JP5098160B2 - Security device, security system and security program - Google Patents

Description

  The present invention relates to a crime prevention device, a crime prevention system, and a crime prevention program provided with an imaging means for imaging a predetermined imaging area where a management target is provided.

  Conventionally, theft destruction damage to the vending machine has occurred frequently for the purpose of theft of money and goods inside the vending machine. For this reason, a crime prevention sensor is installed that detects abnormal actions from the outside to the vending machine, and sends image information that captures the specified warning or abnormal actions to the monitoring center when there is an abnormal action. Devices and security systems have been proposed (see Patent Documents 1 to 3). In such a crime prevention device and crime prevention system, a predetermined warning is given to prevent theft destruction, and by analyzing image information obtained by imaging abnormal acts, a person who has caused the theft destruction to a vending machine is identified. is doing.

JP 2000-149114 A JP 2004-302605 A JP 2005-71232 A

  However, in crime prevention devices equipped with conventional vibration sensors and the like, theft destruction prevention warning is often recognized as a warning only by theft destruction, especially in a short period of theft destruction. May continue to be stolen and destroyed even when there is a warning. For this reason, the theft destruction to the vending machine cannot be surely prevented, and the vending machine main body may be destroyed. Moreover, in a crime prevention system that analyzes image information obtained by imaging a theft destruction act, the image information is often analyzed after the theft destruction act is actually performed. Therefore, in the conventional crime prevention device and crime prevention system, the vending machine main body is often destroyed, and there is a problem that a lot of cost is required to repair the vending machine main body.

  The present invention has been made in view of the above-described drawbacks of the prior art, and provides a crime prevention device, a crime prevention system, and a crime prevention program for predicting an abnormal operation such as theft destruction of a vending machine and preventing the abnormal operation. With the goal.

To solve the above problems and achieve the object, the security device is in a security device provided with imaging means for imaging a predetermined imaging region provided is managed, the imaging area is unattended according to the present invention Is stored as a first population, and the Mahalanobis distance in the MT method between the image group captured when a general operation is performed on the management target and the first population is a second population. Storage means for storing as a population, first calculation means for calculating a Mahalanobis distance in the MT method between an image captured by the imaging means and the first population as a first distance, and the first a second calculating means for calculating the distance between the Mahalanobis distance in the MT method between the second population as a second distance, the calculated second distance to the separation degree by the second calculating means Based on the separation of the, or the operation of a person in the image pickup area is likely to be general, or, a determination unit configured to determine is likely to be abnormal operation, by the image pickup means A first distance is calculated each time an imaging region is imaged, and it is determined whether to calculate the second distance based on the variation of the calculated first distance. Control means for calculating the second distance only when the fluctuation exceeds a certain value, and the determination means is a group of images taken when a general operation is performed on the management target and the management target If the degree of separation is larger than the threshold value set based on the distribution tendency of the second distance group in the image group in which the abnormal operation is performed, the possibility of abnormal operation is high It is characterized by determining.

  The crime prevention apparatus according to the present invention further includes image processing means for dividing and / or compressing the captured image, and the first calculation means divides and separates image data into a plurality of regions by the image processing means. The first distance is calculated based on the compressed image.

  In addition, the security device according to the present invention, when the determination unit determines that there is a high possibility that the motion of the person in the imaging area is an abnormal operation, and / or the voice for warning the abnormal operation and / or the abnormal operation It further comprises output means for outputting a warning display.

  In addition, the security device according to the present invention notifies the external device that the possibility of the abnormal operation is high when the determination unit determines that the operation of the person in the imaging area is likely to be an abnormal operation. It further comprises external communication means for informing the user.

A crime prevention system according to the present invention is a crime prevention system for managing a predetermined management target, based on the crime prevention device according to any one of claims 1 to 4 and information transmitted from the crime prevention device. And a management device that manages the management object.

Further, the crime prevention program according to the present invention is a crime prevention program for managing a predetermined management object.In a crime prevention program for managing a predetermined management object, an imaging procedure for imaging a predetermined imaging area provided with the management object; An image group when the imaging region is unattended is stored as a first population, and an MT of the image group captured when a general operation is performed on the management target and the first population Storage procedure for storing the Mahalanobis distance in the method as a second population, and a first calculation for calculating the Mahalanobis distance in the MT method between the image captured in the imaging procedure and the first population as the first distance procedures and, a second calculation procedure for calculating the Mahalanobis distance as the second distance in the MT method between the first distance and the second population, the second calculation The second distance calculated in order is set as a separation degree, and based on the separation degree, it is highly likely that a person's movement in the imaging region is general or an abnormal movement. A determination procedure for determining whether or not the image pickup area is high, and a first distance is calculated each time the imaging region is imaged by the imaging procedure , and the second distance is calculated based on the calculated variation in the first distance. A control procedure for determining whether or not to calculate, and calculating the second distance only when a variation in the first distance exceeds a certain value. The degree of separation is greater than the threshold set based on the distribution tendency of the second distance group in the image group in which the case of performing an automatic operation and the image group in which the abnormal operation is performed on the management target. If the value is large, there is a possibility of abnormal operation. The Most and judging.

  According to the security device of the present invention, the degree of separation between a captured image obtained by capturing a person's movement in the imaging region and an image when a general operation is performed on a management target is calculated. In addition, it is possible to predict an abnormal operation such as theft destruction on the management target by determining whether the human operation in the imaging region is likely to be general or likely to be an abnormal operation, Such abnormal operation can be prevented.

(Embodiment)
Hereinafter, a crime prevention system and a crime prevention system for a vending machine will be described as an example of the crime prevention system and the crime prevention system according to an embodiment of the present invention with reference to the drawings. Note that the present invention is not limited to the embodiments. In the description of the drawings, the same parts are denoted by the same reference numerals.

  FIG. 1 is a schematic diagram illustrating an overall configuration of a security system including a security device according to the present embodiment. The crime prevention device according to the present embodiment performs a calculation process using the MT method twice, thereby taking a captured image obtained by capturing a person's movement in the imaging region, and an image when a general operation is performed on a management target. Is calculated, and based on this distance, it is predicted whether or not there is a high possibility that the motion of the person in the imaging region is an abnormal operation. As shown in FIG. 1, the crime prevention system 1 according to the embodiment includes a vending machine 10 that is a management target, a crime prevention device 20 that prevents theft destruction of the vending machine 10, and information transmitted from the crime prevention device 20. And a monitoring center 30 having a management device for managing the vending machine 10 based on the above. The security device 20 and the monitoring center 30 are connected via the network 40. Further, the connection mode may be either wired or wireless, and may be directly connected without going through the network 40.

  Whether the crime prevention apparatus 20 is likely to be a general operation such as purchasing a product based on an image obtained by imaging a predetermined imaging area in which the vending machine 10 is provided. Or, it is determined whether or not there is a high possibility of an abnormal operation such as theft or destruction. The security device 20 issues a predetermined warning when it is determined that there is a high possibility that the action of the person in the imaging area is an abnormal action, and transmits the determination result and the captured image information to the monitoring center 30. .

  Based on the information transmitted from the crime prevention device 20, the monitoring center 30 performs a feature analysis of a person who performs an abnormal operation by reporting to an administrator or analyzing image information. The monitoring center 30 may control the merchandise sales process of the vending machine 10, and a determination result indicating that there is a high possibility that the action of the person in the imaging area is abnormal is transmitted from the security device 20. In this case, the vending machine 10 may be instructed to stop the product sales process.

  Next, the crime prevention device 20 will be described. As shown in FIG. 1, the security device 20 includes a control unit 21, a camera 22, a storage unit 23, an image processing unit 24, a distance calculation unit 25, a determination unit 26, a warning output unit 27, and an external communication unit 28.

  The control unit 21 controls each process or operation of the camera 22, the storage unit 23, the image processing unit 24, the distance calculation unit 25, the determination unit 26, the warning output unit 27, and the external communication unit 28. The control unit 21 performs predetermined input / output control on information input / output to / from each of these components, and performs predetermined information processing on the information.

  The camera 22 images a predetermined imaging area where the vending machine 10 is provided under the control of the control unit 21. The camera 22 outputs image information including the captured image, the imaging order, and the imaging date and time to the control unit 21. The camera 22 continuously captures an imaging area and outputs a series of continuous still images to the control unit 21. In addition, the camera 22 has an imaging region set so that a person who purchases a product of the vending machine 10 enters the image together with the vending machine 10.

  The memory | storage part 23 memorize | stores the various information and program which are required for the processing operation of each structure part. The storage unit 23 stores the first reference space and the second reference space required for the calculation process of the distance calculation unit 25.

  Under the control of the control unit 21, the image processing unit 24 divides and / or compresses an image captured by the camera 22 and capturing an image of a person's movement in the imaging region. The image processing unit 24 divides a series of still images captured by the camera 22 into predetermined regions. In the present embodiment, the case where the image processing unit 24 divides the image into nine regions of 3 divisions in the vertical direction and 3 divisions in the horizontal direction is described, but the present invention is not limited to this. You may divide | segment into 4 area | regions of 2 horizontal divisions, and may divide | segment into 12 area | regions of 4 vertical divisions and 3 horizontal divisions. Further, the divided area may be asymmetric. The image processing unit 24 may determine a region to be divided according to the processing capability of the distance calculation unit 25 that performs calculation processing on the image processed by the image processing unit 24. In addition, the image processing unit 24 deletes a series of still images captured by the camera 22 to delete a frame portion other than the analysis target, monochrome processing to process a full-color image into a monochrome image, and reduces image resolution. Perform compression processing such as mosaic processing. In the monochrome processing, the image processing unit 24 uses, for example, information having 256 levels of color density for each hue of R (red), G (green), and B (blue) to 256 gradations from white to black. Reduced data. Further, the image processing unit 24 uses, for example, data obtained by reducing an image having 176 pixels × 144 pixels to 24 pixels × 18 pixels in the mosaic processing. In addition, the image processing unit 24 may thin out a series of still images to such an extent that a person's movement in the imaging region can be determined, and may perform image division processing and image compression processing on the thinned images.

The distance calculation unit 25 calculates the degree of separation between an image captured by the camera 22 and captured of a person's movement in the imaging area, and an image when a general operation is performed on the vending machine 10. To do. The distance calculation unit 25 calculates the Mahalanobis distance (MD) obtained by performing the calculation process using the MT method twice as the distance. The distance calculation unit 25 calculates the MD value between the image processed by the image processing unit 24 and a predetermined population. The MD value D is a distance measure between the reference space as the population and the image information as the sample. In other words, the distance between the reference space as the population and the image as the sample is quantified. The MD value D is obtained by using the following equation (1).
D ² = (x−μ) ^t Σ ⁻¹ (x−μ) (1)
In the above equation (1), x is a vertical vector of sample values, μ is an average vertical vector of a population that is a reference space, t is a transposed matrix, and Σ ⁻¹ is a population vector It is the inverse of the variance / covariance matrix. Here, in the MT method, when the value of D is in the vicinity of 1, a sample having this MD value is highly likely to belong to the population. In the present embodiment, the distance calculation unit 25 includes a first calculation unit 25a that performs the first MT process and a second calculation unit 25b that performs the second MT process, and performs the calculation process using the MT method twice. Is going.

  As the first MT process, the first calculation unit 25a uses a large number of image groups when the imaging region is unattended as a first reference space that is a first population, and uses the MT method to perform the first MT. The MD value between the reference space and the image processed by the image processing unit 24 is calculated as the first MD value that is the first distance. The first calculator 25a calculates a first MD value for each image. In addition, when the image processing unit 24 has divided the image, the first calculation unit 25a obtains a first MD value for each divided region. In addition, the crime prevention apparatus 20 acquires many still images in case an imaging area is unmanned.

  As the second MT process, the second calculation unit 25b uses the first MD value group in a series of still images obtained by capturing a large number of cases where a general operation such as product purchase is performed on the vending machine 10 as the second population. Using the MT method, the MD value between the second reference space and the first MD value obtained by the first calculator 25a is calculated as the second MD value that is the second distance. In the second MT process, all the first MD values in a series of still images for one operation, that is, all the first MD values calculated for one operation are the targets of the second MT process. The distance calculation unit 25b calculates the second MD value, so that the first MD value calculated by the first calculation unit 25a in the first MT process is included in the first MD value group in many general operations as the second reference space. It quantifies how far away. The first MD value calculated in the first MT process by the first calculation unit 25a is calculated based on an image obtained by capturing the motion of a person in the imaging area of the camera 22, and therefore the second MD value is calculated in the imaging area. It can be said that the figure shows how far a person's movement is far from many general movements. In other words, the second MD value corresponds to the degree of separation between the captured image obtained by capturing the motion of the person in the imaging region and the image when the general operation is performed on the vending machine 10. The crime prevention device 20 acquires in advance a series of still images obtained by capturing a large number of cases where a person performs a general operation such as purchasing a product with respect to the vending machine 10, and the first MD for the first reference space is obtained. A second reference space is obtained by calculating each value. In the second MT processing, the first MD value in a series of still images for one motion, that is, all the first MD values calculated for one motion are subject to the processing of the second MT processing. Will be included.

  Based on the degree of separation calculated by the distance calculation unit 25, the determination unit 26 has a high possibility that the person's movement in the imaging area is a general movement or an abnormal movement. Determine. The determination unit 26 uses the second MD value calculated by the second calculation unit 25b as the degree of separation, and it is highly possible that the action of the person in the imaging region is a general action or an abnormal action. To determine if it is high. The second MD value is obtained by quantifying how far the first MD value calculated by the first calculation unit 25a in the first MT process is different from the first MD value group in many general operations which are the second reference space. It is. In other words, it can be said that the distance between the movement of the person in the imaging region and the general movement is quantified. Here, in the MT method, when the MD value is in the vicinity of 1, a sample having this MD value is highly likely to belong to the population. According to the basic principle of the MT method, when the second MD value is close to 1, the first MD value in the motion of the person imaged by the camera 22 is the second reference space and the general motion in the population. The possibility of belonging to the first MD value group is high. For this reason, the determination unit 26 predicts that a person within the imaging range of the camera 22 performs a general operation when the second MD value is near 1. As a result, the determination unit 26 performs normal determination indicating that a person within the imaging range of the camera 22 is likely to perform a general operation such as purchase of a product. On the other hand, when the second MD value is a large value based on the basic principle of the MT method, the first MD value in the motion of the person imaged by the camera 22 is a general motion that is a second reference space and a population. There is a high possibility of not belonging to the first MD value group. For this reason, when the second MD value is larger than 1, for example, 5 or more, the determination unit 26 does not belong to the general operation in the imaging range of the camera 22, and the person in the imaging range. Predict that they will perform abnormal operations such as theft destruction. As a result, the determination unit 26 performs an abnormality determination indicating that a person within the imaging range of the camera 22 is likely to perform an abnormal operation. Note that the threshold value of the second MD value that is a basis for the determination of the general motion and the abnormal motion by the determination unit 26 is the second MD based on a series of still images obtained by imaging a large number of general motions and a large number of abnormal motions. Each of the values may be acquired in advance and set based on the distribution tendency of the second MD value in the general operation and the abnormal operation.

  When the determination unit 26 performs an abnormality determination that the person in the imaging range of the camera 22 predicts that there is a high possibility of performing an abnormal operation, the warning output unit 27 is a voice for warning the abnormal operation and / or the abnormality. Output a warning to alert the operation. Specifically, the warning output unit 27 may be realized using a microphone and a speaker and may output sound. The warning output unit 27 is realized by using a liquid crystal display (LCD), an organic electroluminescence display, and a light emitting diode (LED) that continues to emit light while a current is supplied. A display may be output.

  The external communication unit 28 is implemented using an external communication interface, and performs information communication with the monitoring center 30 and the like. When the determination unit 26 performs an abnormality determination indicating that there is a high possibility that the motion of the person in the imaging region is an abnormal operation, the external communication unit 28 indicates that the motion of the person in the imaging region is an abnormal operation. The monitoring center 30 is notified that the possibility is high, and information including an image captured by the camera 22 is transmitted to the monitoring center 30.

  Next, an operation process of the security device 20 will be described. FIG. 2 is a flowchart for explaining the operation process of the security device 20. As shown in FIG. 2, first, the camera 22 continuously captures still images in a predetermined imaging area provided with the vending machine 10 under the control of the control unit 21, and includes the captured still images. Imaging processing for sequentially outputting the image information to the control unit 21 is performed (step S102). Under the control of the control unit 21, the image processing unit 24 performs image division processing for dividing the image captured by the camera 22 into nine regions (step S104). Thereafter, the image processing unit 24 performs an image compression process for compressing the information amount of the image by performing a monochrome process and a mosaic process on each of the divided areas of the image (step S106).

  Next, the distance calculation unit 25 performs MD value calculation processing. First, the first calculator 25a performs a first MT process using the MT method (step S108). In the first MT process, the first calculation unit 25a obtains a first MD value between the first reference space and the image processed by the image processing unit 24. In this case, the security device 20 acquires in advance a large number of images when the imaging region is unattended. Then, the security device 20 divides these images into nine regions as in the image division process (step S104), and then performs monochrome processing and mosaic processing as in the image compression process (step S106). The security device 20 stores these images in the storage unit 23 as the first reference space in the first MT process. In the first MT processing, the first calculation unit 25a refers to the first reference space stored in the storage unit 23, and obtains the first MD value of each image with respect to the first reference space. In the first MT process, the first MD value is obtained by using the image information of each area of the image subjected to the image compression process (step S106) and the image information group of the same area in many images as the first reference space. Calculate

  Thereafter, the second calculator 25b performs the second MT process using the MT method (step S110). In the second MT process, the second calculator 25b calculates a second MD value in the second reference space and the first MD value calculated in the first calculator 25a. In this case, the crime prevention apparatus 20 captures a large number of cases where a person performs a general operation such as purchasing a product with respect to the vending machine 10 in advance, and acquires a large number of still images in the general operation. Then, the security device 20 divides these images into nine regions as in the image division process (step S104), and then performs monochrome processing and mosaic processing as in the image compression process (step S106). Furthermore, the security device 20 calculates | requires the 1st MD value with respect to 1st reference space about these images, respectively. And the crime prevention apparatus 20 memorize | stores these each 1st MD values in the memory | storage part 23 as 2nd reference space. That is, the second reference space is a first MD value group in many general operations. The second calculation unit 25b refers to the second reference space stored in the storage unit 23 in the second MT process, and the second MD value in the second reference space and the first MD value calculated in the first MT process. Ask for. In the first MT process in step S108 and the second MT process in step S110, the distance calculation unit 25 automatically captures an image captured in the imaging process (step S102) and images a person's action in the imaging region. The second MD value, which is the distance from the image when a general operation is performed on the vending machine 10, is calculated.

  Thereafter, the determination unit 26 has a high possibility that the action of the person in the imaging region of the camera 22 is a general action based on the second MD value calculated by the second calculator 25b, or is an abnormal action. A determination process is performed to determine whether there is a high possibility (step S112). When the second MD value is close to 1, the determination unit 26 indicates that a person within the imaging range of the camera 22 is likely to perform a general operation such as purchasing a product on the vending machine 10. Make a decision. On the other hand, when the second MD value is greater than 1, for example, 5 or more, the determination unit 26 performs abnormality determination indicating that a person within the imaging range of the camera 22 is likely to perform an abnormal operation.

  Next, the control unit 21 determines whether the determination result output from the determination unit 26 is normal determination or abnormality determination (step S114). When the control unit 21 determines that the determination result output from the determination unit 26 is normal determination (step S114: normal determination), the control unit 21 proceeds to step S102, instructs the camera 22 to perform imaging processing, and performs image processing. The unit 24, the distance calculation unit 25, and the determination unit 26 are instructed to perform the processing from step S104 to step S112 on the still image captured by the camera 22.

  On the other hand, when the control unit 21 determines that the determination result output from the determination unit 26 is an abnormality determination (step S114: abnormality determination), there is a high possibility that a person within the imaging range performs an abnormal operation. For this reason, the control unit 21 causes the warning output unit 27 to output a sound for warning of abnormal operation and a display for warning of abnormal operation, and the warning output unit 27 displays a sound for warning of abnormal operation and a display for warning of abnormal operation. The warning process to output is performed (step S116). Then, under the control of the control unit 21, the external communication unit 28 informs that there is a high possibility that the motion of the person in the imaging area is an abnormal operation, and includes information including an image obtained by capturing the predicted abnormal operation Is transmitted to the monitoring center 30 (step S118). The monitoring center 30 notifies the administrator that there is a high possibility that an abnormal operation will be performed using the transmitted information, and analyzes the characteristics of the person who performs the abnormal operation based on the transmitted information.

  As described above, in the crime prevention device 20 and the crime prevention system 1 according to the present embodiment, by performing the MT process twice, the action of the person with respect to the vending machine 10 may be a general operation such as product purchase. It is possible to predict whether or not there is a high possibility of an abnormal operation such as a theft destruction operation. As a result, the security device 20 and the security system 1 perform the warning process and the abnormality notification process before the abnormal operation such as theft destruction operation on the main body of the vending machine 10 is performed. It can be prevented from being destroyed.

  Next, each processing procedure described in FIG. 2 will be specifically described. First, the image division process shown in step S104 and the image compression process shown in step S106 in FIG. 2 will be described. 3 and 4 are diagrams for explaining the image division processing shown in FIG. 2, and FIG. 5 is a diagram for explaining the image compression processing shown in FIG. FIG. 3 is a diagram illustrating an example of an image captured by the camera 22. As shown in an image F <b> 1 in FIG. 3, the camera 22 images a peripheral area including the vending machine 10. The image F <b> 1 shows a case where the person H is in the imaging range of the camera 22. For example, the camera 22 sequentially captures full-color images having a resolution of 177 pixels × 144 pixels, that is, a total of 25488 pixels. The image processing unit 24 divides the image F1 into nine areas A to I as an image Fc1 shown in FIG. 4 as the image division processing in step S104 of FIG. After that, the image processing unit 24 performs monochrome processing and mosaic processing for each of the areas A to I shown in the image Fc1 of FIG. 4 as the image compression processing in step S106 of FIG. 2, for example, an image Fm1 as shown in FIG. get. This image Fm1 has 256 gradations from white to black and is 24 pixels × 18 pixels. This image Fm1 is an image to be processed in the first MT process in the next step S108. As described above, the crime prevention apparatus 20 performs the drawing division process and the drawing compression process to reduce the information amount of the image to be processed, so that the first MT process can be performed even if the entire image has a high resolution. .

  Next, the first MT process shown in step S108 of FIG. 2 will be described. First, the first reference space in the first MT process will be described. 6 to 8 are diagrams illustrating the first reference space in the first MT process. The first reference space in the first MT process includes a large number of images when there is no person in the peripheral area including the vending machine 10 to be managed as in the image F0 shown in FIG. It becomes. As shown in FIG. 7, the security device 20 performs the same division processing as the image division processing shown in step S104 of FIG. 2 on the image F0 in such an unmanned state, and is the same as the image Fc1 shown in FIG. An image Fc0 divided into areas A to I of FIG. Then, as shown in FIG. 8, the security device 20 performs a compression process similar to the drawing compression process shown in step S106 of FIG. 2, and acquires an image Fm0 having the same amount of information as the image Fm1 shown in FIG. The security device 20 sets the image Fm0 as an image constituting the first reference space in the first MT process. The security device 20 acquires a large number of such unmanned images in advance, and sets the image group obtained by performing the imaging process, the image dividing process, and the image compression process on each of the acquired images as a first reference. It memorize | stores in the memory | storage part 23 as space. For example, the security device 20 acquires about 3000 images that are unattended as a reference space.

  Further, the first MT process will be described with reference to FIG. In FIG. 9, the image Fm0 shown in FIG. 8 is shown as an example of an image constituting the first reference space, and a case where the first MT process is performed on the images Fm1 to Fm4 will be described as an example. It is assumed that images are captured by the camera 22 in the order of images Fm1, Fm2, Fm3, and Fm4.

  As illustrated in FIG. 9, the first calculation unit 25a acquires the first MD value for the first reference space including the image Fm0 by using the equation (1) in the first MT process. The first calculator 25a calculates a first MD value for the reference space for each of the regions A to I in the first MT process. As a result, for example, as indicated by an arrow Y1 in FIG. 9, MD1 (1) that is the first MD value in the region A of the image Fm1 is obtained as 41.2. Similarly, for the image Fm2, MD1 (2), which is the first MD value in the region A of the image Fm2, is obtained as 39.6, as indicated by the arrow Y2 in FIG. Also for the image Fm3, as indicated by the arrow Y3 in FIG. 9, MD1 (3), which is the first MD value in the region A, is obtained as 39.2, and for the image Fm4, as indicated by the arrow Y4, MD1 (4), which is the first MD value in the area A, is obtained as 35.0. As described above, the first calculation unit 25a sequentially performs the first MT processing on each region of the image processed by the image processing unit 24 to obtain the first MD value.

  As a result, a calculation result of the first MD value as shown in FIGS. 10 and 11 is obtained. In FIG. 10, for example, the first MD value in each of the areas A to I when a series of captured still images are thinned out to a rate of two frames per second and image division processing, image compression processing, and first MT processing are performed. Indicates. FIG. 10 shows the first MD value of 11 images with an imaging time of 5 seconds. FIG. 11 shows the imaging order dependence of the first MD values in the areas A to I shown in FIG. In FIG. 11, after 3 seconds from the start of imaging, that is, after the imaging order corresponding to the straight line lm is “6”, the person enters the imaging area and changes from the unmanned state to the manned state. As shown in FIG. 11, the first MD value that is in the vicinity of 1 greatly fluctuates to at least 10 or more, with the straight line lm from the unmanned state to the manned state as a boundary. Thus, it can be determined that the imaging region is in a manned state when the first MD value varies greatly. Note that the control unit 21 may determine whether or not to perform the second MT process based on a change in the first MD value in the first MT process. That is, when the first MD value in the first MT process greatly fluctuates, it is determined that the imaging region is in a manned state, the second MT process is performed, and when the first MD value fluctuates in the first MT process is small, It may be determined that the imaging region is in an unattended state, and the process may proceed to the imaging process shown in FIG. 2 without performing the second MT process.

  Next, the second MT process shown in step S110 of FIG. 2 will be described. In the second MT process, the first MD value group when a person performs a general operation such as purchasing a product with respect to the vending machine 10 is set as the second reference space, and the second reference space and the first calculation unit 25a A second MD value in the calculated first MD value is calculated.

  First, the second reference space in the second MT process will be described. The second reference space in the second MT process is a first MD value group when a person performs a general operation on the vending machine 10. Here, the general operation of a person in the vending machine 10 will be specifically described. FIGS. 12 to 14 show the imaging order dependence of the first MD values in a series of still images taken when a person who has entered the imaging area purchases a product that is a general operation for the vending machine 10. FIG. FIGS. 12 to 14 show cases in which different persons purchase products, and a series of still images captured for 5 seconds from when a person enters the imaging area for purchasing products are also shown. And Curve 1B1 in FIG. 12, curve 1B2 in FIG. 13, and curve 1B3 in FIG. 14 show the imaging order dependence of the first MD value in the region B of the image for different persons. As shown by the curve 1B1, the curve 1B2, and the curve 1B3, all of the first MD values greatly increase after the imaging order 5 and return to the values at the start of imaging in the vicinity of the imaging order 10 and are largely similar. Further, when the corresponding curves of the regions A and C to I are compared for each of the same regions, they tend to be similar for each region. That is, even when different persons purchase products, the dependence of the first MD value of each region on the imaging order tends to be similar. For this reason, when a general operation such as product purchase is performed, it is considered that the imaging order dependence of the first MD value is not significantly different from each other.

  On the other hand, with reference to FIGS. 15-17, the case where each different person performs the theft destruction act which is abnormal operation | movement is demonstrated. FIGS. 15 to 17 are diagrams illustrating the imaging order dependence of the first MD values in a series of still images captured when different persons perform a theft destruction operation that is an abnormal operation. In this case, in FIGS. 15 to 17, even if the first MD value of the same region is dependent on the imaging order, the imaging order dependence is greatly different from each other. Further, as shown in FIGS. 15 to 17, there is no point similar to the imaging order dependence of the first MD value corresponding to the product purchase operation shown in FIGS. 12 to 14. In other words, it is considered that the imaging order dependence of the first MD value when the theft destruction operation is performed is significantly different from the imaging order dependence of the first MD value when a general operation such as product purchase is performed.

  Therefore, if the first MD value calculated by the first calculation unit 25a is not significantly different from the population of the first MD value corresponding to the general action, the calculated first MD value may belong to the general action. Is considered high. On the other hand, when the first MD value calculated by the first calculation unit 25a is far from the population of the first MD value corresponding to the general motion, the calculated first MD value belongs to the general motion. Therefore, the possibility of belonging to an abnormal operation including theft destruction operation is high. For this reason, the crime prevention apparatus 20 obtains a large number of first MD values corresponding to general operations, and sets the first MD value group as a second reference space that is a reference space in the second MT process. Next, the second calculator 25b performs the second MT process on the first MD value calculated in the first MT process, and is the distance between the second reference space and the first MD value calculated in the first MT process. A second MD value is obtained. Then, the determination unit 26 may determine whether the person's motion captured by the camera 22 belongs to an abnormal operation such as a theft destruction operation, depending on how far the second reference space is based on the second MD value. It can be determined whether or not is high.

  Actually, the calculation result of the second MD value when the general operation is performed will be described. FIG. 18 is a diagram illustrating the second MT process. The first MD group Da in FIG. 18 is a second reference space in the second MT processing, and is obtained by performing image division processing, image compression processing, and first MT processing on a series of still images obtained by imaging a general operation many times. It is acquired in advance. The first MD value group Dn is a processing target in the second MT process, and is obtained by performing an image division process, an image compression process, and a first MT process on a series of still images obtained by capturing a product purchase operation in the vending machine 10. The first MD value obtained. The first MD group Dn is a first MD value in all the images that have been subjected to the first MT processing, among a series of still images obtained by capturing a product purchase operation. In other words, all the first MD values calculated for one operation are the processing targets of the second MT process. The second calculator 25b calculates the MD values of the first MD group Da and the first MD group Dn that are the second reference space, that is, the second MD value. As a result, MD2 (n) which is the second MD value in the first MD group Dn is 1.2. Thus, the 2nd calculation part 25b calculates a 2nd MD value for every operation | movement.

  Next, the determination process shown in FIG. 2 will be described. FIG. 19 is a table showing the number of data for each value of the second MD value with respect to a general operation, and FIG. 20 is a graph showing the result of FIG. 19, showing the dependence of the number of distribution data on the second MD value. It is. As shown in FIGS. 19 and 20, in a general operation, the data calculated in the second MT process is distributed almost in the vicinity of one. In the MT method, when the MD value is in the vicinity of 1, a sample having this MD value is highly likely to belong to the population. That is, in the general operation, the second MD value is considered to be concentrated in the vicinity of 1 because it is not greatly separated from the first MD group of many general operations that are the basis of the second reference space. For this reason, when the second MD value is near 1, it is considered that the operation corresponding to the second MD value is likely to belong to a general operation.

  FIG. 21 shows the result of calculating the second MD value by actually performing a general operation such as merchandise purchase and an abnormal operation such as theft destruction operation. As shown in FIG. 21, when the general operations 1 to 3 are performed, the second MD values are 0.96, 1.03, and 1.37. Thus, when the general operation is performed, the second MD value is concentrated in the vicinity of 1. On the other hand, when the abnormal operations 1 to 3 such as the theft destruction operation are performed, the second MD values are 5.82, 6.84, and 21.72. Thus, when an abnormal operation is performed, it is considered that the second MD value tends to show a value clearly larger than 1, for example, a value larger than 5. By using such a tendency, the determination unit 26 determines whether the second MD value calculated by the second calculation unit 25b is near 1 or 5 or more, so that the person in the imaging region can be identified. It is determined whether there is a high possibility of performing a general operation or a high possibility of performing an abnormal operation. When the second MD value is close to 1, the determination unit 26 predicts that a person within the imaging range of the camera 22 performs a general operation, and the person within the imaging range of the camera 22 generally purchases a product. A normal determination is made to indicate that there is a high possibility of performing a target action. On the other hand, when the second MD value is 5 or more, the determination unit 26 predicts that the person in the imaging range of the camera 22 performs an abnormal operation such as a theft destruction operation, and the person in the imaging range of the camera 22 is abnormal. An abnormality determination is performed indicating that there is a high possibility of performing the operation. As a result, the crime prevention apparatus 20 outputs the predetermined warning sound and the predetermined warning display before the person in the imaging range performs the theft destruction when the determination unit 26 performs the abnormality determination. Completion of abnormal operations such as the above can be prevented, and destruction of the vending machine 10 can be prevented. In addition, the security device 20 can notify the administrator that there is a person who performs an abnormal operation by the determination unit 26 notifying the monitoring center 30 of the determination result, and the abnormal operation can be performed before the abnormal operation is completed. In some cases, the characteristics of the person to be performed can be analyzed.

  Thus, according to the crime prevention system 1 provided with the crime prevention device 20 according to the present embodiment, it is predicted that there is a high possibility that the action of the person in the imaging area of the camera 22 is an abnormal operation such as theft destruction. Since warning processing or the like can be performed, the completion of abnormal operation can be prevented as compared with a conventional security device. As a result, the crime prevention system 1 can prevent the vending machine 10 to be managed from being destroyed and can appropriately manage the vending machine 10 as compared with the conventional system.

  In addition, the security device 20 and the security system 1 according to the present embodiment obtain the first MD value of the image captured by the camera 22 and confirm that the imaging region is in a manned state based on whether or not the first MD value varies. Can be recognized. For this reason, since the crime prevention device 20 and the crime prevention system 1 according to the present embodiment have a configuration in which an abnormal operation detection sensor such as a vibration sensor is deleted as compared with the conventional crime prevention device and the crime prevention system, it is simpler than before. It can be configured.

  In the present embodiment, a case has been described in which it is determined whether there is a high possibility that the action of a person in the imaging region is an abnormal operation, and a warning process or the like is performed. However, the present invention is not limited to this. For example, the crime prevention apparatus 20 acquires a large number of still images for the operation when he / she is unsure about product purchase as the second reference space, and previously acquires the first MD value group in these images. And the 2nd calculation part 25b calculates | requires 2nd MD value with respect to the 2nd reference | standard space corresponding to the operation | movement in the case of being uncertain about this goods purchase in 2nd MT process. Next, when the second MD value is close to 1, the determination unit 26 determines that there is a high possibility that a person in the imaging region is at a loss in purchasing the product, and based on this determination, the warning output unit 27 May output a voice for introducing a product such as a recommended product.

  Further, in the present embodiment, the case where the crime prevention device 20 includes the image processing unit 24, the distance calculation unit 25, and the determination unit 26 has been described. However, the present invention is not limited to this, and the management device in the monitoring center 30 is the image processing unit 24. The distance calculation unit 25 and the determination unit 26 may be provided, and it may be determined whether or not there is a high possibility that the motion of the person in the imaging region is an abnormal operation. In this case, the control unit 21 in the security device 20 transmits a series of still images captured by the camera 22 to the monitoring center 30 as needed, and performs warning processing in the warning output unit 27 based on instructions from the management device in the monitoring center 30. You may go.

  Moreover, although this embodiment demonstrated the case where the management object was the vending machine 10, it is not restricted to this, It applies to the crime prevention apparatus and crime prevention system generally which prevent money, theft of goods, or destruction of an apparatus main body. It is possible. For example, you may apply to the crime prevention apparatus which prevents destruction of the automatic teller machine main body, or the theft of money in an automatic teller machine.

  The security system 1 and the security device 20 described in the above embodiment can be realized by executing a program prepared in advance on a computer system such as a personal computer or a workstation. Hereinafter, a computer system that executes a program having the same function as the processing operation in the crime prevention system described in the above embodiment will be described.

  FIG. 22 is a system configuration diagram showing the configuration of the computer system using the above-described embodiment, and FIG. 23 is a block diagram showing the configuration of the main body in this computer system. As shown in FIG. 22, a computer system 100 according to the present embodiment includes a main body 101, a display 102 for displaying information such as an image on a display screen 102a according to an instruction from the main body 101, and the computer system. A keyboard 103 for inputting various information to 100 and a mouse 104 for designating an arbitrary position on the display screen 102a of the display 102 are provided.

  Further, as shown in FIG. 10, a main body 101 in the computer system 100 includes a CPU 121, a RAM 122, a ROM 123, a hard disk drive (HDD) 124, a CD-ROM drive 125 that receives a CD-ROM 109, and a flexible disk. An FD drive 126 that accepts an (FD) 108; an I / O interface 127 that connects the display 102, keyboard 103, and mouse 104; and a LAN interface 128 that connects to a local area network or wide area network (LAN / WAN) 106 Prepare.

  Further, a modem 105 for connecting to a public line 107 such as the Internet is connected to the computer system 100, and another computer system (PC) 111 and server 112 are connected via a LAN interface 128 and a LAN / WAN 106. The printer 113 is connected.

  And this computer system 100 implement | achieves a crime prevention system by reading and running the program recorded on the predetermined recording medium. Here, the predetermined recording medium is not limited to “portable physical medium” such as flexible disk (FD) 108, CD-ROM 109, MO disk, DVD disk, magneto-optical disk, IC card, etc. Connected to internal and external hard disk drives (HDD) 124, “fixed physical media” such as RAM 122 and ROM 123, public line 107 connected via modem 105, other computer system 111 and server 112 Any recording medium that records a program that can be read by the computer system 100, such as a “communication medium” that holds the program in a short period of time when the program is transmitted, such as a LAN / WAN 106 that is used.

  That is, the program is recorded on a recording medium such as the above-mentioned “portable physical medium”, “fixed physical medium”, “communication medium” and the like so that the computer system 100 can read the program. A security device and a security system are realized by reading and executing the program from such a recording medium. Note that the program is not limited to being executed by the computer system 100, and when the other computer system 111 or the server 112 executes the program, or when they cooperate to execute the program. In addition, the present invention can be similarly applied.

It is a mimetic diagram showing the whole crime prevention system composition in an embodiment. It is a flowchart explaining the operation | movement process of the crime prevention apparatus shown in FIG. It is a figure explaining the image division process shown in FIG. It is a figure explaining the image division process shown in FIG. It is a figure explaining the image compression process shown in FIG. It is a figure explaining the 1st reference space in the 1st MT processing shown in FIG. It is a figure explaining the 1st reference space in the 1st MT processing shown in FIG. It is a figure explaining the 1st reference space in the 1st MT processing shown in FIG. It is a figure explaining the 1st MT process shown in FIG. It is a figure which illustrates the calculation result of the 1st MD value in the 1st MT processing shown in FIG. It is a figure which shows the imaging order dependence of the 1st MD value in each area | region AI shown in FIG. It is a figure which shows the imaging order dependence of the 1st MD value in the image imaged when the person who entered the imaging region performed general operation | movement with respect to the vending machine. It is a figure which shows the imaging order dependence of the 1st MD value in the image imaged when the person who entered the imaging region performed general operation | movement with respect to the vending machine. It is a figure which shows the imaging order dependence of the 1st MD value in the image imaged when the person who entered the imaging region performed general operation | movement with respect to the vending machine. It is a figure which shows the imaging order dependence of the 1st MD value in the image imaged when the person who entered the imaging region performed abnormal operation | movement with respect to a vending machine. It is a figure which shows the imaging order dependence of the 1st MD value in the image imaged when the person who entered the imaging region performed abnormal operation | movement with respect to a vending machine. It is a figure which shows the imaging order dependence of the 1st MD value in the image imaged when the person who entered the imaging region performed abnormal operation | movement with respect to a vending machine. It is a figure explaining the 2nd MT processing shown in FIG. It is a figure which shows the data number with respect to each value of the 2nd MD value with respect to general operation | movement. It is a figure which shows 2nd MD value dependence of the number of data shown in FIG. It is a figure which shows the 2nd MD value at the time of performing general operation | movement and abnormal operation | movement. It is a block diagram which shows the structure of the computer system using embodiment. It is a block diagram which shows the structure of the main-body part in the computer system shown in FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Security system 10 Vending machine 20 Security device 21 Control part 22 Camera 23 Memory | storage part 24 Image processing part 25 Distance calculation part 25a 1st calculation part 25b 2nd calculation part 26 Judgment part 27 Warning output part 28 External communication part 30 Monitoring center DESCRIPTION OF SYMBOLS 100 Computer system 101 Main body part 102 Display 102a Display screen 103 Keyboard 104 Mouse 105 Modem 106 Local area network or wide area network (LAN / WAN)
107 Public line 108 Flexible disk (FD)
109 CD-ROM
111 Other computer systems (PC)
112 server 113 printer 121 CPU
122 RAM
123 ROM
124 hard disk drive (HDD)
125 CD-ROM drive 126 FD drive 127 I / O interface 128 LAN interface

Claims (6)

In a crime prevention apparatus provided with an imaging means for imaging a predetermined imaging area provided with a management target,
An image method in which an imaging region is unattended is stored as a first population, and an MT method is performed between an image group in which a general operation is performed on the management target and the first population. Storage means for storing the Mahalanobis distance in as a second population;
First calculation means for calculating a Mahalanobis distance in an MT method between an image captured by the imaging means and the first population as a first distance;
Second calculating means for calculating the Mahalanobis distance in the MT method between the first distance and the second population as a second distance;
The second distance calculated by the second calculation means is set as a separation degree, and based on the separation degree, it is highly likely that a person's movement in the imaging region is general or abnormal. A determination means for determining whether or not there is a high possibility of an action;
A first distance is calculated each time the imaging area is imaged by the imaging means, and it is determined whether to calculate the second distance based on the variation of the calculated first distance, Control means for calculating the second distance only when a variation in the first distance exceeds a certain value ,
The determination means is based on the distribution tendency of the second distance group in the image group obtained by imaging the case where the general operation is performed on the management target and the image group obtained when the abnormal operation is performed on the management target. A crime prevention device characterized by determining that the possibility of an abnormal operation is high when the degree of separation is greater than a set threshold value. Further comprising image processing means for dividing and / or compressing the captured image;
2. The first calculation unit according to claim 1, wherein the first calculation unit calculates the first distance based on an image obtained by dividing and / or compressing image data in a plurality of regions by the image processing unit. Security device as described in 4.   An output means for outputting a sound for warning of the abnormal action and / or a display for warning of the abnormal action when the judging means determines that there is a high possibility that the action of the person in the imaging region is an abnormal action; The crime prevention device according to claim 1, wherein the security device is provided.   When the determination unit determines that the action of the person in the imaging area is likely to be an abnormal operation, the communication unit further includes an external communication unit that notifies an external device that the possibility of the abnormal operation is high The crime prevention device according to any one of claims 1 to 3, wherein In a crime prevention system that manages a predetermined management target,
The crime prevention device according to any one of claims 1 to 4,
A management device for managing the management object based on information transmitted from the security device;
Security system characterized by having In a crime prevention program that manages a predetermined management target,
An imaging procedure for imaging a predetermined imaging area in which the management object is provided;
An image group when the imaging region is unattended is stored as a first population, and an MT of the image group captured when a general operation is performed on the management target and the first population A memory procedure for storing the Mahalanobis distance in the law as a second population;
A first calculation procedure for calculating the Mahalanobis distance in the MT method between the image captured in the imaging procedure and the first population as a first distance;
A second calculation procedure for calculating the Mahalanobis distance in the MT method between the first distance and the second population as a second distance;
The second distance calculated in the second calculation procedure is set as a separation degree, and based on the separation degree, it is highly likely that a person's movement in the imaging region is general or abnormal. A determination procedure for determining whether or not there is a high possibility of an action,
Calculating the first distance each time the imaging region is imaged by the imaging procedure, determining whether to calculate the second distance based on the variation of the calculated first distance, A control procedure for calculating the second distance only when a variation in the first distance exceeds a certain value ,
The determination procedure is based on a distribution tendency of a second distance group in an image group obtained by imaging a case where a general operation is performed on the management target and an image group obtained by imaging an abnormal operation on the management target. A crime prevention program characterized by determining that there is a high possibility of abnormal operation when the degree of separation is larger than a set threshold value.