Person estimation system, person estimation program, and person estimation method

Abstract

This invention provides a person estimation system, a person estimation program, and a person estimation method that can estimate the actual size information of a person and improve the accuracy of person identification. [Solution] The person estimation system according to one embodiment estimates the person who was at the scene. The person estimation system comprises: an image acquisition unit that acquires an image showing the person who was at the scene and the background of the scene; a 3D coordinate acquisition unit that acquires the 3D coordinates of the background in the image; a person display unit that displays a person image P4 on a 3D coordinate acquired image P3, which is an image from which the 3D coordinates have been acquired by the 3D coordinate acquisition unit; a person image adjustment unit that adjusts the shape, size, and posture of the person image P4 relative to the background B of the 3D coordinate acquired image P3; and a person model generation unit that generates a 3D model of the person from the person image P4 adjusted by the person image adjustment unit.

Description

Technical Field

[0001] The present disclosure relates to a person estimation system, a person estimation program, and a person estimation method for estimating a person present at a scene.

Background Art

[0002] Non-Patent Document 1 describes SMPL-X and SMPLify-X related to models that capture the body together with the face and hands. In the technology described in Non-Patent Document 1, a 3D human body shape model is estimated from a single image. Non-Patent Document 2 describes a technology for restoring a 3D model of a human body from a single image. Non-Patent Document 3 describes a biometric technology for analyzing the walking of a person. In this technology, a silhouette is used. Non-Patent Document 4 describes a gait authentication method that robustly performs gait authentication even when clothing conditions are different.

Prior Art Documents

Non-Patent Documents

[0003]

Non-Patent Document 1

Non-Patent Document 2

Non-Patent Document 3

Non-Patent Document 4

[0004] The technologies described in Non-Patent Documents 1 and 2 can estimate a three-dimensional human body shape model, but they have the problem that the dimensions are not determined. In all of the aforementioned technologies, it is currently difficult to estimate actual size information from images of people. In the technologies described in Non-Patent Documents 3 and 4, there is no actual size information of the person, and it can be difficult to perform accurate personal identification, so there is room for improvement in the accuracy of person identification.

[0005] This disclosure aims to provide a person estimation system, a person estimation program, and a person estimation method that can estimate the actual size information of a person and improve the accuracy of person identification. [Means for solving the problem]

[0006] (1) The person estimation system relating to this disclosure estimates the person who was at the scene. The person estimation system comprises: an image acquisition unit that acquires an image showing the person who was at the scene and the background of the scene; a 3D coordinate acquisition unit that acquires the 3D coordinates of the background in the image; a person display unit that displays the person's image on the 3D coordinate acquisition image, which is an image from which the 3D coordinates have been acquired by the 3D coordinate acquisition unit; a person image adjustment unit that adjusts the shape, size, and posture of the person's image relative to the background of the 3D coordinate acquisition image; and a person model generation unit that generates a 3D model of the person from the person image adjusted by the person image adjustment unit.

[0007] In this person estimation system, the image acquisition unit acquires images showing the person present at the scene and the background of the scene, and the 3D coordinate acquisition unit acquires the 3D coordinates of the background in the image. The person display unit displays the person's image on the image with acquired 3D coordinates, and the person image adjustment unit allows adjustment of the shape, size, and posture of the person's image relative to the background. Therefore, by adjusting the shape, size, and posture of the person's image relative to the background of the image with acquired 3D coordinates, accurate information on the person's height, weight, and posture can be obtained. Thus, the person's actual size information can be easily and accurately estimated. From the person image adjusted by the person image adjustment unit, the person model generation unit generates a 3D model of the person. Thus, the person's actual size information can be estimated, and the person can be easily and accurately identified.

[0008] (2) In (1) above, the person estimation system may include a distortion correction unit that corrects lens distortion in the image. The 3D coordinate acquisition unit may acquire the 3D coordinates of the background in the image from which lens distortion has been corrected by the distortion correction unit. In this case, the distortion correction unit can eliminate the lens distortion in the image.

[0009] (3) In (1) or (2) above, the person model generation unit may estimate the person's height and weight from the person image adjusted by the person image adjustment unit. In this case, it becomes possible to estimate the person's height and weight with higher accuracy.

[0010] (4) In any of (1) to (3) above, the person model generation unit may perform one-to-one authentication of a person from the person image adjusted by the person image adjustment unit. In this case, the identification of the person can be performed with higher accuracy.

[0011] (5) The person estimation program relating to this disclosure estimates the person who was at the scene. The person estimation program causes a computer to perform the following steps: acquire an image showing the person who was at the scene and the background of the scene; acquire the three-dimensional coordinates of the background in the image; display the person's image on the image from which the three-dimensional coordinates were acquired in the step of acquiring the three-dimensional coordinates; adjust the shape, size, and posture of the person's image relative to the background of the image from which the three-dimensional coordinates were acquired; and generate a three-dimensional model of the person from the person's image adjusted in the adjustment step.

[0012] (6) The person estimation method relating to this disclosure estimates the person who was at the scene. The person estimation method comprises the steps of: acquiring an image showing the person who was at the scene and the background of the scene; acquiring the three-dimensional coordinates of the background in the image; displaying the person image on the three-dimensional coordinate acquisition image from which the three-dimensional coordinates were acquired in the step of acquiring the three-dimensional coordinates; adjusting the shape, size, and posture of the person image relative to the background of the three-dimensional coordinate acquisition image; and generating a three-dimensional model of the person from the person image adjusted in the adjustment step.

[0013] In the person estimation program and method described above, images showing the people present at the scene and the background are acquired, and the 3D coordinates of the background in those images are obtained. A person's image is displayed on the image with acquired 3D coordinates, and the shape, size, and posture of the person's image are adjusted relative to the background. Then, a 3D model of the person is generated from the adjusted person's image. Therefore, similar to the person estimation system described above, the shape, size, and posture of the person's image can be adjusted relative to the background of the image with acquired 3D coordinates, allowing for accurate information on the person's height, weight, and posture, and enabling easy and highly accurate estimation of the person's actual size. In addition to estimating the person's actual size, it is also possible to easily and accurately identify the person. [Effects of the Invention]

[0014] According to the present disclosure, it is possible to estimate the actual size information of a person and improve the accuracy of identifying the person.

Brief Description of the Drawings

[0015] [Figure 1] FIG. 1 is a block diagram showing the functional configuration of a person estimation system and a person estimation program according to an embodiment. [Figure 2] FIG. 2 is a diagram showing an example of an image acquired by an image acquisition unit. [Figure 3] FIG. 3 is a diagram showing an example of a background 3D model for acquiring 3D coordinates. <\(0000076\)><\(0000077\)>FIG. 4 is a diagram showing examples of an image with acquired 3D coordinates and a person image. <\(0000078\)>[[ID=,19]]<\(0000079\)>FIG. 5 is a diagram showing an example of a person image. <\(0000080\)><\(0000081\)>FIG. 6 is a diagram showing an example of adjustment of a person image with respect to the background and the principle of actual scaling. <\(0000082\)><\(0000083\)>FIG. 7 is a flowchart showing an example of the steps of a person estimation method according to an embodiment. <\(0000084\)><\(0000085\)>

Modes for Carrying Out the Invention

[0018] "Person estimation" indicates identifying a person. In the person estimation system, person estimation program, and person estimation method according to this embodiment, an image showing the persons present at the scene and the background of the scene is acquired. "Background" indicates the scenery other than the persons in the captured image.

[0019] FIG. 1 is a block diagram showing a person estimation system 1 as an example of this embodiment. For example, the person estimation system 1 includes an information terminal 2 and a person estimation program 10. The information terminal 2 is an example of a computer. The person estimation program 10 is, for example, an application installed in the information terminal 2 and operable by a user. The person estimation program 10 may be, for example, a download application downloaded from a server, or an application including a function as a communication application.

[0020] The information terminal 2 is, for example, a device capable of executing the functions of the person estimation system 1 and the person estimation program 10. The information terminal 2 may be, for example, a mobile terminal. The mobile terminal indicates a portable terminal such as a mobile phone including a smartphone, a tablet, a camera, or a notebook computer. The information terminal 2 may be a terminal other than a mobile terminal, and may be, for example, a personal computer or a workstation.

[0021] Information terminal 2 comprises a processor (e.g., CPU) that runs an operating system and software (applications), a main memory unit composed of ROM and RAM, an auxiliary memory unit composed of flash memory, a communication control unit composed of a wireless communication module, input devices such as buttons, and output devices such as a display. However, the configuration of information terminal 2 is not limited to the above and can be changed as appropriate. Hereinafter, the main memory unit and the auxiliary memory unit may be collectively referred to as the memory unit.

[0022] On information terminal 2, the person estimation program 10 is executed as an application. The person estimation program 10 is, for example, an application executed on information terminal 2. However, the person estimation program 10 may also be executed on a server. The person estimation program 10 may also be downloaded from the server. Below, we will describe an example in which the person estimation program 10 is an application installed on information terminal 2, and the functions of the person estimation program 10 are executed on information terminal 2.

[0023] Each function of the person estimation system 1 is realized by loading the person estimation program 10 into the processor or memory unit and executing the person estimation program 10. The processor operates the aforementioned communication control unit, input device, or output device according to the person estimation program 10, and reads and writes data to the memory unit. The data used for processing by the information terminal 2 is stored in the memory unit.

[0024] The person estimation program 10 may be a distributed processing system executed by multiple computers, a client-server system, or a cloud system. The person estimation program 10 may include, for example, a main module, a data acquisition module, a judgment module, and an output module.

[0025] Each functional element of the person estimation program 10 functions when the data acquisition module, judgment module, and output module are executed. The person estimation program 10 may be provided, for example, by being permanently recorded on a tangible storage medium such as a CD-ROM, DVD-ROM, or semiconductor memory. The person estimation program 10 may also be provided via a communication network as a data signal superimposed on a carrier wave.

[0026] For example, the person estimation program 10 includes, as functional components, an image acquisition unit 11, a line recognition unit 12, a distortion correction unit 13, a 3D model acquisition unit 14, a 3D coordinate acquisition unit 15, a camera information acquisition unit 16, a person display unit 17, a person image adjustment unit 18, and a person model generation unit 19. Some functions of the image acquisition unit 11, line recognition unit 12, distortion correction unit 13, 3D model acquisition unit 14, 3D coordinate acquisition unit 15, camera information acquisition unit 16, person display unit 17, person image adjustment unit 18, and person model generation unit 19 are implemented, for example, by open-source software.

[0027] As shown in Figures 1 and 2, the image acquisition unit 11 acquires an image P1 that shows a person M who was at site A and the background B of site A. For example, the image acquisition unit 11 displays the image P1 on the display 2b of the information terminal 2. For example, a camera C (see Figure 6) is located at site A, and the image acquisition unit 11 acquires the image P1 captured by camera C. As an example, camera C is a security camera. In this case, the person estimation system 1 and the person estimation program 10 are applied to the person's image from the security camera. However, camera C may also be the camera of a mobile terminal, and the type of camera C is not particularly limited.

[0028] The straight line recognition unit 12 recognizes straight lines in the background B shown in image P1. Image P1 is, for example, a two-dimensional image. In image P1, parts that are actually straight lines may appear as curves. The straight line recognition unit 12 has the function of recognizing parts that are actually straight lines as straight lines. "Parts that are straight lines" are, for example, the boundary line between a wall and the ground, or one side of a rectangular structure.

[0029] The straight line recognition unit 12 recognizes straight lines in the image P1 based on the operator's actions on the image P1. For example, the operator moves the cursor to a part that is actually a straight line and clicks to place multiple points T1. The straight line recognition unit 12 recognizes that the area where multiple points T1 are lined up is a straight line.

[0030] The distortion correction unit 13 corrects the lens distortion D of image P1. Image P1 captured by camera C may have lens distortion D. The distortion correction unit 13 displays image P1 with the lens distortion D removed on display 2b. For example, the distortion correction unit 13 displays image P1 with the parts recognized as straight lines by the straight line recognition unit 12 as straight lines. By correcting the lens distortion D, the distortion correction unit 13 makes parts that were not visible when lens distortion D was present visible (see Figure 3, etc.).

[0031] The 3D model acquisition unit 14 acquires a 3D model (background 3D model) of background B. A "3D model" refers to a representation of the shape, dimensions, and positional relationships of the objects captured in the background as 3D space (3D data). As shown in Figures 2 and 3, the 3D model acquisition unit 14 acquires, for example, a background 3D model P2, which is a 3D model of background B. The 3D model acquisition unit 14 generates the background 3D model P2, for example, using SfM Photogrammetry.

[0032] The 3D model acquisition unit 14 acquires real-scale information of background B in advance. The 3D model acquisition unit 14 acquires a 3D model from multiple background images of background B that are stored in advance, for example. For example, multiple background images (100 for example) are stored in the information terminal 2 by taking pictures of background B from all angles in advance.

[0033] The 3D model acquisition unit 14 acquires a background 3D model P2 from multiple background images stored in the information terminal 2. The background images are, for example, color images. In this case, the 3D model acquisition unit 14 can recognize the 3D information of background B in color, making it possible to acquire a background 3D model P2 with higher accuracy.

[0034] Furthermore, the 3D model acquisition unit 14 may acquire a 3D model of background B by photographing background B together with a measuring tape. In addition, the 3D model acquisition unit 14 may acquire a 3D model of background B from measured values ​​of each part of background B obtained by measurement. This measurement is performed, for example, by a laser scanner. Thus, the method by which the 3D model acquisition unit 14 acquires a 3D model of background B is not particularly limited.

[0035] The 3D coordinate acquisition unit 15 acquires the 3D coordinates of background B in image P1. The 3D coordinate acquisition unit 15 acquires the 3D coordinates of background B in image P1 from image P1 and a 3D model, for example, after lens distortion D has been corrected by the distortion correction unit 13. The 3D coordinate acquisition unit 15 acquires the 3D coordinates of background B from image P1 and background 3D model P2.

[0036] For example, the 3D coordinate acquisition unit 15 displays image P1 and background 3D model P2 on display 2b. For example, the operator places a point T2 on an arbitrary part of image P1 and a point T3 on the background 3D model P2 at the same location as point T2. When multiple points T2 are placed on image P1 and multiple points T3 are placed at the same location as point T2 on background 3D model P2, the 3D coordinate acquisition unit 15 acquires the 3D coordinates of background B.

[0037] The camera information acquisition unit 16 performs, for example, calibration of camera C. The camera information acquisition unit 16 acquires camera parameters of camera C relative to the background 3D model P2. The camera parameters are, for example, the position of camera C, the orientation of camera C, and the focal length of camera C (angle of view of camera C). The camera information acquisition unit 16 may, for example, transfer the camera parameters to software and simulate the reproduction of what kind of camera camera C is. Alternatively, the camera information acquisition unit 16 may acquire the 3D coordinates of background B by estimating the camera parameters.

[0038] As shown in Figure 4, for example, the 3D coordinate acquisition unit 15 displays the 3D coordinate-acquired image P3, which is the image P1 from which the 3D coordinates have been acquired, on the display 2b. The person display unit 17 displays the person image P4 on the 3D coordinate-acquired image P3. The person image P4 is an image used to identify person M by acquiring the 3D data of person M depicted in image P1. The person image P4 is, for example, a 3D human obtained by the SMPL-X model.

[0039] As shown in Figures 4 and 5, the person display unit 17 initially displays, for example, a person image P4 in a T-pose. However, the person display unit 17 may initially display a person image P4 in an upright position, and the posture of the person image P4 that the person display unit 17 initially displays is not particularly limited.

[0040] The person image adjustment unit 18 adjusts the shape, size, and posture of the person image P4 relative to the background B of the image P3 for which 3D coordinates have been acquired. The person image adjustment unit 18 allows the operator to change the shape, size, and posture of the person image P4. For example, the person image adjustment unit 18 displays the image P3 for which 3D coordinates have been acquired together with image P1. In this case, the operator can look at the person M projected onto image P1 and adjust the shape, size, and posture of the person image P4 to match the person M.

[0041] The operator, for example, uses the mouse of information terminal 2 to perform operations that change the shape, size, and posture of the person image P4. "Adjusting the shape, size, and posture of the person image" means, for example, changing the position, orientation, angle, length, and thickness of the face, torso, arms, and legs of the person image P4, as well as changing the posture of the person image P4.

[0042] The person image adjustment unit 18 may have an input section for inputting the height and weight of person M. The input section may be, for example, a text box displayed on the display 2b. Alternatively, the input section may be a pull-down menu displayed on the display 2b. Initial values ​​for the height and weight of person M may be entered into the input section.

[0043] For example, the person display unit 17 displays a person image P4 having a shape and size corresponding to the height and weight entered into the input unit. In this case, the person image P4 can be easily displayed on the display 2b. In particular, if there is some idea of ​​who person M is and their height and weight can be determined to some extent, the display and adjustment of the person image P4 can be easily performed by inputting information into the input unit.

[0044] For example, as shown in Figures 4 and 6, the person image adjustment unit 18 adjusts the size of the person image P4 so that the contact point of the person image P4 in the background B matches the contact point T4 of person M in the background B. The contact point T4 is, for example, the point where person M's body (feet, for example) touches the ground. Alternatively, the contact point T4 may be the point where person M's hand touches a wall, and the type of contact point is not particularly limited.

[0045] For example, if the person image P4 is too small, it will be far from the contact point T4, and if the person image P4 is too large, part of it will be embedded in the contact point T4. When the person image P4 is far from the contact point T4, it can be seen that the person image P4 is too small, and when part of the person image P4 is embedded in the contact point T4, it can be seen that the person image P4 is too large.

[0046] The person image adjustment unit 18 aligns the contact portion of the person image P4 with the background B (for example, the toes) with the actual contact point T4, thereby obtaining a person image P4 of the correct size, i.e., a person image P4 at actual scale. In this way, by aligning the contact portion of the person image P4 with the contact point T4 of the background B (image P3 with acquired 3D coordinates) which has been scaled to actual scale by the 3D coordinate acquisition unit 15, it becomes possible to scale the person M to actual scale.

[0047] The human model generation unit 19 generates a three-dimensional model of person M from the human image P4 adjusted by the human image adjustment unit 18. "Generating a three-dimensional model of a person" means representing the shape and dimensions (e.g., body shape) of person M as three-dimensional space (three-dimensional data). The human model generation unit 19 estimates the height and weight of person M from the human image P4 adjusted by the human image adjustment unit 18. The human model generation unit 19 may also estimate the height and weight of person M from the estimated three-dimensional model of person M. In this case, more accurate height and weight for person M can be obtained.

[0048] The human model generation unit 19 may, for example, reshape the adjusted human image P4 into a T-pose and perform analysis on the human image P4. The human model generation unit 19 may, for example, perform one-to-one authentication of person M from the human image P4 adjusted by the human image adjustment unit 18. This makes it possible to identify person M. For example, the human model generation unit 19 may compare multiple 3D models estimated from multiple images to evaluate whether they represent the same person or not. In this case, a more quantitative evaluation of person M becomes possible.

[0049] Next, the person estimation method according to this embodiment will be explained with reference to the flowchart in Figure 7. Figure 7 is a flowchart showing an example of the steps of the person estimation method. The person estimation method according to this embodiment is executed, for example, by the person estimation program 10 described above. The person estimation program 10 causes a computer (for example, information terminal 2) to execute each of the steps described later.

[0050] First, an image P1 is acquired that shows person M at site A and background B at site A (step S1, step S1). At this time, the image acquisition unit 11 acquires image P1, which is a picture taken by camera C showing person M and background B at site A. The image P1 acquired by the image acquisition unit 11 is displayed on display 2b, for example.

[0051] Next, the lens distortion D of image P1 is corrected (step S2). At this time, the distortion correction unit 13 corrects the lens distortion D of image P1. For example, the parts that the straight line recognition unit 12 recognizes as straight lines are made straight lines, and the image P1 with the lens distortion D eliminated is displayed on the display 2b.

[0052] Furthermore, a 3D model of background B is acquired (step S3). Step S3 may be performed before step S1, after step S2, or between step S1 and step S2. In step S3, the 3D model acquisition unit 14 acquires a 3D model of background B (background 3D model P2).

[0053] The 3D model of background B includes information about the actual scale of background B. The 3D model of background B may be obtained based on multiple pre-stored background images of background B, or by photographing background B together with a measuring tape. Furthermore, the 3D model of background B may be obtained from measured values ​​of various parts of background B obtained through measurement.

[0054] Next, the 3D coordinates of background B in image P1 are obtained (step S4, step of obtaining 3D coordinates). For example, the 3D coordinates of background B are obtained from image P1 and background 3D model P2. Then, the 3D coordinate acquisition unit 15 displays the image P3 with the acquired 3D coordinates on the display 2b.

[0055] The person image P4 is displayed on the 3D coordinate image P3 (step of displaying the person image, process of displaying the person image). For example, the person display unit 17 displays the person image P4 in its initial state on the 3D coordinate image P3. The posture of the person image P4 in its initial state is, for example, a T-pose or an upright posture.

[0056] The shape, size, and posture of the person image P4 relative to the background B of the 3D coordinate acquisition image P3 are adjusted (adjustment step, adjustment process). At this time, the shape, size, and posture of the person image P4 are reproduced (step S5). Specifically, initial values ​​for height and weight are entered into the input of the person image adjustment unit 18, the person image P4 is displayed, and the shape, size, and posture of the person image P4 are adjusted. As described above, the part of the person image P4 that is in contact with the background B (e.g., the toes) is aligned with the contact point T4. This is how the person M is scaled to its actual size.

[0057] A 3D model of person M is generated from the adjusted person image P4 (step S6, step S6). At this time, the person model generation unit 19 estimates the height and weight of person M from the person image P4, and one-to-one authentication of person M is performed. After these steps, the series of steps of the person estimation method is completed.

[0058] Next, the effects and benefits obtained from the person estimation system 1, person estimation program 10, and person estimation method according to this embodiment will be described. In the person estimation system 1, person estimation program 10, and person estimation method according to this embodiment, an image P1 is obtained that shows a person M who was at site A and the background B of site A, and the three-dimensional coordinates of background B in image P1 are obtained. A person image P4 is displayed in the image P3 from which the three-dimensional coordinates have been obtained, and the shape, size, and posture of the person image P4 relative to background B can be adjusted.

[0059] Therefore, by adjusting the shape, size, and posture of the person image P4 relative to the background B of the 3D coordinate image P3, accurate information on the height, weight, and posture of person M can be obtained. Thus, the actual size information of person M can be easily and accurately estimated. A 3D model of person M is generated from the adjusted person image P4. Therefore, the actual size information of person M can be estimated, and person M can be easily and accurately identified. Furthermore, it becomes possible to estimate the actual size of the 3D human body shape from a single image P1.

[0060] As mentioned above, the person estimation system 1 (person estimation program 10) may include a distortion correction unit 13 that corrects the lens distortion D of the image P1. The 3D coordinate acquisition unit 15 may acquire the 3D coordinates of the background B in the image P1 after the lens distortion D has been corrected by the distortion correction unit 13. In this case, the lens distortion D of the image P1 can be eliminated by the distortion correction unit 13.

[0061] As mentioned above, the human model generation unit 19 may estimate the height and weight of person M from the human image P4 adjusted by the human image adjustment unit 18. In this case, it becomes possible to estimate the height and weight of person M with higher accuracy.

[0062] As mentioned above, the person model generation unit 19 may perform one-to-one authentication of person M from the person image P4 adjusted by the person image adjustment unit 18. In this case, the identification of person M can be performed with higher accuracy.

[0063] As mentioned above, the person estimation system 1 (person estimation program 10) may include a 3D model acquisition unit 14 that acquires a 3D model of background B. The 3D coordinate acquisition unit 15 may acquire the 3D coordinates of background B in image P1 from the 3D model. In this case, the 3D coordinates of background B can be acquired from the 3D model of background B (for example, background 3D model P2), making it easy to acquire the 3D coordinates of background B.

[0064] As mentioned above, the 3D model acquisition unit 14 may acquire a 3D model from multiple background images B that are stored in advance. In this case, acquiring a 3D model from multiple background images that are stored in advance makes it easier to acquire a 3D model.

[0065] As mentioned above, the person image adjustment unit 18 may have an input unit for inputting the height and weight of person M. The person display unit 17 may display a person image P4 with a shape and size corresponding to the height and weight entered in the input unit. In this case, since a person image P4 with a shape and size corresponding to the height and weight entered in the input unit is displayed, the shape, size, and posture of the person image P4 can be easily adjusted.

[0066] Embodiments of the person estimation system, person estimation program, and person estimation method relating to this disclosure have been described above. However, this disclosure is not limited to the embodiments described above and may be modified within the scope of the gist described in the claims. That is, the configuration and functions of each part of the person estimation system and person estimation program, as well as the content and sequence of steps of the person estimation method, can be appropriately changed within the scope of the gist described above.

[0067] For example, in the embodiment described above, a person estimation program 10 (person estimation system 1) having an image acquisition unit 11, a line recognition unit 12, a distortion correction unit 13, a 3D model acquisition unit 14, a 3D coordinate acquisition unit 15, a camera information acquisition unit 16, a person display unit 17, a person image adjustment unit 18, and a person model generation unit 19 was described. However, the person estimation system and person estimation program may not have any of the image acquisition unit 11, line recognition unit 12, distortion correction unit 13, 3D model acquisition unit 14, 3D coordinate acquisition unit 15, camera information acquisition unit 16, person display unit 17, person image adjustment unit 18, and person model generation unit 19.

[0068] In the embodiments described above, an example was described in which the person estimation program 10 has a distortion correction unit 13. However, depending on the type of camera C, lens distortion D may not occur, in which case the distortion correction unit 13 can be omitted. Furthermore, in cases where three-dimensional measurement of the background B is possible, the three-dimensional model acquisition unit 14 can be omitted.

[0069] Furthermore, examples of gait analysis being performed in the person estimation system, person estimation program, and person estimation method were described, as well as examples of the person estimation system and person estimation program being applied to person footage from security cameras. However, the person estimation system, person estimation program, and person estimation method relating to this disclosure can be applied to various fields, such as person estimation in the science of human movement, or person estimation in medicine or healthcare. [Explanation of Symbols]

[0070] 1...Person estimation system, 2...Information terminal, 2b...Display, 10...Person estimation program, 11...Image acquisition unit, 12...Line recognition unit, 13...Distortion correction unit, 14...3D model acquisition unit, 15...3D coordinate acquisition unit, 16...Camera information acquisition unit, 17...Person display unit, 18...Person image adjustment unit, 19...Person model generation unit, A...Scene, B...Background, C...Camera, D...Lens distortion, M...Person, P1...Image, P2...Background 3D model, P3...Image with acquired 3D coordinates, P4...Person image, T1, T2, T3...Points, T4...Contact point.

Claims

1. This is a person estimation system that estimates the people who were at the scene. An image acquisition unit that acquires images showing the people present at the scene and the background of the scene, A 3D coordinate acquisition unit that acquires the 3D coordinates of the background in the aforementioned image, A person display unit that displays a person image on the image obtained by the three-dimensional coordinate acquisition unit, which is the three-dimensional coordinate acquired image, A person image adjustment unit adjusts the shape, size, and posture of the person image relative to the background of the three-dimensional coordinate acquisition image, A person model generation unit generates a three-dimensional model of a person from the person image adjusted by the person image adjustment unit, Equipped with, The person image adjustment unit adjusts the size of the person image so that the contact points of the person image in the background of the three-dimensional coordinate acquisition image match the contact points of the person in the background. Person estimation system.

2. It includes a distortion correction unit that corrects the lens distortion of the aforementioned image, The three-dimensional coordinate acquisition unit acquires the three-dimensional coordinates of the background in the image in which lens distortion has been corrected by the distortion correction unit. The person estimation system according to claim 1.

3. The person model generation unit estimates the person's height and weight from the person image adjusted by the person image adjustment unit. A person estimation system according to claim 1 or claim 2.

4. The person model generation unit performs one-to-one authentication of the person based on the person image adjusted by the person image adjustment unit. A person estimation system according to claim 1 or claim 2.

5. This is a person estimation program that estimates the people who were at the scene. On the computer, The steps include obtaining images showing the people present at the scene and the background of the scene, The steps include obtaining the three-dimensional coordinates of the background in the aforementioned image, The step of obtaining the three-dimensional coordinates includes a step of displaying a person's image on the image from which the three-dimensional coordinates have been obtained, which is the image with the three-dimensional coordinates obtained. The steps include adjusting the shape, size, and posture of the person image relative to the background of the three-dimensional coordinate acquisition image, The adjustment step involves generating a three-dimensional model of the person from the adjusted person image, Make it run, In the adjustment step, the size of the person image is adjusted so that the contact points of the person image in the background of the three-dimensional coordinate acquisition image match the contact points of the person in the background. A person estimation program.

6. This is a person estimation method for estimating the people who were at the scene. The process of obtaining images showing the people present at the scene and the background of the scene, A step of obtaining the three-dimensional coordinates of the background in the aforementioned image, The process of acquiring the three-dimensional coordinates includes a step of displaying a person's image on the three-dimensional coordinate-acquired image, A step of adjusting the shape, size, and posture of the person image relative to the background of the three-dimensional coordinate acquisition image, The process of generating a three-dimensional model of a person from the person image adjusted in the adjustment process, Equipped with, In the adjustment step, the size of the person image is adjusted so that the contact points of the person image in the background of the three-dimensional coordinate acquisition image match the contact points of the person in the background. Person estimation method.

Images