JP2011081735A - Suspicious person monitoring device and suspicious person monitoring method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a suspicious person monitoring device capable of appropriately specifying a suspicious person in a monitoring area. <P>SOLUTION: The suspicious person monitoring device 10 includes: a storage unit 102 for storing person identification information for identifying a person in the monitoring area and reliability of a person in association with each other, and related information indicating relevance among a plurality of persons; a reliability estimation unit 108 for estimating reliability of a target person; a related person specification unit 110 for specifying a related person in association with the target person; a reliability calculation unit 112 for calculating reliability of the related person based on the reliability of the target person; a reliability update unit 114 for updating the reliability of the target person to the reliability estimated by the reliability estimation unit 108 and the reliability of the related person to the reliability calculated by the reliability calculation unit 112; and a suspicious person determination unit 116 for determining that the person is a suspicious person when the reliability in association with the person identification information in the storage unit 102 is less than a threshold value. <P>COPYRIGHT: (C)2011,JPO&amp;INPIT

Description

  The present invention relates to a suspicious person monitoring apparatus and a suspicious person monitoring method for monitoring a suspicious person in a monitoring area.

  Conventionally, in order to prevent a suspicious person from entering the monitoring area, a security gate is provided at the entrance of the monitoring area. However, a so-called companion who illegally enters a room following a person who has the authority to enter the room or an intrusion from a place where no security gate is provided such as a window is assumed, and the security gate alone is not sufficient.

  In order to solve this problem, it is necessary to authenticate all persons in the area, not just the entrance to the monitoring area. As such a technique, for example, an apparatus for detecting a suspicious person by detecting abnormal behavior from a user's behavior pattern is known (see, for example, Patent Documents 1 and 2).

JP 2006-65572 A JP 2004-328622 A

  However, it is difficult to judge a person who has not been authenticated to have the right to enter the room, but it is difficult to actively determine that the person is a suspicious person, and it is difficult to appropriately evaluate the person in the monitoring area. It was.

  The present invention has been made in view of the above, and an object of the present invention is to provide a suspicious person monitoring apparatus and a suspicious person monitoring method capable of appropriately identifying a suspicious person in a monitoring area.

  In order to solve the above-described problems and achieve the object, the present invention is a suspicious person monitoring apparatus that stores person identification information for identifying a person in a monitoring area and the reliability of the person in association with each other. In addition, a storage unit that stores related information indicating the relationship between a plurality of persons, a reliability estimation unit that estimates the reliability of a predetermined target person based on a detection result of an external sensor, and the related information A related person specifying unit for specifying a related person associated with the target person, and a reliability for calculating the reliability of the related person based on the reliability of the target person estimated by the reliability estimating unit When the reliability of the target person is estimated by the calculation unit and the reliability estimation unit, the reliability of the target person stored in the storage unit is estimated by the reliability estimation unit Updated every time A reliability update unit that updates the reliability of the related person stored in the storage unit to the reliability calculated by the reliability calculation unit, and is associated with the person identification information in the storage unit A suspicious person determination unit that compares the reliability with a predetermined threshold and determines that the person identified by the person identification information is a suspicious person when the reliability is less than the threshold; and When the suspicious person judging unit judges that the person is a suspicious person, the suspicious person judging unit is provided with an output unit that outputs that the suspicious person may have entered.

  Another embodiment of the present invention is a suspicious person monitoring method executed by a suspicious person monitoring apparatus, wherein the suspicious person monitoring apparatus includes person identification information for identifying a person in a monitoring area, and the reliability of the person. Is stored in association with each other, and further includes a storage unit for storing related information indicating the relationship between a plurality of persons, and a reliability estimation step of estimating the reliability of a predetermined target person based on the detection result of the external sensor And a related person specifying step of specifying a related person associated with the target person in the related information, and the reliability of the related person based on the reliability of the target person estimated in the reliability estimation step. When the reliability of the target person is estimated in the reliability calculation step for calculating the degree and the reliability estimation step, the reliability of the target person stored in the storage unit is set as the reliability A reliability update step of updating the reliability estimated in the determination step, and updating the reliability of the related person stored in the storage unit to the reliability calculated in the reliability calculation step; The reliability associated with the person identification information in the storage unit is compared with a predetermined threshold, and when the reliability is less than the threshold, the person identified by the person identification information is suspicious. A suspicious person determination step for determining that the person is a suspicious person, and an output step for outputting that the suspicious person may have entered if the suspicious person determination unit determines that the person is a suspicious person. And

  According to the present invention, there is an effect that it is possible to suppress erroneous detection of a person other than a suspicious person as a suspicious person appropriately in the monitoring area.

FIG. 1 is a diagram showing an overall configuration of a suspicious person monitoring system 1 according to a first embodiment of the present invention. FIG. 2 is a diagram illustrating a data configuration of the storage unit 102. FIG. 3 is a diagram illustrating a data configuration of the association degree table. FIG. 4 is a diagram for explaining the processing of the related person specifying unit 110. FIG. 5 is a flowchart showing the suspicious person monitoring process. FIG. 6 is a flowchart showing processing when a new reliability is estimated. FIG. 7 is a diagram for explaining processing when a new reliability is estimated. FIG. 8 is a flowchart showing a process when a relevance or the like is newly estimated. FIG. 9 is a diagram for explaining processing when new relevance or the like is estimated. FIG. 10 is a diagram for specifically explaining the process of updating the reliability. FIG. 11 is a diagram for specifically explaining the process of updating the reliability. FIG. 12 is a diagram for specifically explaining the process of updating the reliability. FIG. 13 is a diagram for specifically explaining the process of updating the reliability. FIG. 14 is a diagram for specifically explaining the process of updating the reliability. FIG. 15 is a diagram for specifically explaining the process of updating the reliability. FIG. 16 is a diagram for specifically explaining the process of updating the reliability. FIG. 17 is a diagram for specifically explaining the process of updating the reliability. FIG. 18 is a diagram for specifically explaining the process of updating the reliability. FIG. 19 is a diagram for specifically explaining the process of updating the reliability. FIG. 20 is a diagram for specifically explaining the process of updating the reliability. FIG. 21 is a diagram for specifically explaining the process of updating the reliability.

  Exemplary embodiments of a suspicious person monitoring device and a suspicious person monitoring method according to the present invention will be described below in detail with reference to the accompanying drawings.

  FIG. 1 is a diagram showing an overall configuration of a suspicious person monitoring system 1 according to an embodiment of the present invention. The suspicious person monitoring system 1 includes a laser distance meter 2, a camera 3, an authentication device 4, and a suspicious person monitoring device 10. These devices are connected via the network 5. The authentication device 4 is installed at the entrance / exit of a monitoring area for monitoring the intrusion of a suspicious person. The authentication device 4 is an authentication device that reads information on an IC chip such as an employee ID card by using, for example, RFID (Radio Frequency IDentification).

  As another example, the authentication device 4 may be a camera that captures a face image of a person and authenticates the person by matching with a pre-registered face image. As described above, the authentication device 4 is not limited to the embodiment as long as it can authenticate a person.

  The laser rangefinder 2 and the camera 3 as external sensors are installed in the monitoring area. At least one laser rangefinder 2 and camera 3 are installed in the monitoring area, and detect the position of a person in the monitoring area. Furthermore, the laser distance meter 2 and the camera 3 track each person, identify a plurality of persons, and detect a movement path, a movement speed, and the like of each person.

  The suspicious person monitoring device 10 includes a person registration unit 100, a storage unit 102, a relevance estimation unit 104, a related information update unit 106, a reliability estimation unit 108, a related person identification unit 110, and a reliability calculation unit. 112, a reliability update unit 114, a suspicious person determination unit 116, and an output unit 118.

  When the laser distance meter 2 or the camera 3 detects a person in the monitoring area, the person registration unit 100 assigns a person ID for identifying the person to the detected person, and stores the person ID in the storage unit 102.

  FIG. 2 is a diagram illustrating a data configuration of the storage unit 102. The storage unit 102 stores a person ID, reliability, and related information in association with each other. When a new person is detected by the laser distance meter 2 or the like, a new line is created for this person and the person ID is registered. The person ID may be a sequential number in the registration order by the person registration unit 100. The reliability is a value indicating the probability that the person is authorized to enter the monitoring area. The reliability ranges from 0 to 100, and the higher the value, the higher the reliability. The reliability is set to 50 as a default value, and the reliability of a person newly registered in the storage unit 102 is set to 50. When the authentication by the authentication device 4 is successful, the reliability is set to 100 because the person has the entry authority.

The related information is information indicating the relationship between a plurality of persons. For example, when the person P ₁ and the person P ₂ exchange greetings, it is understood that the person P ₁ and the person P ₂ are related. In this case, in association with the person ID of the person P ₁ , the person ID of the person P ₂ having a relationship with the person P ₁ is stored in the storage unit 102 as related information. Similarly, in association with the person ID of the person P _2, the person ID of the person P ₁ having a relationship with a person P ₂ as the relevant information is stored.

In addition to relevance, the related information includes relevance indicating the degree of relevance. The degree of relevance is a value determined according to the degree of contact between persons, for example, 0.5 when greetings are exchanged and 0.1 when they are within a predetermined distance range. A value in the range of 1 to 1. When the person P ₁ and the person P ₂ are related and the degree of association is 0.5, as shown in FIG. 2, the degree of association with the person P ₂ is associated with the person ID of the person P _1. 0.5 is stored. Likewise, in association with the person ID of the person _{P 2,} relevance 0.5 between the person _{P 1} is stored. That is, the related information between persons is expressed as a weighted undirected graph.

  In the present embodiment, related information regarding each person is stored in the storage unit 102 in association with the person ID of each person. The related information is information indicating the relationship and the degree of association between persons. The storage unit 102 may store it as information independent of the table in which the person ID and the reliability shown in FIG. In this case, the related information only needs to include person IDs of a plurality of persons having relevance and their degrees of relevance.

  The relevance estimation unit 104 estimates at least one of relevance and relevance between a plurality of persons based on the detection result of the laser distance meter 2 or the camera 3. Specifically, the relevance estimation unit 104 detects a predetermined interaction from the detection result of the laser distance meter 2 or the camera 3. Here, the interaction is the content of exchange between a plurality of persons estimated from the detection result of the laser distance meter 2 or the camera 3. Examples of the interaction include exchanging greetings, having a conversation, entering the field of view, and being in the vicinity. The relevance estimation unit 104 further includes a relevance level table illustrated in FIG. 3, and refers to the relevance level table to estimate the relevance level based on the detected interaction. In the association degree table, the interaction and the association degree are associated with each other. Further, the degree of association may be estimated according to the number of times of interaction and time.

  For example, suppose that it is detected from the movement paths of a plurality of persons by the laser distance meter 2 and the camera 3 that the plurality of persons are moving in the same direction. In this case, it is presumed that the user is talking. Therefore, it can be seen that the plurality of persons are relatively close relatives such as colleagues. Since this interaction is shown in the relevance level table shown in FIG. 3, the relevance estimation unit 104 estimates that there is relevance. Further, in this case, the relevance is estimated to be 0.7 associated with this interaction in the relevance table. In this way, a relatively high degree of association is set for the interaction that a plurality of persons are moving in the same direction.

  As shown in FIG. 3, in addition to the above, there is an interaction that faces each other for a certain time. In this case, it is presumed that greetings are being exchanged or talking. Therefore, it is estimated that there is a relevance also in this case. Furthermore, the relevance is estimated to be 0.5.

  Whether or not a plurality of persons are facing each other is detected based on, for example, the result of photographing a face image by the camera 3. As another example, it may be estimated that there is a person's face in the direction of travel from the previous movement path.

  In addition to this, interactions such as entering the field of view and existing within a specified distance are set. Whether or not it is in the field of view is determined by specifying the orientation of the person's face and assuming that the viewing angle of the person is 100 °, for example. Then, a region within a predetermined distance from the person and within the range of the viewing angle is set as the view range. Then, when a person is in the field of view, it is determined that the person is in the field of view.

  When the relevance estimation unit 104 estimates relevance and relevance between a plurality of persons, the relevance information update unit 106 stores the estimated relevance and relevance in the storage unit 102 as related information. In the case where the related information of the person who is the target of the relevance estimation is already stored in the storage unit 102, the storage unit is based on the relevance and relevance newly estimated by the relevance estimation unit 104. 102 related information is updated.

  The related information update unit 106 may further update the related information according to the attribute of the place where the person is detected, the degree of congestion, the attribute of the person in contact, and the like. For example, at the entrance of a building, it is possible that irrelevant persons move in the same direction in the vicinity of the flow line. Therefore, in this case, the relevance when moving in the same direction may be estimated to be lower than normal, such as 0.5.

  In a crowded situation, the situation may continue for a while after an unrelated person approaches. This can happen both when moving and when waiting. Therefore, in such a situation, the relevance may be estimated at a low level.

  In the case of contact with a person whose attribute is estimated, the degree of association may be varied depending on the attribute of the person. For example, if the specified person is a receptionist or a guide, even if there is an approach or conversation, the degree of relevance is estimated to be low. Also, if the identified person is a security officer or the like, the security officer checks whether or not the contacted person is a suspicious person. Good.

  The reliability estimation unit 108 estimates the reliability of the person based on the authentication result by the authentication device 4 and the like. For example, when the authentication by the authentication device 4 is successful, the reliability is estimated as 100.

  The reliability estimation unit 108 may estimate the reliability according to a plurality of other conditions. For example, if the movement path of the target person is identified from the detection results of the laser distance meter 2 and the camera 3, and the movement path is away from a general movement path registered in advance, a place where there are few people Since it is presumed that it is heading to a different place from other people, a relatively low reliability is estimated. Thus, the reliability is estimated based on the movement history.

  Further, the reliability is estimated from the appearance of the person obtained as a result of photographing by the camera 3. Specifically, general clothes are learned in advance, and when a person with a different appearance obtained by photographing is detected, a low reliability is estimated for this person. As a result, for example, a low reliability can be estimated for a suspicious person whose face is hidden by a mask or sunglasses, or who is wearing a coat in summer.

  Furthermore, as another example, the reliability estimation unit 108 may estimate the reliability based on the behavior of a person obtained from the photographing result by the camera 3 or the like. Specifically, a general behavior is learned in advance, and when the behavior of a person obtained from a photographing result or the like is different from this, a low reliability is estimated for the person. Thereby, for example, a low reliability can be estimated for a person who has taken a suspicious action such as checking a surveillance camera or placing an object.

  When the reliability of the predetermined person (hereinafter referred to as a target person) is estimated by the reliability estimation unit 108, the related person specifying unit 110 is a person (hereinafter referred to as a person who is related to the target person in the related information in the storage unit 102). , Referred to as a related person). Further, when the related information is updated by the related information update unit 106, the related person specifying unit 110 selects a person who has a relationship in the updated related information, that is, a related person of the person (target person) to be updated. Identify.

FIG. 4 is a diagram for explaining the processing of the related person specifying unit 110. As illustrated in FIG. 4, for example, when the reliability of the person P ₁ is estimated when the person P ₁ and the person P ₂ are related, the related person specifying unit 110 sets the person P ₂ as the target person. identified as relevant person P _1. Related personal identification unit 110 refers to the related information of the person P _2, the person P ₃ is the case with the relevant person P ₂ further identify a person P ₃ as a related person of the person P _1. As described above, the related person specifying unit 110 specifies not only a person directly related to the target person but also a person indirectly related to the target person as a related person based on the relationship between the related person and the related person. .

  For a person who is indirectly related via a related person, the number of links is set in advance, and for example, up to three persons (3 links) from the target person are specified as related persons. As a result, it is possible to avoid specifying a person who is substantially less related to the target person as a related person.

Related personal identification unit 110 also, when the related information of the person P ₁ and if the associated information of the person P ₂ is updated and the person P ₁ and the person P ₂ is newly created, the person P ₁ and the person P ₂ is specified as a related person, and a person having a relationship with each of the two target persons is specified as a related person.

When the reliability is estimated by the reliability estimation unit 108 and the related person is specified by the related person specifying unit 110, the reliability calculation unit 112 calculates a new target person P _j estimated by the reliability estimation unit 108. New reliability S _{j of} each related person according to (Equation 1) based on the degree of reliability S _j ′ and the related degree R _ij between each related person P _i specified by the related person specifying unit 110 and the target person. _i 'is calculated.
S _i ′ = S _i + R _ij × (S _j ′ −S _j ) (Formula 1)

Note that the degree of association between the target person and the person connected by two or more links is the product of the degree of association of each link connecting the persons. That is, the degree of association between the person P ₁ and the person P ₃ shown in FIG. 4 is R ₁₂ × R ₂₃ .

  The new reliability of the related person may be a value determined on the basis of the amount of change before and after the estimation of the reliability of the target person and the reliability of the target person and the related person. It is not limited.

In addition, when the related information is updated by the related information update unit 106, the reliability calculation unit 112 determines the relevance according to (Expression 2) based on the relevance R _ij ′ indicated in the updated related information. A new reliability S _i ′ of the target person P _i is calculated.
S _i ′ = S _i + R _ij ′ × (S _j −S _i ) (Formula 2)
Here, S _i and S _j are the reliability of the target persons P _i and P _j before the related information is updated.

  Note that the new reliability of the target person may be a value determined based on the degree of association after the related information is updated and the reliability of the target person indicated in the related information. It is not limited.

  When the related information is updated by the related information update unit 106, the reliability calculation unit 112 further uses the new reliability of the target person obtained by (Expression 2) to The reliability with respect to the related person about the object person concerning information is calculated.

  When the reliability is estimated by the reliability estimation unit 108, the reliability update unit 114 updates the reliability of the target person stored in the storage unit 102 to the reliability estimated by the reliability estimation unit 108. . In addition, the reliability of the related person is updated to the reliability calculated by the reliability calculation unit 112. When the related information in the storage unit 102 is updated by the related information update unit 106, the reliability update unit 114 calculates the reliability of the target person and the related person in the updated related information by the reliability calculation unit 112. Update to the specified confidence level. The reliability update unit 114 further updates the reliability when a predetermined period of time has elapsed since the previous update. Specifically, the reliability value is reduced by a certain amount.

  Each time the storage unit 102 is updated, the suspicious person determination unit 116 compares the reliability of each person with a predetermined threshold, and when the reliability is equal to or lower than the threshold, the suspicious person determination unit 116 determines that the person is a suspicious person. judge. When the suspicious person determination unit 116 determines that the predetermined person is a suspicious person, the output unit 118 outputs that the suspicious person may have entered. Specifically, it may be displayed on the display screen or may be output as audio. Further, when the suspicious person monitoring device 10 is connected to the monitoring center or the like via the network 5, the monitoring center may be notified to that effect.

  FIG. 5 is a flowchart showing the suspicious person monitoring process. When the reliability is updated by the reliability update unit 114 (Yes in step S100), the suspicious person determination unit 116 compares the updated reliability with a threshold value, and the updated reliability is less than the threshold value. (Step S102, Yes), the output unit 118 outputs a warning indicating that a suspicious person may have entered (step S104). If the updated reliability is greater than or equal to the threshold (No in step S102), the output is not performed by the output unit 118 and the process ends.

  When the reliability is updated in step S100, when the reliability is newly estimated by the reliability estimation unit 108, at least one of the new relevance and the new relevance is estimated by the relevance estimation unit 104. There are three cases in which a certain period of time has elapsed since the last update of reliability. Hereinafter, a process until the reliability is updated when a new reliability is estimated and when a new relationship or the like is estimated will be described.

  FIG. 6 is a flowchart showing processing when a new reliability is estimated. When the new reliability is estimated by the reliability estimation unit 108 (step S200), the related person specifying unit 110 refers to the storage unit 102 and relates to the target person whose reliability is estimated by the reliability estimation unit 108. A person is specified (step S202). Next, the reliability calculation unit 112 calculates the reliability of the related person specified by the related person specifying unit 110 based on the reliability of the target person newly estimated by the reliability estimation unit 108 (step S204). ). Next, the reliability update unit 114 updates the reliability of the target person to the reliability estimated by the reliability estimation unit 108, and further updates the reliability of the related person to the reliability calculated by the reliability calculation unit 112. Update (step S206). This completes the process when a new reliability is estimated.

FIG. 7 is a diagram for explaining processing when a new reliability is estimated. As shown in FIG. 7, when the reliability of the person P ₁ is updated, relevance indicated in the relevant information, that is, the reliability of transmission to knotted related personal link occurs. In the present embodiment, since three links are propagated, the related person specifying unit 110 specifies persons P _{2 to} P ₆ as related persons of the person P ₁ (step S202). The reliability calculation unit 112 calculates the reliability of the associated character _P 2 to P ₆ by (Equation 1) (step S204). Then, the reliability update unit 114 updates the reliability of the target person _{P 1} and related personal _P 2 to P ₆ (step S206).

  In this way, when a new reliability is estimated, the new reliability is always reflected in the reliability of not only the target person but also the other person who is related to the target person. Reliable reliability can be evaluated.

Note that the person P _3, there are two links indirectly tied with the person P ₁ via the link and the person P ₂ tied directly from the person P _1. Thus, when updating the reliability via another person, the reliability is calculated by the following formula.
S _i ′ = S _i + T _ij × (S _j ′ −S _j ) (Expression 3)
Here, T _ij is a propagation rate, and represents the influence from the person P _j to the person P _i . This T _ij is calculated using R _ij or R _ik × R _kj that is a combination of a plurality of relevance levels.

In addition, for a related person having a relationship with the target person through a plurality of links, the shortest path having the number of nodes (related persons) is used for T _ij in (Expression 3). That is, in the example shown in FIG. 7, R ₁₃ is used as the propagation rate T ₁₃ .

For related persons connected to the target person by a plurality of links, the value of any link may be used, and the method for determining the propagation rate T _{ij in} (Equation 3) in this case is described in the embodiment. It is not limited.

Specifically, the higher propagation rate obtained for each link may be used as the propagation rate T _ij . When the relevance R ₁₃ = 0.6, R ₁₂ = 0.8, and R ₂₃ = 0.5, in the direct link from P ₁ to P ₃ , the propagation rate T ₁₃ = R ₁₃ = 0.6 It becomes. On the other hand, in the _{P 2} over the link, the propagation rate _{T 13 = R 12 × R 23} = 0.8 × 0.5 = 0.4. By comparing these, the one with the higher propagation rate, that is, the relevance of the direct link from P ₁ is used as T _ij .

  Furthermore, as another example, for a related person who is connected to a target person by a plurality of links, a new reliability may be calculated in consideration of being connected by a plurality of links.

  FIG. 8 is a flowchart showing a process when a relevance or the like is newly estimated. First, when at least one of a new relevance and relevance degree is estimated by the relevance estimation unit 104 (step S210), the related information update unit 106 estimates the related information stored in the storage unit 102. (Step S212). Next, the related person specifying unit 110 specifies a target person in the related information updated by the related information update unit 106, and further specifies a related person for the target person (step S214). Next, the reliability calculation unit 112 calculates the reliability of the target person and the related person (step S216). Next, the reliability update unit 114 updates the reliability of the target person and the related person to the reliability calculated by the reliability calculation unit 112 (step S218). This completes the processing when a new relevance or the like is estimated.

FIG. 9 is a diagram for explaining processing when new relevance or the like is estimated. As illustrated in FIG. 9, when the degree of association (R ₁₂ ) between the person P ₁ and the person P ₂ is updated, first, the related person specifying unit 110 has relevance in the updated related information. identifying the person _{P 1} and the person _{P 2} as the target person. Further, the related persons of the person P ₁ and the person P ₂ are specified (step S214). Here, the persons P _{3 to} P ₆ are identified as related persons. Next, the reliability calculation unit 112 calculates the reliability of the target person using (Equation 2) based on the new relevance. The reliability calculation unit 112 further calculates the reliability of the related person by (Equation 1) based on the calculated reliability of the target person (step S216). Next, the reliability update unit 114 updates the reliability of the target person and the related person (step S218).

  In this way, when the relevance between certain persons is updated, the reliability of the person who is related to that person also changes accordingly. calculate. Thereby, when the relevance is updated, the reliability of the related person is updated.

Next, a process for updating the reliability will be specifically described. As illustrated in FIG. 10, information such as a person ID is not stored in the storage unit 102 at the start of processing of the suspicious person monitoring system 1. It is assumed that one person enters the monitoring area. In this case, the person registration unit 100 assigns the person ID of P ₁ to this person and registers it in the storage unit 102. At this time, the reliability is set to the initial value 50. As a result, as shown in FIG. 11, the person ID and the reliability of the person P ₁ are registered in the storage unit 102.

Next, it is assumed that one more person enters the monitoring area. In this case, the person registration unit 100, a _{P 2} was assigned as a person ID to the second child of the person is registered in the storage unit 102. Further, when the person P ₁ and the person P ₂ exchange greetings and the relevance estimation unit 104 detects an interaction, the relevance estimation unit 104 has a relevance between the person P ₁ and the person P ₂ , and the degree of relevance thereof. Is estimated to be 0.5. Then, as shown in FIG. 12, the storage unit 102, in association with the person ID of the person _{P 1} and the person _{P 2,} relevance 0.5 are registered with one another person ID. In this case, the reliability calculation unit 112 calculates a new reliability according to (Equation 2). Reliability of the person _{P 1} and the person _{P 2} is obtained by (Equation 4) and (Equation 5).
S ₁ ′ = S ₁ + R ₁₂ × (S ₂ −S ₁ )
= 50 + (50-50) × 0.5 = 50 (Formula 4)
S ₂ ′ = S ₂ + R ₁₂ × (S ₁ −S ₂ )
= 50 + (50-50) × 0.5 = 50 (Formula 5)
That is, all the reliability values are the same values as before the update, and the update is not performed.

Here, the authentication device 4 has been successfully authenticated person P _1. In this case, the reliability estimation unit 108 estimates the reliability of the person P ₁ as 100, and the reliability update unit 114 updates the reliability of the person P ₁ to 100 as shown in FIG. Further, at this time, the reliability of the person P ₂ having relevance to the person P ₁ is also updated by (Equation 1). Reliability of the person _{P 2} is obtained by (Equation 6). Reliability of the person P ₂ is updated to a new value 75 as shown in FIG. 14.
_{S 2 '= S 2 + R} 12 × (S 1' -S 2)
= 50 + (100-50) × 0.5 = 75 (Expression 6)

  In this way, each time a new relevance or the like is estimated or a new reliability is estimated, a new reliability is also calculated for the related person.

As shown in FIG. 15, in the state where the information of five persons is stored in the storage unit 102, when a new person enters the monitoring area, as shown in FIG. The assigned person ID “P ₆ ” and the reliability 50 are newly registered in the storage unit 102. Thereafter, when a certain period of time elapses without estimation of new reliability or new relevance, the reliability update unit 114 stores the reliability stored in the storage unit 102 based on the deviation degree of the movement route. Update the degree. In the present embodiment, the reliability update unit 114 may decrease each reliability stored in the storage unit 102 by a predetermined value every time a certain period elapses.

  In addition, when the reliability is updated after such a fixed period has passed, the reliability update of the related person based on the updated reliability, that is, the update of the reliability update may be performed. Well, you do n’t have to. Here, for convenience of explanation, it is assumed that propagation is not performed.

When the time elapses without detecting the interaction between the person P ₆ and another person and the reliability of the person P ₆ decreases to 30 as shown in FIG. 17, the person P ₆ passes the person P ₄ , Suppose an interaction is detected. In this case, the related information between the person P ₄ and the person P ₆ is updated, and the new reliability of the person P ₄ and the person P ₆ is calculated by (Equation 2) based on the updated related information. Specifically, the new reliability of the person P ₄ and the person P ₆ is obtained by (Expression 7) and (Expression 8).
S ₄ ′ = S ₄ + R ₄₆ × (S ₆ −S ₄ )
= 60 + (30-60) × 0.1 = 57 (Expression 7)
S ₆ ′ = S ₆ + R ₄₆ × (S ₄ −S ₆ )
= 30 + (60-30) × 0.1 = 33 (Expression 8)

Furthermore, the new reliability of the person P ₃ who is the related person of the person P ₄ is calculated by (Equation 1). Specifically, a new confidence of the person P ₃ is obtained by (Equation 9).
_{S 3 '= S 3 + R} 34 × (S 4' -S 4)
= 70 + (57-60) × 0.9 = 67 (Equation 9)
With the above processing, as shown in FIG. 18, the reliability of the person P ₃ and the person P ₄ and the person P ₆ is updated.

Further, as shown in FIG. 19, in a state where the information of the storage unit 102 to 6 persons is stored, the person P ₆ is photographed by the camera 3, and hidden face with sunglasses, reliability below a threshold Assume that 16 is estimated. In this case, the reliability of the person P ₄ having a relationship with the person P ₆ is updated by (Equation 1). New confidence of the person _{P 4} is obtained by (Equation 10).
_{S 4 '= S 4 + R} 46 × (S 6' -S 6)
= 100 + (16-30) × 0.1 = 99 (Equation 10)
Thus, as shown in FIG. 20, the reliability of the person P ₄ is updated.

Further, the reliability of the person P ₃ having a relationship with the person P ₄ is also updated by (Equation 1). New confidence of the person _{P 3} is obtained by (Equation 11).
_{S 3 '= S 3 + R} 46 × (S 4' -S 4)
= 100 + (99-100) × 0.9 = 99 (Expression 11)
Thus, as shown in FIG. 21, the reliability of the person P ₃ is updated. In addition, the reliability of a person having relevant to the person P ₃ is updated.

At this point, since the reliability of the person P ₆ is below the threshold value 20, based on the result of the comparison by the suspicious individual judgment unit 116, the output unit 118 outputs the effect that there is a possibility that the suspicious person intrudes .

  In addition to what has been described above, the evaluation of the detected action may be conditioned, and the reliability may be updated. For example, if a person other than the related person enters a single area, or if the staying time is longer than expected, the suspicious behavior is registered, and if detected, the reliability of the person is changed. You may do it.

  As described above, according to the suspicious person monitoring system 1 according to the embodiment, based on the reliability of the person who has been authenticated by the authentication apparatus, not only the person who has been authenticated but also any person who has been authenticated. Therefore, it is possible to appropriately determine whether an unauthenticated person is a suspicious person.

DESCRIPTION OF SYMBOLS 1 Suspicious person monitoring system 2 Laser distance meter 3 Camera 4 Authentication apparatus 5 Network 10 Suspicious person monitoring apparatus 100 Person registration part 102 Storage part 104 Relevance estimation part 106 Related information update part 108 Reliability estimation part 110 Related person specific part 112 Trust Degree calculation unit 114 Reliability update unit 116 Suspicious person determination unit 118 Output unit

Claims (9)

A storage unit for storing the person identification information for identifying the person in the monitoring area and the reliability of the person in association with each other, and storing the related information indicating the relation between the plurality of persons;
A reliability estimation unit that estimates the reliability of a predetermined target person based on the detection result of the external sensor;
A related person specifying unit for specifying a related person associated with the target person in the related information;
A reliability calculation unit that calculates the reliability of the related person based on the reliability of the target person estimated by the reliability estimation unit;
When the reliability of the target person is estimated by the reliability estimation unit, the reliability of the target person stored in the storage unit is updated to the reliability estimated by the reliability estimation unit. A reliability update unit that updates the reliability of the related person stored in the storage unit to the reliability calculated by the reliability calculation unit;
The reliability associated with the person identification information in the storage unit is compared with a predetermined threshold, and when the reliability is less than the threshold, the person identified by the person identification information is A suspicious person determination unit that determines that the person is a suspicious person
A suspicious person monitoring device, comprising: an output unit that outputs that there is a possibility that a suspicious person has invaded when the suspicious person determination unit determines that the person is a suspicious person.   2. The suspicious person monitoring apparatus according to claim 1, wherein the suspicious person determination unit compares the reliability with the threshold every time the reliability is updated. A relevance estimation unit that estimates relevance between a plurality of persons based on a detection result of an external sensor;
A relevance update unit that updates the related information stored in the storage unit based on an estimation result by the relevance estimation unit when relevance between a plurality of persons is estimated by the relevance estimation unit And further comprising
3. The reliability calculation unit according to claim 1, wherein the reliability calculation unit calculates the reliability of the plurality of persons and the reliability of each of the plurality of persons based on the updated relevance. The suspicious person monitoring device described.   2. The personal information registration apparatus according to claim 1, further comprising: a person registration unit that registers the person identification information for identifying the new person in the storage unit when the external sensor detects a new person in the monitoring area. The suspicious person monitoring device according to any one of items 1 to 3. The related information further includes a degree of association indicating a degree of association between a plurality of persons,
The suspicious person monitoring apparatus according to claim 1, wherein the reliability calculation unit further calculates the reliability based on the relevance between a plurality of persons.   The suspicious person monitoring device according to claim 1, wherein the reliability update unit further updates the reliability according to a time elapsed after the update of the reliability.   The said reliability estimation part estimates the reliability of the said subject person based on the authentication result of the said subject person obtained by the authentication apparatus, The Claim 1 characterized by the above-mentioned. Suspicious person monitoring device.   The said reliability estimation part estimates the reliability of the said subject person based on the movement history of the said subject person obtained by the external sensor, The Claim 1 characterized by the above-mentioned. Suspicious person monitoring device. A suspicious person monitoring method executed by a suspicious person monitoring apparatus,
The suspicious person monitoring device stores person identification information for identifying a person in a monitoring area and the reliability of the person in association with each other, and further stores related information indicating the relation between a plurality of persons. With
A reliability estimation step for estimating the reliability of a predetermined target person based on the detection result of the external sensor;
A related person specifying step of specifying a related person associated with the target person in the related information;
A reliability calculation step of calculating the reliability of the related person based on the reliability of the target person estimated in the reliability estimation step;
When the reliability of the target person is estimated in the reliability estimation step, the reliability of the target person stored in the storage unit is updated to the reliability estimated in the reliability estimation step. A reliability update step of updating the reliability of the related person stored in the storage unit to the reliability calculated in the reliability calculation step;
The reliability associated with the person identification information in the storage unit is compared with a predetermined threshold, and when the reliability is less than the threshold, the person identified by the person identification information is determined. A suspicious person determination step for determining that the person is a suspicious person;
A suspicious person monitoring method, comprising: an output step of outputting that there is a possibility that a suspicious person has invaded when the suspicious person judging unit judges that the person is a suspicious person.

Images