JP5373128B2 - Person identification device, person identification method, and person identification program - Google Patents

Description

  The present invention relates to a technique for specifying a person included in image information, for example.

  Conventionally, there is a technique called face recognition as a technique for identifying a person included in image information. Face recognition is a technique for specifying the position of a human face included in image information. There is a person identification technology using face recognition. In this technique, a face image of a person registered in advance and a face image included in image information acquired by a camera or the like are compared, and a person indicated by the face image included in the acquired image information is specified. In this technique, face images are compared based on a ratio of distances between parts (for example, eyes, nose, mouth) forming a face and a ratio of sizes of the parts.

Japanese Patent Laid-Open No. 2005-084815 JP 2001-283224 A Japanese Patent Laid-Open No. 2001-27396 Japanese Unexamined Patent Publication No. 2007-4767 JP 2006-284372 A Japanese Patent Laid-Open No. 08-257017

In the conventional technique for specifying a person included in image information, a person cannot be specified unless a face image is included in the image information. In particular, it is difficult to specify a person unless the front image of the face is included in the image information. However, the face information is not always included in the image information. In particular, the front image of the face is not always included.
For example, when the number of visitors is limited to a specific person, the vicinity of the entrance is photographed with a camera installed at the entrance of a store or the like, and the person included in the photographed image is identified. If the specified person is not permitted to enter, the entry is restricted. In this case, if the person is trying to enter maliciously even though entry is not permitted, it may be possible to enter with the face hidden so that the face is not reflected on the camera with luggage or the like. In addition, even general visitors may enter while facing sideways. Therefore, it is not always possible to acquire a front image of the face of a person who enters conveniently.
An object of the present invention is to make it possible to specify a person even from image information that does not include a face front image, for example.

The person specifying device according to the present invention is, for example,
As information for identifying each person of a plurality of persons, 3D model information having a bone structure of the whole body of the person and having a length between the joints of the person is stored. A three-dimensional model storage unit stored in the apparatus;
An image acquisition unit that captures an image of the person by photographing the person with the imaging device;
The image information indicates after bending and deforming a joint of a bone structure included in the 3D model information stored in the 3D model storage unit so as to match the contour of the image information of the person acquired by the image acquisition unit. The contour of the person is compared with the contour of the deformed three-dimensional model information, the length between the predetermined joints of the person obtained from the image information of the person acquired by the image acquisition unit, and the three-dimensional model information A person specifying unit that specifies a person indicated by the image information by comparing the length between the predetermined joints of the image information.

When the image acquisition unit detects that the person is in the first position by a predetermined detection device, the image acquisition unit takes an image of the person by an imaging device installed in the second position;
The person specifying unit calculates a length between the joints of the person from the first position, the second position, and the image information of the person.

The person identifying device further includes:
A texture storage unit that stores a plurality of textures for representing the surface of a human image in a storage device;
A texture extraction unit that extracts a texture for expressing the image information of the person from a plurality of textures stored in the texture storage unit by a processing device;
The person specifying unit pastes the texture extracted by the texture extracting unit on the three-dimensional model information, and compares the contour of the three-dimensional model information pasted with the texture with the contour of the human image information. A person indicated by the image information is specified.

The person specifying unit specifies the person's posture by transforming the three-dimensional model information according to the person's posture in the person's image information,
The person identifying device further includes:
An operation specifying unit is provided that specifies a motion of the person by a processing device based on the posture of the person specified by the person specifying unit.

The person specifying method according to the present invention is, for example,
As information for identifying each person of a plurality of persons, 3D model information having a bone structure of the whole body of the person and having a length between the joints of the person is stored. A person identification method using the storage device
An image acquisition step in which the imaging device captures a person and acquires image information of the person;
The processing apparatus bends and deforms the joint of the bone structure included in the three-dimensional model information stored in the storage device so as to match the contour of the person's image information acquired in the image acquisition step. The contour of the person to be shown is compared with the contour of the deformed three-dimensional model information, the length between the predetermined joints of the person obtained from the image information of the person acquired in the image acquisition step, and the three-dimensional model And a person specifying step of specifying a person indicated by the image information by comparing the length between the predetermined joints of the information.

The person specifying program according to the present invention is, for example,
As information for identifying each person of a plurality of persons, 3D model information having a bone structure of the whole body of the person and having a length between the joints of the person is stored. A person identification program using the storage device
An image acquisition process for photographing a person and acquiring the image information of the person;
The contour of the person indicated by the image information is obtained by bending and deforming a bone structure joint included in the three-dimensional model information stored in the storage device so as to match the contour of the image information of the person acquired by the image acquisition process. And the contour of the deformed 3D model information, the length between the predetermined joints of the person obtained from the image information of the person acquired by the image acquisition process, and the 3D model information By comparing the length between predetermined joints, the computer is caused to execute a person specifying process for specifying a person indicated by the image information.

  The person specifying device according to the present invention specifies a person by comparing the image of the person included in the image information with the three-dimensional model information having the bone structure of the whole body of the person stored in advance. Therefore, it is possible to specify a person even when a face image is not included in the image information.

FIG. 3 is a functional block diagram showing functions of the person identification device 100 according to the first embodiment. The flowchart which shows the production | generation process of three-dimensional model information. The figure which shows the condition which acquires the person's image information for three-dimensional model information creation. The figure which shows an example of a base model. The figure which shows an example of the texture of clothes. The figure which shows an example of three-dimensional model information. The flowchart which shows a matching process. FIG. 4 is a functional block diagram illustrating functions of a person identification device 100 according to a second embodiment. The flowchart which shows an operation | movement detection process. The figure which shows an example of the hardware constitutions of the person specific device 100.

Embodiment 1 FIG.
First, 3D model information generation processing will be described. Next, a person specifying process based on the generated three-dimensional model information will be described.

FIG. 1 is a functional block diagram showing functions of the person identification device 100 according to this embodiment.
The person identification device 100 includes an image acquisition unit 110, a three-dimensional model generation unit 120, a three-dimensional model storage unit 130 (person information storage unit), a person identification unit 140, a texture extraction unit 150, and a texture storage unit 160.
The image acquisition unit 110 captures a person with a photographing device such as a camera and obtains the image information of the person.
Based on the image information of the person acquired by the image acquisition unit 110, the 3D model generation unit 120 generates the 3D model information of the person by the processing device. The three-dimensional model information is obtained by pasting the texture extracted by the texture extracting unit 150 (described later) on the bone model information generated by calculating the length of a predetermined bone (main bone) of the person from the image information. The information represents the person in a three-dimensional model. Since the length of the bone and the thickness of each part of the body are different for each person, the three-dimensional model information is information that is different for each person, and is information that can identify the person.
The 3D model storage unit 130 stores the 3D model information generated by the 3D model generation unit 120 in a storage device together with personal information such as the person's name and ID.
The person specifying unit 140 is configured to store the 3D model information stored in the 3D model storage unit 130 and the image information acquired by the image acquisition unit 110 (usually an image different from the image information used when creating the 3D model information). The person included in the (information) is compared to identify who the person is.
The texture extraction unit 150 extracts a texture for representing a person included in the image information acquired by the image acquisition unit 110 from the texture stored in the texture storage unit 160 described later by the processing device.
The texture storage unit 160 stores a plurality of textures for expressing the surface of a human image in a storage device. The texture storage unit 160 stores, for example, a plurality of textures representing human skin and the like. Here, the texture storage unit 160 stores a plurality of patterns of clothing texture as a three-dimensional model. A texture is an image that is pasted to represent the texture of the surface of an object in three-dimensional computer graphics.

For example, the image acquisition unit 110 acquires a person's image information by photographing a person from at least three directions with a camera at the entrance of the store. Then, the 3D model generation unit 120 generates 3D model information of the person based on the image information, and stores the 3D model information generated by the 3D model storage unit 130.
Thereafter, when the person visits the store again, the image acquisition unit 110 acquires the image information by photographing at the entrance of the store. Then, the person specifying unit 140 specifies who is the person who has visited this time based on the three-dimensional model information generated from the image information acquired when the user visited the store before. That is, the person specifying device 100 can specify who the store visitor is based on the three-dimensional model information generated when the store visited previously.
Or after generating 3D model information based on image information acquired at the entrance of a store, the person who visited the store moves in the store. Then, the image acquisition unit 110 acquires a new image information by photographing a person in the store with a camera. The person specifying unit 140 specifies a person included in the image information newly acquired by the image acquisition unit 110 based on the 3D model information stored in the 3D model storage unit 130. That is, the person identifying device 100 identifies who is included in the image information acquired by photographing the inside of the store based on the three-dimensional model information generated from the image information obtained by photographing the entrance of the store. be able to. That is, the person identification device 100 can track a person who moves in the store based on the three-dimensional model information.

First, 3D model information generation processing will be described. FIG. 2 is a flowchart showing a process for generating three-dimensional model information.
(S11: Image acquisition step)
The image acquisition unit 110 acquires image information by photographing a person with a camera.
FIG. 3 is a diagram illustrating a situation in which the image acquisition unit 110 acquires human image information for creating three-dimensional model information. As shown in FIG. 3, the image acquisition unit 110 acquires image information by photographing a person who has passed through at least three cameras installed at different positions. That is, the image acquisition unit 110 acquires image information by photographing a person from at least three directions (for example, the front direction, the horizontal direction, and the upward direction). If there is image information from three directions, a person can be represented as a solid (three-dimensional model). The more cameras, the more accurate 3D model information can be generated.
The image acquisition unit 110 captures a person at a predetermined position from a predetermined position. If the position of the photographed person and the photographed position (camera position) are known, the distance between them can be known, and the size of the person can be calculated from the image information obtained by photographing. In other words, from the image information, human height, foot length, foot thickness, hand length, hand thickness, shoulder width, face size, face part size, distance between face parts Etc. can be calculated. Therefore, for example, a passage sensor is installed near the entrance of the store, and when the passage sensor detects that a person has passed, the image is captured by at least three cameras installed at predetermined positions. A person in the position can be photographed. The position of the camera can naturally be known because it is the position where it was installed.

(S12: Shape adjustment step)
The three-dimensional model generation unit 120 matches the shape of the base model (pseudo trunk) serving as the base of the three-dimensional model information with the image information acquired by the image acquisition unit 110 in (S11) using the processing device.
FIG. 4 is a diagram illustrating an example of the base model. In FIG. 4, the base model is shown in a plan view, but is actually expressed in three dimensions. In addition, the base model has a minimum bone structure necessary for representing human movement (in FIG. 4, only a part of bones is shown for simplicity). Therefore, the base model can be transformed into any posture taken by a human. In FIG. 4, the face portion of the base model is omitted for simplicity except for the jaw portion. In FIG. 4, bones are represented by lines, and joints are represented by circles.
The three-dimensional model generation unit 120 transforms the base model into a human posture in the image information acquired by the image acquisition unit 110. The three-dimensional model generation unit 120 can, for example, extract a person's contour from image information and transform the base model to match the extracted contour, thereby transforming the base model into a person's posture in the image information.

(S13: Size adjustment step)
The three-dimensional model generation unit 120 changes the size of the base model deformed by the three-dimensional model generation unit 120 in (S12) to the size of a person obtained from the image information acquired by the image acquisition unit 110 in (S11).
That is, the three-dimensional model generation unit 120 calculates the bone length and shape of the person from the image information acquired by the image acquisition unit 110 in (S11). Then, the three-dimensional model generation unit 120 changes the bone length and shape of the base model deformed by the three-dimensional model generation unit 120 in (S12) to the calculated bone length and shape by the processing device. Thereby, the three-dimensional model generation unit 120 generates a three-dimensional bone model (bone model information) corresponding to the person in the image information.
Since people are wearing clothes, the length and shape of some bones may not be specified from the image information acquired by the image acquisition unit 110 in (S11). For example, when a person is wearing large (loose) clothes, the length and shape of the part covered by the clothes may not be specified. In this case, for example, the approximate length or shape may be estimated and calculated from the length or shape of other bones, or the length or thickness of a portion where the length or thickness cannot be specified is ignored and the three-dimensional Model information may be generated.

(S14: Texture pasting step)
The texture extraction unit 150 extracts textures for expressing a person in the image information extracted by the image acquisition unit 110 in (S11) from the plurality of textures stored in the texture storage unit 160.
Here, it is assumed that the texture extraction unit 150 stores one type of skin texture and a plurality of clothing textures. FIG. 5 is a diagram illustrating an example of clothing texture stored in the texture storage unit 160. In FIG. 5, for the sake of simplicity, the texture of clothes is shown as a plane, but the texture of clothes is a three-dimensional model. Therefore, the texture of the clothes stored in the texture storage unit 160 can be freely deformed by bending the sleeve portion. That is, when the clothing texture is pasted on the 3D bone model information and the 3D bone model is dressed, the shape of the clothing texture can be changed in accordance with the posture change of the 3D bone model. In other words, if the arm of the 3D bone model is bent, the sleeve of the texture of the clothes will also bend. As shown in FIG. 5, the texture storage unit 160 stores textures of a plurality of clothes such as a short sleeve shirt, a long sleeve shirt, a jacket, a jumper, long pants, and a skirt. The texture storage unit 160 stores, for example, textures of various shapes such as collared, collarless, slender, standard, thick, crew neck, and V-neck even for a short-sleeved shirt. That is, it is assumed that the texture storage unit 160 stores a texture expressing the silhouette of general clothes.
Therefore, the texture extracting unit 150 first separates the clothes and the parts other than the clothes from the contours for each color of the human image in the image information by the processing device. Then, the texture extracting unit 150 selects and extracts the texture of the clothes corresponding to the clothes worn by the person in the image information from the outline or the like of the clothes portion by the processing device. Further, the texture extraction unit 150 extracts skin texture.
Then, the three-dimensional model generation unit 120 attaches the texture extracted by the texture extraction unit 150 to the bone model information generated in (S13). The texture extraction unit 150 applies the skin texture in accordance with the size (thickness, etc.) of the corresponding portion, and deforms and applies the clothing texture according to the posture of the person and the size of the clothing in the image information. To do. The pasting of the texture of clothes means that the clothes indicated by the texture of clothes are put on the bone model indicated by the bone model information. As a result, the three-dimensional model generation unit 120 generates three-dimensional model information.
FIG. 6 is a diagram illustrating an example of the three-dimensional model information. In FIG. 6, the face portion is omitted for simplicity. The three-dimensional model information represents the posture of a person in the image information, the length and thickness of each part (foot, torso, arms, neck, etc.), and the silhouette of the clothes worn as a three-dimensional model. As described above, the three-dimensional model information is a three-dimensional model, and is generated based on a veil model having a minimum bone structure necessary for representing human movement. It can be transformed into any posture. At this time, the texture of the affixed clothes is also deformed according to the posture.
Then, the 3D model storage unit 130 stores the generated 3D model information in the storage device. In addition, when it is specified who the person indicated by the 3D model information is, the 3D model storage unit 130 stores the personal information of the person in association with the 3D model.

In (S12) and (S13), the three-dimensional model generation unit 120 changes the posture and size within a range that can be specified from the image information acquired by the image acquisition unit 110 in (S11). That is, the posture and size of a portion that becomes a blind spot of the camera and cannot be specified from the image information may be ignored. In addition, it is memorize | stored with 3D model information whether it was the part which can be specified from image information. This is because matching is performed based only on a part that can be specified from the image information in the person specifying process described later.
In the above description, the description has been made on the assumption that three-dimensional model information of the whole body of a person is generated. However, the present invention is not limited to this, and three-dimensional model information of only a part of a human body may be generated. For example, three-dimensional model information from the base of a person to the heel or three-dimensional model information of a face may be generated. In this case, in (S12), the three-dimensional model generation unit 120 first needs to cut out an image of a part to be generated of the three-dimensional model information from the image information acquired by the image acquisition unit 110 in (S11).
Further, a plurality of image information may be acquired by performing photographing a plurality of times. Thereby, the image information of the person's different posture can be obtained. By obtaining image information of different postures, it is possible to specify from which position each part is deformed and in which direction. That is, the deformation direction (direction of bending of the joint) and the movable shaft portion (joint position) can be specified. As a result, it is possible to more accurately identify which joint is each joint, and it is possible to more accurately obtain the length between the joints (the length of each part).

Next, a person specifying process (matching process) using the three-dimensional model information will be described. FIG. 7 is a flowchart showing the matching process.
(S21: Image acquisition step)
The image acquisition unit 110 acquires image information by photographing a person with a camera. Here, unlike the case of generating the three-dimensional model information, the image acquisition unit 110 may acquire the image information by photographing a person from one direction.
Similarly to (S11), the image acquisition unit 110 captures a person at a predetermined position with a camera installed at the predetermined position. For example, a person who visits the store again is photographed with a camera installed at a predetermined position when the passage sensor detects that the person has passed, as in (S11). Thereby, the position of the photographed person can be known, and the size of the person can be calculated from the acquired image information. In addition, the image acquisition unit 110 may calculate the position of a person who is photographed from the position of another object included in the photographed image. In other words, the position of the person included in the image information obtained by photographing with the camera for photographing the inside of the store is calculated from the position of other objects (for example, desks, chairs, planting, etc.) included in the image information. Also good.

(S22: 3D model adjustment step)
The person specifying unit 140 transforms the 3D model information stored in the storage device by the 3D model storage unit 130 into a human posture included in the image information acquired by the image acquisition unit 110 in (S21) by the processing device.
For example, the person identification unit 140 converts the 3D model information into image information by changing the posture of the 3D model information so that the outline of the 3D model information matches the outline of the person included in the image information. It can be transformed into the posture of the person involved. At this point, the person specifying unit 140 may change the orientation of the 3D model information, and if the outline of the 3D model information cannot be matched with the outline of the person included in the image information, the image information It is determined that the person included in the list does not match the person indicated by the 3D model information.

(S23: Comparison step)
The person identification unit 140 compares the person included in the image information acquired by the image acquisition unit 110 in (S21) with the three-dimensional model information deformed in (S22) by the processing device.
For example, the person specifying unit 140 compares the person included in the image information with the three-dimensional model information based on the size. For example, the length (bone length) from the knee to the heel between the person included in the image information and the three-dimensional model information is compared. In addition, the size of other portions, for example, the length from the elbow to the wrist (bone length) may be compared. Since the length of the bone varies from person to person, it is possible to identify who the person is by finding 3D model information in which the bone length matches that of the person included in the image information. In particular, the specific accuracy can be increased by finding a three-dimensional model in which the lengths of a plurality of bones match.
Moreover, you may compare not with the length of a bone but with the thickness of the predetermined part. For example, the comparison may be made based on the thickness of the knee circumference or the thickness of the ankle. Moreover, the specific accuracy can be further increased by comparing both the length of the bone and the thickness of the predetermined portion.
Furthermore, you may compare about the part of clothes by the shape of clothes. In other words, the outline of the clothes worn by the person included in the image information and the outline of the texture of the clothes (three-dimensional model information of the clothes) deformed as the 3D model information is transformed into the posture of the person. You may compare. In addition to the comparison by the length of the bone and the thickness of the predetermined portion, the specific accuracy can be further increased by performing the comparison by the shape of the clothes. Each visitor's clothes will change. Therefore, the comparison based on the shape of clothes identifies who is included in the image information acquired by photographing the inside of the store based on the three-dimensional model information generated from the image information obtained by photographing the entrance of the store. However, it may not function effectively when identifying who the store visitor is based on the three-dimensional model information generated when the store visited previously.
As a result of the comparison, if the person included in the image information matches the three-dimensional model information, the person included in the image information can be identified as the person indicated by the three-dimensional model information. On the other hand, if they do not match, it is possible to specify that the person included in the image information is not the person indicated by the three-dimensional model information.

As described above, the person identifying apparatus 100 can identify a person by using an image of another part of the human body even when the human face image is not included in the image information. Of course, the person may be specified by the face image of the person.
In particular, the person identifying apparatus 100 does not perform comparison based on the distance between the facial parts or the size ratio of the facial parts as in the conventional face recognition, but performs comparison based on the size of a part of the body. Therefore, even when a person is specified by a person's face image, it is possible to compare only the portion included in the image information even if it is not a front image of a person's face but a profile image.
Further, since the person specifying device 100 transforms and compares the three-dimensional model information with the posture of the person indicated by the image information, the person can be specified regardless of the posture of the person. In addition, the person specifying device 100 can specify a person and the posture of the person.

  In addition, since the person specifying device 100 can specify a person regardless of the posture of the person, the person specifying the person at the entrance of the store or the like is specified to restrict entrance, etc. It is possible to perform high-accuracy admission restrictions (unauthorized persons can enter or admission is not restricted). In other words, it is possible to specify who is hiding his face so that his / her face does not appear on the camera with luggage, etc., or who is entering while facing sideways, and can restrict entry with high accuracy. . Similarly, it is possible to specify who the person moving in the store is in any posture, and the person moving in the store can be tracked with high accuracy.

  In the above description, the person specifying unit 140 has been described as comparing the person included in the image information with the three-dimensional model information based on the size. Therefore, it is necessary to specify the position of the person included in the image information and the position of the photographing apparatus that acquired the image information. However, when the position of the person included in the image information or the position of the photographing apparatus that acquired the image information cannot be specified, the person included in the image information and the three-dimensional model information are connected to the length of the bone or the thickness of a predetermined part. Comparison may be made according to the ratio. For example, the person specifying unit 140 compares the length of the person included in the image information with the three-dimensional model information from the knee to the heel, the length from the base of the foot to the knee, and the length from the elbow to the wrist. You may do that. In this case, in (S21), the image acquisition unit 110 does not need to acquire image information that can calculate the size of a person. In this case as well, a comparison according to the shape of clothes may be performed.

In the above description, the three-dimensional model information is transformed in (S22) and compared in (S23). Here, when the 3D model storage unit 130 stores a plurality of pieces of 3D model information, when checking which 3D model information matches among 3D model information, each 3D model is stored. Information (S22) and (S23) are processed to find matching three-dimensional model information.
When the 3D model storage unit 130 stores a large amount of 3D model information, it takes a lot of time to perform the processes (S22) and (S23) for all the 3D model information. Therefore, the three-dimensional model storage unit 130 may store information (feature information) indicating human characteristics indicated by the three-dimensional model information together with the three-dimensional model information. The feature information is, for example, sex, age, and the like. The person identifying unit 140 identifies the person's gender and approximate age from the image information, and narrows down the three-dimensional model information based on the identified gender and age. Note that any method may be used to specify the sex and the approximate age. Then, the person specifying unit 140 may perform the processes (S22) and (S23) only for the narrowed-down three-dimensional model information.

  In addition, the 3D model generation unit 120 may identify who is the store visitor from the image information acquired when visiting the store, and may update the 3D model information based on the image information acquired when visiting the store. By repeatedly acquiring image information of the same person from various angles, the three-dimensional model information can be made more accurate. Also, even when the size changes due to growth or the thickness of the body changes, the 3D model information can be made more accurate by updating the 3D model information each time a visit is made. it can.

Embodiment 2. FIG.
In this embodiment, a person identifying apparatus 100 that detects a person who performs a specific action and issues a warning based on the three-dimensional model information will be described.

FIG. 8 is a functional block diagram showing functions of the person identifying apparatus 100 according to this embodiment. The person identification device 100 according to this embodiment includes an operation storage unit 170 and an operation identification unit 180 in addition to the functions of the person identification device 100 according to the first embodiment.
The motion storage unit 170 stores information (script) indicating a predetermined motion (a motion of a predetermined bone model) in a storage device.
The operation specifying unit 180 detects the operation stored in the operation storage unit 170 from the image information acquired by the image acquisition unit 110 by the processing device. Then, when detecting the operation stored in the operation storage unit 170, the operation specifying unit 180 notifies the administrator or the like of the person specifying device 100.

A motion detection process using the three-dimensional model information will be described. FIG. 9 is a flowchart showing the operation detection process.
(S31: Image acquisition step)
Similarly to (S21), the image acquisition unit 110 acquires image information by photographing a person with a camera.
(S32: 3D model adjustment step)
Similarly to (S22), the person specifying unit 140 uses the 3D model information stored in the storage device by the 3D model storage unit 130 as the posture of the person included in the image information acquired by the image acquisition unit 110 in (S21). It is deformed by the processing device. Thereby, the person specifying unit 140 specifies the posture of the person included in the image information.
(S33: Operation specifying step)
The action specifying unit 180 compares the posture of the person included in the image information specified by the person specifying unit 140 with the posture of the person one hour before, and specifies the action of the person from one hour before to the current time. .
And it returns to (S31) and repeats a process. That is, acquisition of image information at the next time, identification of posture, and identification of motion are repeated. Thereby, the action of the person at a predetermined time can be specified.
(S34: Operation comparison step)
The action specifying unit 180 compares the action specified in (S34) with the action indicated by the information stored in the action storage unit 170. As a result of the comparison, the operation specifying unit 180 matches the operation specified in (S34) with the operation indicated by the information stored in the operation storage unit 170, that is, the operation indicated by the information stored in the operation storage unit 170. If detected, the administrator of the person identifying apparatus 100 is notified. For example, the action specifying unit 180 outputs a warning message, sound, light, or the like to the terminal of the administrator of the person specifying device 100.

  That is, since the person specifying apparatus 100 according to this embodiment views a person as a three-dimensional model, the person is tracked by the three-dimensional model and the posture of the person is specified. The movement of the person is specified by specifying the posture of the person continuously. For example, when a predetermined movement is detected, an alarm can be output. For example, a warning can be issued when an operation when attempting to shoot a pistol is detected. Further, by specifying the movement of the person, it is possible to specify the situation (for example, mental situation) of the person. In other words, it is possible to identify a situation such as frustration.

The above is summarized as follows.
The person specifying device according to the present invention is, for example,
For each of a plurality of people, as information for identifying the person, a 3D model storage unit that stores the 3D model information of the person in a storage device;
An image acquisition unit that captures an image of the person by photographing the person with the imaging device;
A person specifying unit that specifies a person indicated by the image information by comparing the image information of the person acquired by the image acquiring unit with the 3D model information stored by the 3D model storage unit; Features.

The person specifying unit specifies a person indicated by the image information by transforming the three-dimensional model information in accordance with the posture of the person in the image information of the person and comparing it with the image information of the person. It is characterized by.

The 3D model storage unit stores 3D model information having information on a predetermined bone length of a person,
The person specifying unit calculates the length of the predetermined bone of the person from the image information of the person acquired by the image acquisition unit, and the bone length indicated by the 3D model information stored in the 3D model storage unit And the person indicated by the image information is specified.

When the image acquisition unit detects that the person is in the first position by a predetermined detection device, the image acquisition unit takes an image of the person by an imaging device installed in the second position;
The person specifying unit calculates a length of the predetermined bone of the person from the first position, the second position, and the image information of the person.

The person identifying device further includes:
A texture storage unit that stores a plurality of textures for representing the surface of a human image in a storage device;
A texture extraction unit that extracts a texture for expressing the image information of the person from a plurality of textures stored in the texture storage unit by a processing device;
The person specifying unit pastes the texture extracted by the texture extracting unit on the three-dimensional model information, and compares the three-dimensional model with the texture pasted with the person's image information to obtain the image information. The person who shows is specified.

The 3D model storage unit stores a plurality of 3D model information, and stores specific information for specifying a person indicated by each 3D model of the plurality of 3D model information,
The person specifying unit narrows down the three-dimensional model information based on the person's image information and the specific information, and compares the person's image information with the narrowed-down three-dimensional model information. It is characterized in that the person indicated by is identified.

The person specifying unit specifies the person's posture by transforming the three-dimensional model information according to the person's posture in the person's image information,
The person identifying device further includes:
An operation specifying unit is provided that specifies a motion of the person by a processing device based on the posture of the person specified by the person specifying unit.

The person identifying device further includes:
An operation storage unit storing information indicating a predetermined operation in a storage device;
The operation specifying unit outputs a warning when the specified operation is an operation stored in the operation storage unit.

The person specifying method according to the present invention includes, for example, an image acquisition step in which a photographing apparatus photographs a person and acquires image information of the person;
A processing device that compares the image information of the person acquired in the image acquisition step with the three-dimensional model information of the person stored in advance in the storage device, thereby specifying a person specifying step of specifying the person indicated by the image information; It is characterized by providing.

The person specifying program according to the present invention is, for example,
An image acquisition process for photographing a person and acquiring the image information of the person;
Causing the computer to execute a person specifying process for specifying the person indicated by the image information by comparing the image information of the person acquired in the image acquiring process with the three-dimensional model information stored in the storage device in advance. Features.

The person specifying device according to the present invention is, for example,
For each person of a plurality of persons, as information for identifying the person, a person information storage unit that stores the size of a predetermined part of the person in a storage device;
An image acquisition unit that, when a predetermined detection device detects that a person is in the first position, takes an image of the person and acquires image information of the person by an imaging device installed in the second position;
The size of the predetermined portion of the person is calculated from the image information of the person acquired by the image acquisition unit, the first position, and the second position, and compared with the size stored in the person information storage unit And a person specifying unit for specifying a person indicated by the image information.

Next, the hardware configuration of the person identification device 100 in the embodiment will be described.
FIG. 10 is a diagram illustrating an example of a hardware configuration of the person identification device 100.
As shown in FIG. 10, the person identification device 100 includes a CPU 911 (also referred to as a central processing unit, a central processing unit, a processing unit, an arithmetic unit, a microprocessor, a microcomputer, and a processor) that executes a program. The CPU 911 is connected to the ROM 913, the RAM 914, the LCD 901 (Liquid Crystal Display), the keyboard 902 (K / B), the communication board 915, and the magnetic disk device 920 via the bus 912, and controls these hardware devices. Instead of the magnetic disk device 920, a storage device such as an optical disk device or a memory card read / write device may be used.

  The ROM 913 and the magnetic disk device 920 are examples of a nonvolatile memory. The RAM 914 is an example of a volatile memory. The ROM 913, the RAM 914, and the magnetic disk device 920 are examples of a storage device (memory). The keyboard 902 and the communication board 915 are examples of input devices. The communication board 915 is an example of a communication device. Furthermore, the LCD 901 is an example of a display device.

  An operating system 921 (OS), a window system 922, a program group 923, and a file group 924 are stored in the magnetic disk device 920 or the ROM 913. The programs in the program group 923 are executed by the CPU 911, the operating system 921, and the window system 922.

The program group 923 includes “image acquisition unit 110”, “3D model generation unit 120”, “3D model storage unit 130”, “person identification unit 140”, “texture extraction unit 150”, “ Stored are software, programs, and other programs that execute the functions described as the texture storage unit 160, the operation storage unit 170, the operation specifying unit 180, and the like. The program is read and executed by the CPU 911.
In the file group 924, information, data, signal values, variable values, and parameters stored in “image information”, “3D model information”, “texture”, etc. in the above description are stored in “file” or “database”. Stored as each item. The “file” and “database” are stored in a recording medium such as a disk or a memory. Information, data, signal values, variable values, and parameters stored in a storage medium such as a disk or memory are read out to the main memory or cache memory by the CPU 911 via a read / write circuit, and extracted, searched, referenced, compared, and calculated. Used for the operation of the CPU 911 such as calculation / processing / output / printing / display. Information, data, signal values, variable values, and parameters are temporarily stored in the main memory, cache memory, and buffer memory during the operation of the CPU 911 for extraction, search, reference, comparison, calculation, calculation, processing, output, printing, and display. Is remembered.

In the above description, the arrows in the flowchart mainly indicate input / output of data and signals, and the data and signal values are recorded in a memory of the RAM 914, other recording media such as an optical disk, and an IC chip. Data and signals are transmitted online by a bus 912, signal lines, cables, other transmission media, and radio waves.
In addition, what is described as “to part” in the above description may be “to circuit”, “to device”, “to device”, “to means”, and “to function”. It may be “step”, “˜procedure”, “˜processing”. In addition, what is described as “˜device” may be “˜circuit”, “˜device”, “˜device”, “˜means”, “˜function”, and “˜step”, “ ~ Procedure "," ~ process ". Furthermore, what is described as “to process” may be “to step”. That is, what is described as “˜unit” may be realized by firmware stored in the ROM 913. Alternatively, it may be implemented only by software, or only by hardware such as elements, devices, substrates, and wirings, by a combination of software and hardware, or by a combination of firmware. Firmware and software are stored in a recording medium such as ROM 913 as a program. The program is read by the CPU 911 and executed by the CPU 911. That is, the program causes a computer or the like to function as the “˜unit” described above. Alternatively, the computer or the like is caused to execute the procedures and methods of “to part” described above.

  DESCRIPTION OF SYMBOLS 100 Person specific apparatus, 110 Image acquisition part, 120 3D model production | generation part, 130 3D model memory | storage part, 140 Person specific part, 150 Texture extraction part, 160 Texture memory | storage part, 170 Motion storage part, 180 Motion specification part

Claims (6)

As information for identifying each person of a plurality of persons, 3D model information having a bone structure of the whole body of the person and having a length between the joints of the person is stored. A three-dimensional model storage unit stored in the apparatus;
An image acquisition unit that captures an image of the person by photographing the person with the imaging device;
The image information indicates after bending and deforming a joint of a bone structure included in the 3D model information stored in the 3D model storage unit so as to match the contour of the image information of the person acquired by the image acquisition unit. The contour of the person is compared with the contour of the deformed three-dimensional model information, the length between the predetermined joints of the person obtained from the image information of the person acquired by the image acquisition unit, and the three-dimensional model information A person specifying apparatus comprising: a person specifying unit that specifies a person indicated by the image information by comparing a length between the predetermined joints of the image information. When the image acquisition unit detects that the person is in the first position by a predetermined detection device, the image acquisition unit takes an image of the person by an imaging device installed in the second position;
The person specifying device according to claim 1, wherein the person specifying unit calculates a length between the joints of the person from the first position, the second position, and the image information of the person. . The person identifying device further includes:
A texture storage unit that stores a plurality of textures for representing the surface of a human image in a storage device;
A texture extraction unit that extracts a texture for expressing the image information of the person from a plurality of textures stored in the texture storage unit by a processing device;
The person specifying unit pastes the texture extracted by the texture extracting unit on the three-dimensional model information, and compares the contour of the three-dimensional model information pasted with the texture with the contour of the human image information. The person specifying apparatus according to claim 1, wherein the person specified by the image information is specified. The person specifying unit specifies the person's posture by transforming the three-dimensional model information according to the person's posture in the person's image information,
The person identifying device further includes:
The person specifying device according to any one of claims 1 to 3, further comprising an action specifying unit that specifies a motion of the person by a processing device based on the posture of the person specified by the person specifying unit. As information for identifying each person of a plurality of persons, 3D model information having a bone structure of the whole body of the person and having a length between the joints of the person is stored. A person identification method using the storage device
An image acquisition step in which the imaging device captures a person and acquires image information of the person;
The processing apparatus bends and deforms the joint of the bone structure included in the three-dimensional model information stored in the storage device so as to match the contour of the person's image information acquired in the image acquisition step. The contour of the person to be shown is compared with the contour of the deformed three-dimensional model information, the length between the predetermined joints of the person obtained from the image information of the person acquired in the image acquisition step, and the three-dimensional model A person specifying method comprising: a person specifying step for specifying a person indicated by the image information by comparing a length between the predetermined joints of the information. As information for identifying each person of a plurality of persons, 3D model information having a bone structure of the whole body of the person and having a length between the joints of the person is stored. A person identification program using the storage device
An image acquisition process for photographing a person and acquiring the image information of the person;
The contour of the person indicated by the image information is obtained by bending and deforming a bone structure joint included in the three-dimensional model information stored in the storage device so as to match the contour of the image information of the person acquired by the image acquisition process. And the contour of the deformed 3D model information, the length between the predetermined joints of the person obtained from the image information of the person acquired by the image acquisition process, and the 3D model information A person specifying program for causing a computer to execute a person specifying process for specifying a person indicated by the image information by comparing a length between predetermined joints.

Images