Method for generating basic object information in bird's eye view

Abstract

The present invention relates to a method for generating basic object information reflected in a bird's eye view. According to a preferred embodiment of the present invention, the method for generating basic object information in a bird's eye view comprises the steps in which: a service device identifies an object in a to-be-read image and extracts individual object information about the identified object; and the service device generates basic object information in a bird's eye view by matching the extracted individual object information.

Description

How to generate basic object information from a bird's-eye view

[0001] The present invention relates to a method for generating basic object information reflected in a bird's-eye view.

[0002] Intersections and parking lots are monitored by CCTV. CCTVs capture real-time footage of the surveillance area. CCTVs are installed on walls or pillars and capture images diagonally across the surveillance area. To eliminate blind spots, multiple CCTVs are sometimes installed in the same surveillance area. These cameras can equally divide the surveillance area or overlap all or part of it.

[0003] The manager supervises control information based on video footage from the control area. The control information may include at least one of the following: vehicle compliance with traffic laws, human compliance with traffic laws, traffic volume, vehicle speed, vehicle location, vehicle direction, and vehicle movement status in the parking lot. The present invention does not impose any limitations on the control area. Control information may vary greatly depending on the control area.

[0004] Administrators monitor footage from multiple CCTV cameras using multiple monitors, each dedicated to a specific CCTV camera. This significantly reduces control efficiency and makes it difficult to detect specific events in real time. Of course, CCTV footage serves as valuable evidence for specific events.

[0005] If video footage of a controlled area is provided in a bird's-eye view format, the manager of that control area only needs to monitor that bird's-eye view. Therefore, providing bird's-eye view information can increase control efficiency. Furthermore, this bird's-eye view information can enable automatic, real-time monitoring. The bird's-eye view format is ideal for simulating actual control areas. Furthermore, CCTV footage matched to this bird's-eye view can serve as valuable evidence for specific events.

[0006] Accordingly, the present invention proposes a basic technology that can improve the efficiency of control.

[0007] The present invention proposes a method for generating object information that serves as the basis for forming a bird's-eye view used to identify control matters in a control area.

[0008] The present invention proposes a method for generating data that serves as the basis for forming bird's-eye views applicable to a variety of applications. The basic object information generated by the present invention can be applied to a variety of applications that utilize basic object information, beyond the purpose of control.

[0009] The present invention can also be used to generate object information in various application services based on user (object) location.

[0010] The present invention can also be used in fields of monitoring various objects for various purposes in addition to control purposes.

[0011] There may be no restrictions on the objects that are the target of generating basic object information of the present invention.

[0012] A method for generating basic object information in a bird's-eye view according to a preferred embodiment of the present invention includes a step in which a service device recognizes an object in a target image to be read and extracts individual object information for the recognized object; and a step in which the service device generates basic object information in the bird's-eye view by matching the extracted individual object information.

[0013] Here, the individual object information may include individual floor center point information, and the basic object information may include basic floor center point information.

[0014] In addition, the service device can calculate the base floor center point information as an average value of a plurality of individual floor center point information.

[0015] Additionally, the service device can extract individual object information by converting floor center point information using an algorithm that generates a bird's-eye view.

[0016] Additionally, the algorithm for generating the bird's-eye view may include Inverse Perspective Mapping (IPM).

[0017] Additionally, the floor center point information may be a point on the floor that can specify the location of an object in the object area.

[0018] Additionally, the above-mentioned target image for reading can be generated by capturing images taken by each of a plurality of image capturing devices.

[0019] Additionally, the plurality of video recording devices may be CCTVs.

[0020] Additionally, the plurality of video recording devices can capture images of an intersection.

[0021] Since the present invention recognizes the location of an object using the center point of the floor, the location of an object in a bird's-eye view can be more accurately determined.

[0022] In addition, since the present invention applies an algorithm that generates a bird's-eye view of the floor center point itself, distortion of the position of a tall object can be eliminated.

[0023] In addition, the present invention can make the position of an object more accurate by using the average value of the multiple bottom center points as the position of the object when there are multiple bottom center points for a single object.

[0024] The present invention can enable more reliable control by providing basic object information on a bird's-eye view.

[0025] The present invention can provide various control data services based on the location of highly reliable objects.

[0026] Figure 1 is a schematic diagram of a system to which a basic object information generation method according to a preferred embodiment of the present invention is applied.

[0027] Figure 2 shows a flow chart of a method for generating basic object information according to a preferred embodiment of the present invention.

[0028] Figure 3 is a detailed flow chart of the reading target image generation step of Figure 2.

[0029] Figure 4 is a detailed flow chart of the object recognition and individual object information extraction steps of Figure 2.

[0030] Figure 5 is a drawing for explaining the object recognition and individual object information extraction steps of Figure 4.

[0031] Figure 6 is a drawing for explaining the individual object information correction step of Figure 4.

[0032] Figure 7 is a detailed flow chart of the basic object information creation step.

[0033] Figure 8 is a drawing for explaining duplicate objects.

[0034] Figure 9 is a drawing for explaining the duplicate object determination step of Figure 7.

[0035] The present invention is susceptible to various modifications and embodiments. Specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, but rather to encompass all modifications, equivalents, and alternatives falling within the spirit and technical scope of the present invention.

[0036] In describing each drawing, similar reference numerals are used to designate similar components. In describing the present invention, detailed descriptions of related known technologies are omitted if they are deemed to obscure the gist of the present invention.

[0037] Although terms such as "first," "second," etc. may be used to describe various components, these components should not be limited by these terms. These terms are used solely to distinguish one component from another.

[0038] For example, without departing from the scope of the present invention, the first component could be referred to as the second component, and similarly, the second component could also be referred to as the first component.

[0039] The term and / or includes any combination of a plurality of related described items or any one of a plurality of related described items.

[0040] The terminology used in this application is for the purpose of describing specific embodiments only and is not intended to limit the present invention.

[0041] Singular expressions include plural expressions unless the context clearly dictates otherwise. In this application, terms such as "comprise" or "have" are intended to indicate the presence of a feature, number, step, operation, component, part, or combination thereof described in the specification, but should be understood not to preclude the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

[0042] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs.

[0043] Terms defined in commonly used dictionaries should be interpreted to have a meaning consistent with their meaning in the context of the relevant technology, and will not be interpreted in an idealized or overly formal sense unless expressly defined in this application.

[0044] 1. Glossary of Terms

[0045] * Floor center point: This can refer to the lower center point of the object under control. This lower center point can be the lower center of the front or rear of the object, or the lower center of the object itself. Depending on the designer's intention, the floor center point of the object can be selected at an appropriate location below the object. The floor center point can be used to determine the object's position in a bird's-eye view. There are no restrictions on the floor center point, as long as it represents the lower floor within the object's area. If the object is a vehicle, it can be a point on the road directly below the center of the front of the object (the front of the vehicle). Determining the floor center point relative to the road in this way can minimize distortion of the object's position when applying a bird's-eye view generation algorithm, such as the IPM transformation. In other words, the floor center point is preferably located directly below the front center of the object on the ground surface (e.g., a road surface or parking lot floor). The accuracy of the floor center point can be improved through AI learning from numerous images.

[0046] Alternatively, the floor center point may be the point where the object touches the floor. For reference, if the floor center point belonging to the floor is subjected to IPM transformation, distortion of the floor center point can be eliminated. The floor has characteristics that are clearly distinct from the object. Therefore, setting the floor center point as the point where the object touches the floor can significantly improve the recognition rate of the AI ​​floor center point. The term "floor center point" in the present invention may encompass all points on the floor that can identify the location of an object in the object area.

[0047] 2. Introduction

[0048] To create a bird's eye view, one might consider transforming closed-circuit television (CCTV) images into Inverse Perspective Mapping (IPM).

[0049] However, the IPM conversion assumes that all objects have a height of "0 (zero)." Therefore, when converting an image of a tall object, such as a car, to IPM, significant distortion occurs in the image.

[0050] For example, when creating a bird's-eye view of an intersection by combining multiple CCTV images installed at the intersection, there is a technical limitation that each IPM-converted CCTV image has significant distortion and cannot create a bird's-eye view for controlling the intersection.

[0051] However, a bird's-eye view, like one actually captured from above the control area, isn't necessarily required for control purposes. As long as control information can be grasped, the bird's-eye view can be in the form of a virtual space or a digital twin. Furthermore, if the bird's-eye view clearly depicts control items and real-time control targets (objects), the control objectives can be sufficiently achieved. Furthermore, even without generating a bird's-eye view based on the information available on the bird's-eye view, the control objectives can be sufficiently achieved by processing the information available on the bird's-eye view.

[0052] To this end, the present invention proposes a method for generating basic object information that can be used in a bird's-eye view. This basic object information can be used to represent a control target (or object) on a bird's-eye view or to identify control information. This basic object information can be used to identify control information regardless of the bird's-eye view. However, the basic object information generated using the method described below can significantly enhance compatibility when providing a bird's-eye view format.

[0053] The basic object information may include at least one of the object's location and the object's movement direction.

[0054] Below, a method for generating basic object information using images captured by a camera device is described in detail. To clarify the gist of the present invention, descriptions of previously known matters are omitted or simplified.

[0055] 3. System Overview

[0056] Referring to FIG. 1, a system to which a basic object creation method is applied may include an image capturing device and a service device.

[0057] Here, the video recording device (1) may be configured in multiple units. The video recording device may be, for example, a CCTV. The present invention is not limited to the video recording device and can be applied to any means capable of recording video.

[0058] The service device (2) may be configured as a server. The server may consist of a single server or multiple servers. Furthermore, the image analysis device may be configured as a dedicated server or a cloud computing environment.

[0059] The service device (2) can receive images captured by the image capture device (1). In this case, the images can be received in real time. The service device (2) can be equipped with an artificial intelligence model for performing the following method.

[0060] The service device (2) can generate basic object information according to the method described below.

[0061] 4. How to create basic object information

[0062] Referring to FIG. 2, a method for generating basic object information of an object may include a step of generating a reading target image (S100), a step of recognizing an object and extracting individual object information (S200), and a step of generating basic object information (S300).

[0063] (1) Step of creating an image to be read (S100)

[0064] Referring to FIG. 3, the service device (2) can receive images from N (N is a natural number greater than or equal to 2) image capturing devices (1). The images may be video streaming information. At this time, the service device (2) can capture images (S110). The service device (2) can capture images for each of the plurality of image capturing devices (1). Here, the N image capturing devices (1) may be devices that capture the same control area. The areas captured by the N image capturing devices may all or partly overlap. Alternatively, the areas captured by the N image capturing devices (1) may be captured by dividing the control area so that they do not overlap each other.

[0065] In addition, the service device (2) can correct the captured image to generate a target image to be read (S120). At this time, the service device (2) can correct the image distortion of the camera lens using previously stored calibration information.

[0066] S100 may be performed on each image received from multiple image capturing devices. For image alignment, S100 may be performed on images received at the same time from multiple image capturing devices.

[0067] (2) Object recognition and individual object information extraction step (S200)

[0068] The object recognition and individual object information extraction step (S200) can be performed for each target image to be read.

[0069] (A) Object recognition step (S210)

[0070] The service device (2) can recognize an object to be controlled (hereinafter referred to as “object to be controlled”) in the image to be read (S210). Here, the object to be controlled may be a control object in the control area. For example, the object to be controlled may be a person or a vehicle. For object recognition, an object recognition artificial intelligence model may be used. For example, the object recognition artificial intelligence model may be YOLO (You Only Look Once). Through the object recognition process, the service device (2) can extract an object identifier (ID), attribute information of the object (in other words, the type of object), and location information of the object area (coordinates of the object area in the image to be read, for example, a Bounding BOX).

[0071] (B) Object-specific mapping element extraction step (S220)

[0072] And, the service device (2) can extract mapping elements for each control target object (S220). At this time, the service device (2) can extract mapping elements for each object using a mapping element extraction artificial intelligence model. The mapping element extraction artificial intelligence model can receive a reading target image, an object identifier, object attribute information, and object area location information as input, and output at least one of floor center point information and direction of movement information for each object. The floor center point information may be two-dimensional coordinates (x, y). As described below, the floor center point information extracted in S220 can be converted into floor center point information on a bird's eye view through a correction and alignment process. In order to distinguish it from the floor center point hereinafter, the floor center point information extracted in S220 is referred to as “temporary floor center point information.”

[0073] The mapping element extraction AI model can be a model trained using the target image, object identifier, object attribute information, object area location information, floor center point information, and direction information. Fig. 5a is an example screen where the floor center point (P) has been extracted by the mapping element extraction AI model. In Fig. 5a, the object attribute (K) is a vehicle. In Fig. 5a, the letter “A” represents the object area location information.

[0074] (C) Individual object information extraction step (S230)

[0075] In addition, the service device (2) can extract individual object information (S230). The individual object information may be object information on a bird's-eye view. The individual object information may include at least one of the following: position information of the object on the bird's-eye view (in other words, information on the center point of the floor on the bird's-eye view), position information of the object area on the bird's-eye view, and information on the direction of movement of the object on the bird's-eye view.

[0076] In order to distinguish the position information of objects on the bird's-eye view extracted at other stages, the position information of object areas on the bird's-eye view, and the progress direction information of objects on the bird's-eye view, the position information of objects on the bird's-eye view extracted at S230, the position information of object areas on the bird's-eye view, and the progress direction information of objects on the bird's-eye view are respectively referred to as individual object position information, individual object area position information, and individual object progress direction information.

[0077] First, the process of extracting individual object location information (individual floor center point information) and individual object area location information is explained.

[0078] First, the service device (2) can extract individual floor center point information (in other words, individual object location information) and individual object area location information by converting the object area location information extracted from S210 and the floor center point information extracted from S220 using an algorithm for generating a bird's-eye view. The individual object area location information and the individual floor center point information may be in the form of two-dimensional coordinates.

[0079] For example, Inverse Perspective Mapping (IPM) and Perspective Transform can be used as algorithms for generating bird's-eye views. The IPM transform generates a bird's-eye view image of the target image, and the Perspective Transform can generate coordinates on the bird's-eye view image. The algorithm for generating bird's-eye views is a well-known technology, so a detailed description thereof will be omitted.

[0080] The present invention can minimize distortion of object positions by converting the coordinates of the floor center point itself using an algorithm that generates a bird's-eye view. The present invention can minimize distortion of object area position information by converting the object area position information itself using an algorithm that generates a bird's-eye view. Specifically, since the position of an object is identified based on the coordinates of the floor and the identified coordinates of the floor are converted into IPM, coordinate distortion due to height can be significantly eliminated during IPM conversion. Fig. 5b is an example screen showing the state of the floor center point (P) and the position information (A) of the object area after being converted into IPM. Fig. 5c is an example screen showing the state of the floor center point (P) and the position information (A) of the object area after being perspective converted.

[0081] S210 to S230 may be performed on each of the target images to be read from the images received from the plurality of image capturing devices.

[0082] However, when performing a conversion using an algorithm that generates a bird's-eye view (specifically, IPM), distortion of the floor center point and object area location information may occur, as shown in Figure 6. The degree of this distortion may vary depending on the angle and distance at which multiple video recording devices capture the image, and the height of the object being captured.

[0083] (D) Individual object information correction step (S240)

[0084] Accordingly, the present invention can perform correction on individual object information extracted in S230 (S240).

[0085] S240 can be performed on each image received from a plurality of image capturing devices. The directions in which the plurality of image capturing devices capture are different. And, the angles at which the plurality of image capturing devices capture may be the same or different. Considering this, the service device (2) can be equipped with a correction algorithm for each of the plurality of image capturing devices. And, the correction algorithm can perform correction on individual object information using a correction coefficient for each of the plurality of image capturing devices. Specifically, the correction algorithm can be an equation that includes a correction coefficient and uses individual object information as an independent variable. And, the correction coefficient can be a coefficient for correcting a two-dimensional coordinate. For example, when the coordinates of an individual floor center point extracted from an image captured from an image of a first image capturing device are (x, y), the service device (2) can correct the individual floor center point to (x-k1, y-k2). Here, k1 and k2 may be correction coefficients used to correct individual object information (particularly, floor center point) extracted from an image captured from an image of the first image capturing device. The correction coefficients may vary depending on the image capturing device. In Fig. 6, “A” is floor center point information before correction, “P” is object area location information before correction, “A'” is floor center point information after correction, and “P'” is object area location information after correction.

[0086] Through the above process, highly reliable individual object information can be extracted. S240 can be optionally performed.

[0087] Individual object information can be extracted or generated for each target image. Individual object information can be identified by the capture time of the image that forms the basis of the target image and the imaging device that captured the image.

[0088] (3) Basic object information creation step (S300)

[0089] By matching individual object information, basic object information can be generated from a bird's-eye view. To this end, individual object information for images captured from multiple imaging devices can be matched.

[0090] (A) Duplicate object determination step (S310)

[0091] First, the service device (2) can determine overlapping objects using individual object information (S310). Referring to FIG. 8, when multiple cameras capture the same control area, multiple cameras can capture the same object at the same time. In FIG. 8, A1 is object area location information among individual object information extracted from an image captured from an image of the first image capturing device, and P1 is floor center point information among individual object information extracted from an image captured from an image of the first image capturing device. In addition, in FIG. 8, A2 is object area location information among individual object information extracted from an image captured from an image of the second image capturing device, and P2 is floor center point information among individual object information extracted from an image captured from an image of the second image capturing device.

[0092] Referring to Figure 9, the process of matching object information for each guest is described in detail.

[0093] In FIG. 9, A1 is object area position information among individual object information extracted from a first reading target image extracted from an image of a first image capturing device, and P1 is bottom center point information among individual object information extracted from a first reading target image extracted from an image of the first image capturing device. In addition, in FIG. 9, A2 is object area position information among individual object information extracted from a second reading target image extracted from an image of a second image capturing device, and P2 is bottom center point information among individual object information extracted from a second reading target image extracted from an image of the second image capturing device. In addition, in FIG. 9, A3 is object area position information among individual object information extracted from a third reading target image extracted from an image of a third image capturing device, and P3 is bottom center point information among individual object information extracted from a third reading target image extracted from an image of the third image capturing device. Here, the first to third reading target images are captured images taken at the same time.

[0094] Hereinafter, for convenience of explanation, the floor center point information among the individual object information is referred to as individual floor center point information, and the object area location information among the individual object information is referred to as individual object area location information.

[0095] The present invention can determine whether an object extracted from an image captured from an image of a plurality of image capturing devices is a single object or multiple objects by using individual floor center point information and individual object area location information.

[0096] In Fig. 9, the first case illustrates a case where three video capturing devices capture the same object. The first case illustrates a case where three individual object information (specifically, individual floor center point information and individual object area position information) represent the same object (K). Looking at the first case, it can be seen that the three individual floor center point information are quite close together, and the three individual object area position information extends in different directions centered around an overlapping area depending on the shooting direction.

[0097] In Fig. 9, the second case is when the first individual object area location information (A1) includes the second individual object area location information (A2). In this case, two different objects can be recognized. In the second case, the final floor center point (basic floor center point information) for object K1 is BP1, and the final floor center point (basic floor center point) for object K2 is BP2.

[0098] In the third case of FIG. 9, it can be seen that the second individual object area location information (A2) and the third individual object area location information (A3) extend in different directions based on the overlapping area, and the first individual object area location information (A1) slightly overlaps the edge area of ​​the second individual object area location information (A2). In addition, it can be seen that the second and third individual floor center points (P2, P3) are quite close, and the first individual floor center point (P1) is relatively far from the second and third individual floor center points (P2, P3) compared to the distance between the second and third individual floor center points (P2, P3). In this case, the service device (2) can recognize the object by distinguishing it into a first object identified by the first individual object information and a second object identified by the second and third individual object information.

[0099] In addition, the shooting angle of the imaging device, which affects the coordinates of individual object information, can also affect the identification of duplicate objects. This is because the distortion of object information can vary depending on the shooting angle. Furthermore, the properties of the object can also affect the coordinates of individual object information. The recognition rate of the floor center point can also vary depending on the width of the object and the shape of the front of the object. Furthermore, distortion caused by the algorithm that generates the bird's-eye view can also affect the coordinates of individual object information. Therefore, identifying a single object by combining object information before and after processing by the algorithm that generates the bird's-eye view can help minimize the effects of distortion caused by the algorithm that generates the bird's-eye view.

[0100] Thus, depending on the distribution of information about multiple individual objects, objects can be recognized in varying numbers, and the number of possible cases can be vast. Therefore, it may be desirable to identify factors that influence the identification of duplicate objects and use artificial intelligence to determine them.

[0101] Accordingly, the service device (2) can be equipped with a duplication removal artificial intelligence model that distinguishes duplication of objects.

[0102] The deduplication AI model can receive object information and at least one piece of information helpful in determining whether the object information is duplicated. Hereinafter, "at least one piece of information helpful in determining whether the object information is duplicated" is referred to as "duplication-related information."

[0103] The deduplication AI model can receive the following information as object information. The object information below may be extracted from target images captured by multiple imaging devices at the same time.

[0104] 1) Object location information and object area location information before conversion by an algorithm that extracts the target image from the video taken at the same time and creates a bird's eye view.

[0105] 2) Object location information and object area location information after converting the information in 1) into an algorithm that generates a bird's eye view.

[0106] The deduplication AI model can receive the following information as input for duplicate-related information. The information below may be extracted from target images captured by multiple imaging devices at the same time.

[0107] 1) Object property information (Class)

[0108] 2) Video recording device shooting angle information

[0109] 3) Shooting direction information of the video recording device

[0110] In some cases, at least one of the redundant related information may be omitted from the input of the deduplication artificial intelligence model.

[0111] 1) Object location information and object area location information before conversion by an algorithm that extracts the target image from the video taken at the same time and creates a bird's eye view.

[0112] In the above 1) information, object area location information may be omitted.

[0113] 2) Object location information and object area location information after converting the information in 1) into an algorithm that generates a bird's eye view.

[0114] In the above 2) information, object area location information may be omitted.

[0115] And, the deduplication artificial intelligence model can output object-related information that reflects the deduplication status. Hereinafter, the 'object-related information that reflects the deduplication status' is referred to as object-related information. Here, the object-related information may be individual object information for each object in a deduplication status. Here, the individual object information for each object may be at least one individual object information. If there are multiple individual object information for a single object, there may be multiple individual object information for that object (for example, in the first case of FIG. 9, there may be three individual object information for object K). The 'object-related information that reflects the deduplication status' may be very diverse. The 'object-related information that reflects the deduplication status' may include any one of an object-specific identifier determined through deduplication, individual object location information for each object, and location information of an object area for each object. Any information that can identify a duplicate object may be included in the output of the 'deduplication artificial intelligence model' of the present invention.

[0116] To perform such tasks, a redundancy control artificial intelligence model can be trained by object information, redundancy-related information, and object-related information reflecting the deduplicated state.

[0117] (B) Object-specific basic object information generation step (S320)

[0118] The service device (2) can generate basic object information using individual object information for each object in a state where duplication has been removed. Corresponding to the individual object information, the basic object information may be basic object location information, basic object area location information, and basic object progress direction information.

[0119] Here, the individual object information may be at least one of individual object location information, individual object area location information, and individual object progress direction information.

[0120] First, we describe the steps for generating basic object information using individual object location information.

[0121] When there is a plurality of pieces of individual object location information for a single object, the service device (2) can calculate basic object location information as an average value of the plurality of pieces of individual object location information. The average value can be calculated for each of the x-coordinate and the y-coordinate. Here, the individual object location information may be individual floor center point information, and the basic object location information may be basic floor center point information.

[0122] The present invention generates object information displayed on a bird's-eye view using the average value of object information extracted from target images for multiple image capture devices positioned at divided angles. Therefore, the object information displayed on the bird's-eye view can be highly reliable within an acceptable margin of error. This allows the bird's-eye view to sufficiently achieve control objectives. If the object information is generated using the average value of object information in a situation where image capture devices are installed at evenly divided angles, the reliability of the object information can be further enhanced.

[0123] The base floor center point information can act as a factor in determining the position of an object in the bird's-eye view form.

[0124] [Explanation of symbols]

[0125] 1: Video recording device

[0126] 2: Service device

[0127]

Claims

1. A step in which a service device recognizes an object in a target image to be read and extracts individual object information about the recognized object; and A method for generating basic object information in a bird's-eye view, comprising a step of generating basic object information in a bird's-eye view by having the service device match the extracted individual object information.

2. In paragraph 1, The above individual object information includes individual floor center point information, A method for generating basic object information in a bird's-eye view, characterized in that the basic object information includes basic floor center point information.

3. In paragraph 2, A method for generating basic object information, characterized in that the service device calculates basic floor center point information as an average value of a plurality of individual floor center point information.

4. In paragraph 1, A basic object information generation method characterized in that the above service device extracts individual object information by converting floor center point information using an algorithm for generating a bird's-eye view.

5. In paragraph 4, A method for generating basic object information, characterized in that the algorithm for generating the above bird's-eye view includes IPM (Inverse Perspective Mapping).

6. In paragraph 4, A method for generating basic object information, characterized in that the above floor center point information is a point on the floor that can specify the location of an object in an object area.

7. In paragraph 1, A basic object information generation method characterized in that the above-mentioned reading target image is generated by capturing images taken by each of a plurality of image capturing devices.

8. In paragraph 7, A method for generating basic object information, characterized in that the plurality of video recording devices are CCTVs.

9. In paragraph 7, A method for generating basic object information, characterized in that the plurality of image capturing devices capture an intersection.

Images