks is calculated, for example as shown in FIG. 7(a). A flag is attached to the name of the classification result of each similarity rank to indicate whether it is a detection target or not, and based on the flag, the small building degree, which serves as a real estate index, is calculated by dividing (total probability of detection target) by (total probability of similarity rank 1 to similarity rank 5).

In the next step S7, a graph showing the relationship as shown in FIG. 8 is plotted based on the small building index calculated in step S6 and the sales (financial indicator) of the company being audited. If the plotting results in a range where the sales are high but the small building index is also high, it is determined that there is a possibility of fraudulent transactions.

Next, the operations of the accounting risk assessment support system 1, the accounting risk assessment support method, and the accounting risk assessment support program of this embodiment will be described.
The accounting risk assessment support system 1, accounting risk assessment support method, and accounting risk assessment support program of this embodiment configured as described above acquire an aerial image PA including location information designated as geographical information such as the location of a company, estimate the type of building from the relevant object image PB extracted from the aerial image PA, calculate real estate indices such as a small building degree, and output the relationship between the real estate indices and financial indices such as sales.

This makes it possible to evaluate relationships such as whether a company's geographical information is appropriate compared to sales, and whether there is the possibility of fraudulent transactions such as fictitious transactions, without having to visit the company's location or holdings.

Below, an accounting risk assessment support system of another embodiment different from the accounting risk assessment support system 1 of the embodiment described above will be described with reference to Figures 10 to 13. Note that the same terms or the same reference numerals will be used to describe the same or equivalent parts as those described in the embodiment described above.

In the accounting risk assessment support system 1 described in the above embodiment, real estate indexes are estimated by performing image recognition on the relevant object image PB extracted from the photographic image, but in the accounting risk assessment support system of Example 1, real estate indexes are estimated only from the map image.

In other words, the accounting risk assessment support system of Example 1 will be described with reference to FIG. 1. The calculation processing unit 3 is mainly composed of an image acquisition unit 31 that acquires a map image, an image extraction unit 32 that extracts target object images from the map image, a real estate index estimation unit 33 that estimates real estate indices of the object images, and a relationship output unit 34 that outputs the relationship between the real estate index and the financial index.

The real estate index estimation unit 33 of the first embodiment calculates the building area of the building identified in the relevant object image and the number of buildings in the surrounding area of the relevant object image as a real estate index. FIG. 10 is a flowchart explaining the process of calculating the real estate index and subsequent steps in the process flow of the accounting risk assessment support system of the first embodiment. The process from input of the location information to binarization of the map image is the same as steps S1 to S3 of FIG. 1.

Here, FIG. 11 is a diagram explaining the surrounding area image TA, which is a binary image of the surrounding area set based on the location information. For example, an area of about 300 m vertical x 400 m horizontal with the location information at the center of the map image can be set as the surrounding area.

As described above, buildings can be detected with high accuracy from the surrounding area image TA, in which everything other than buildings is painted black and binarized. Therefore, in step S11, the number of buildings BC detected in the surrounding area image TA (number of buildings) is calculated, as shown in FIG. 11(a).

Next, in step S12, the building closest to the map center of the surrounding area image TA is identified as the relevant building MB (Figure 11 (b)). For the building identified as the relevant building MB, the number of pixels in the part of the surrounding area image TA that corresponds to the building (gray part) is calculated, and the building area is calculated based on this number of pixels.

Figure 12 is an explanatory diagram showing an example of calculating the number of buildings in the surrounding area image TA and the building area of the corresponding building MB. As shown in this table, even if the range of the surrounding area is constant (in this example, approximately 300m long x 400m wide), the relationship between the number of buildings and the building area can vary. Here, we will focus on cases where both the number of buildings and the building area are small, and assign a flag of "1".

On the other hand, Figure 13 shows an example of a relationship matrix that can be used to determine fraudulent transactions based on the relationship between the number of buildings in the surrounding area and the floor area of the relevant building. This relationship matrix changes depending on the scale of the financial indicators being audited, location information, type of building, etc., so it is merely an example.

Figure 13 shows the relationship that for a certain sales volume (financial indicator), if the building area is large, the risk of fraudulent transactions is small, but if the building area is medium-sized (about the size of a few detached houses) or small (the size of a detached house), the risk of fraudulent transactions increases the fewer the number of buildings. For example, if the building MB in question is a single-family home in the middle of a forest, there will be few buildings in the surrounding area and the building area will also be small. If a company located in such a place has high sales volume, the existence of fraudulent transactions such as fictitious transactions will be suspected.

Therefore, in step S13, the number of buildings and the building area calculated in steps S11 and S12 are applied to a relationship matrix such as that shown in FIG. 13. Then, in step S14, it is determined from the results whether or not there is a possibility of fraudulent transactions.

The accounting risk assessment support system, accounting risk assessment support method, and accounting risk assessment support program of Example 1 configured in this manner can output the relationship between real estate indicators and financial indicators and determine the possibility of fraudulent transactions, etc., without performing image recognition based on photographic images.

The rest of the configuration and effects are similar to those of the previous embodiment, so a detailed description will be omitted.

The above describes the embodiments of the present invention in detail with reference to the drawings, but the specific configuration is not limited to these embodiments and examples, and design changes that do not deviate from the gist of the present invention are included in the present invention.

For example, in the above embodiment and Example 1, the case where a map image is binarized is described as an example, but the present invention is not limited to this. A binarization process is performed on a photographic image to separate buildings and non-buildings, and the above-mentioned process can be applied to the binary image (FIG. 4(a), FIG. 11(a)) thus generated.

In addition, in the above embodiment, an example was described in which a relevant building MB is identified from a map image, and a relevant object image PB is extracted from a photographic image based on that, but the present invention is not limited to this. It is also possible to perform machine learning on a photographic image to distinguish between buildings and non-buildings, and then extract the building closest to the center of the acquired photographic image as the relevant building.

1: Accounting risk assessment support system 31: Image acquisition unit 32: Image extraction unit 33: Real estate index estimation unit 34: Relationship output unit PA: Aerial image (photo image)
PB: Matching image MA: Binarized image (map image)
MB: Building in question TA: Surrounding area image

Claims (8)

An accounting risk assessment support system for evaluating the relationship between geographical information and financial information of a company, comprising:
an image acquisition unit that acquires a photographic image including location information designated as the geographical information;
an image extraction unit that extracts a corresponding image from the photographic image based on the position information;
a real estate index estimation unit that estimates a real estate index related to the geographical information by performing image recognition on the relevant object image;
An accounting risk assessment support system comprising a relationship output unit that outputs the relationship between the real estate index and a financial index that is information related to finances. The accounting risk assessment support system according to claim 1, characterized in that the real estate index is a building degree, which is an evaluation of the size of a building. The image extraction unit uses a map image when extracting the relevant object image,
3. The accounting risk assessment support system according to claim 1, wherein the map image is one in which buildings are distinguished from other areas by binarization. The accounting risk assessment support system according to claim 3, characterized in that buildings are detected from the binarized map image, the building closest to the location information among the detected buildings is detected as the relevant building, and the relevant object image is extracted based on the area in which the relevant building is included. The accounting risk assessment support system according to claim 1 or 2, characterized in that the real estate index estimation unit performs image recognition using a machine-learned model. 1. A method for supporting accounting risk assessment for evaluating a relationship between geographical information and financial information of a company, comprising:
acquiring a photographic image including location information designated as the geographical information;
extracting a relevant image from the photographic image based on the position information;
estimating real estate metrics related to the geographic information from the relevant images;
An accounting risk assessment support method comprising a step of outputting a relationship between the real estate index and a financial index which is information related to finances. An accounting risk assessment support program for evaluating the relationship between a company's geographical information and financial information,
obtaining a photographic image including location information designated as the geographical information;
extracting a relevant image from the photographic image based on the position information;
A step of estimating a real estate index related to the geographical information by performing image recognition on the relevant object image;
and outputting the relationship between the real estate index and a financial index that is information related to the finances. An accounting risk assessment support system for evaluating the relationship between geographical information and financial information of a company, comprising:
an image acquisition unit that acquires a map image including location information designated as the geographical information;
an image extraction unit that extracts a corresponding object image from the map image based on the position information;
a real estate index estimation unit that calculates a building area of a building specified in the relevant object image and a number of buildings in a surrounding area of the relevant object image as real estate indexes;
An accounting risk assessment support system comprising a relationship output unit that outputs the relationship between the real estate index and a financial index that is information related to finances.