JP2012103809A - Composite-type sensor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve a problem that accuracy of anomaly determination is reduced by inaccurate positional relationship between a sensed area by an image sensor and that by a human body sensor which sets a part of monitored space as sensed space, in a composite-type sensor that includes both sensors and senses an invasion anomaly by determining a result of sensing by the sensors in an integrated manner.SOLUTION: A composite-type sensor executes: a step for conducting a walk test through movement of a worker within monitored space to acquire measurement data resulting from associating a human image position within an image at the time of measurement with a signal level of a sensing signal of a space sensor part; a step for selecting the measurement data corresponding to a peak value Pmax as a representative point at a characteristic 150 given by the measurement data; a step for correcting a human image position Xp according to a moving speed of the worker to obtain a correction peak position Xpc; a step for setting a composite sensed area so as to center the correction peak position Xpc; and a step for specifying a remaining image as an area solely sensed by an image sensor.

Description

  The present invention relates to an image monitoring sensor that includes an image sensor and a human body sensor that detects a human body in a space, and monitors the presence or absence of an intruder in the monitoring space.

  As a type of sensor that monitors an intruder into a surveillance space, a composite sensor that uses a camera that captures the surveillance space and a passive infrared ray (PIR) sensor is known (Patent Document 1 below). ). In this composite sensor, two sensors having different detection principles can detect a person in a monitoring space, and can perform abnormality determination with high accuracy by integrally determining the detection results. That is, in the composite sensor, the detection ranges of the two sensors are basically set to be the same.

  On the other hand, the detection range of the two sensors may be different in the composite sensor. For example, since the angle of view of the camera is likely to be larger than the viewing angle of the PIR sensor, the efficiency of checking the status of the surveillance space with the camera image when the PIR sensor detects an intruder is considered. The angle of view can be set wider than the viewing angle of the PIR sensor. When it is desired to monitor a part of the monitoring space with priority, only the part may be set as the detection range of the PIR sensor. In addition, when there is a false alarm factor of the PIR sensor such as a background object whose temperature changes, for example, the viewing angle of the PIR sensor has to be narrowed.

  Thus, when the detection range of the PIR sensor is set to a part of the shooting range of the camera, it is necessary to grasp the detection area of the PIR sensor set on the image shot by the camera in order to perform abnormality determination. . However, since the relationship between the angle of view of the camera and the direction of the detection range of the PIR sensor can change depending on the situation of the installation location of the device, it is necessary to specify the detection area on the image for each installation location and set it in the device Occurs. There is a walk test as a method for easily performing this work (Patent Document 2 below). In the walk test, for example, the installation worker walks around the monitoring space and records the position of the human image on the image when the PIR sensor reports. Then, the set of human images is specified as the detection area of the PIR sensor.

JP 2000-33252 A JP 2006-178515 A

  A PIR sensor intended for human body detection is generally configured using a pyroelectric element. The pyroelectric PIR sensor causes infrared rays from a plurality of zones to enter the pyroelectric element with a lens disposed in front of the pyroelectric element. A set of a plurality of zones defines the detection range of the pyroelectric PIR sensor.

  The pyroelectric element outputs a detection signal according to a change in the amount of incident infrared rays. Because of this principle, a delay time is generated to detect the change in the amount of infrared rays. Therefore, the position of the zone in the detection area where the human body passes and the position where the human body exists at the timing when the PIR sensor detects the human body (on the image) It does not necessarily match the position of the person who appears. That is, there is a problem that the detection area of the PIR sensor set on the image taken by the camera may not be specified accurately.

  The present invention has been made to solve the above-mentioned problems, and the relationship between the image of the camera corresponding to the detection area of the image sensor and the detection area of the human body sensor on the image is suitably specified and appropriately It is an object of the present invention to provide a composite sensor that can perform abnormal determination.

  The composite sensor according to the present invention has a function of detecting a human body from an image obtained by capturing a monitoring space, and a detection signal obtained from a detection space set as a part of the monitoring space. A human image extracting unit for extracting a human image from the image, and a monitoring space for determining whether or not there is an intruder in the monitoring space based on the detection result of the human body detecting function. A recording unit for recording measurement data in which the position of the human image on the image at each measurement timing and the signal level by the human body sensor are associated with each other, and movement of the human body in the monitoring space A speed estimation unit for estimating speed, and the measurement data having the signal level equal to or higher than a predetermined level at which a human body is detected, the human image position according to the moving speed Corrected to generate corrected measurement data, and based on the corrected measurement data, an area specifying unit for specifying a composite detection area corresponding to the detection space on the image, and a human image extracted from the image is the composite image And an abnormality determination unit that executes different abnormality determination processing depending on whether the detection area or the single detection area that is other than the detection area exists. In the different abnormality determination process, for example, an abnormality is determined only by the image sensor outside the composite detection area, and the abnormality is determined by only the human sensor or the integrated evaluation of the human body sensor and the image sensor in the composite detection area. It is also preferable to change the evaluation criteria of the image sensor inside and outside the composite detection area.

  Another composite sensor according to the present invention includes a direction estimation unit that estimates a moving direction of the human body in the monitoring space based on the image, and the area specifying unit is configured to detect the human image position of the measurement data. Then, the corrected measurement data is generated by correcting the position of the distance according to the moving speed in the direction opposite to the moving direction.

  According to the present invention, in a composite sensor including an image sensor and a human body sensor, the relationship between the detection area of the image sensor and the detection area of the human body sensor is suitably specified, and appropriate abnormality determination is possible. Appropriate abnormality determination is possible.

It is a block diagram which shows the structure of the outline of the security system comprised using the composite type | mold sensor which is embodiment of this invention. It is a schematic diagram which shows an example of the positional relationship of the monitoring space image | photographed with the camera of an imaging part, and the detection space of the PIR sensor of a sensor part. It is an outline processing flow figure explaining operation of the vigilance mode. It is a schematic processing flowchart explaining operation | movement of registration mode. It is a schematic diagram explaining the process which specifies the composite detection area in an image based on the intensity | strength characteristic in an image. It is a schematic diagram explaining the other process which specifies the composite detection area in an image based on the intensity | strength characteristic in an image.

  Hereinafter, a composite sensor 2 according to an embodiment of the present invention (hereinafter referred to as an embodiment) will be described with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of a security system configured using the composite sensor 2. The security system includes a composite sensor 2, a security device 4, an operation display 6, a communication network 8, and a monitoring center 10. When the composite sensor 2 detects a suspicious person who has entered the monitoring area, an abnormality notification is made. An image of the monitoring area is sent to the monitoring center 10 together with the signal. Upon receiving the abnormality report, the monitoring center 10 can take appropriate measures such as monitoring the situation based on the sent image and dispatching a guard.

  The composite sensor 2 is installed in a property to be monitored. For example, the composite sensor 2 is installed in the monitoring area of the yard of the house to be monitored or the outer periphery of the building. The composite sensor 2 can also be used for indoor monitoring. When the composite sensor 2 detects a suspicious person by an image sensor or a human body sensor in the monitoring area, it makes an abnormality determination based on the detection result, and then sends an abnormality notification signal and images for a predetermined period before and after the abnormality notification signal to the security device 4. Output.

  The security device 4 is basically installed for each property to be monitored. The security device 4 is connected to the composite sensor 2 and the operation indicator 6 so that they can communicate with each other via wireless / wired communication, and is connected to a remote monitoring center 10 via a communication network 8. The security device 4 can be set to a monitoring mode and a monitoring cancellation mode, and relays an abnormality notification signal and an image from the composite sensor 2 to the monitoring center 10 in the monitoring mode. On the other hand, in the monitoring release mode, even if an abnormality notification signal is received from the composite sensor 2, relaying to the monitoring center 10 is not performed.

  The security device 4 can be connected to a plurality of composite sensors 2 installed in the property to be monitored, and can be connected to other sensors. The security device 4 may integrally determine signals from a plurality of sensors, perform final abnormality determination, and may perform processing such as reporting to the monitoring center 10 based on the result. For example, when a reader for detecting an RFID (Radio Frequency IDentification) tag to be carried by an entrance permitter is installed in the monitoring area, and the composite sensor 2 issues a report when the reader does not detect the tag. The security device 4 determines that there is an intrusion abnormality.

  The operation display 6 includes a monitor that displays a monitoring image. For example, the user of this security system, such as a resident of a monitoring object house, can confirm a monitoring area with an image at any time with the operation indicator 6. The operation indicator 6 can be provided with input means such as buttons, switches, and numeric keys for performing various operations on the security device 4, and a speaker for outputting a voice guide or an alarm when an abnormality is detected.

  The communication network 8 is responsible for communication between the security device 4 and the monitoring center 10. The communication network 8 is configured using mobile communication in addition to the public line network.

  The configuration of the composite sensor 2 will be described with reference to FIG. The composite sensor 2 includes an imaging unit 20, a space sensor unit 22, a monitoring processing unit 24, a storage unit 26, a display unit 28, an operation unit 30, and a communication unit 32.

  The imaging unit 20 is a camera that captures an image of the monitoring space and acquires an image. For example, the imaging unit 20 generates an image signal of the monitoring region using a CCD (Charge Coupled Device) imaging device or the like. The imaging unit 20 and the monitoring processing unit 24 constitute an image sensor that detects a human body from an image of the monitoring space.

  The space sensor unit 22 includes an infrared sensor that detects an infrared change obtained from a detection space set as a part of the monitoring space as a human body sensor, and detects the human body in the detection space according to the signal level of the detection signal. In this embodiment, the space sensor unit 22 includes a PIR sensor using a pyroelectric element as an infrared sensor. The PIR sensor collects infrared rays by an optical system such as a mirror or a lens and receives the light by a pyroelectric element. For example, an optical system including a plurality of Fresnel lenses arranged to face the pyroelectric elements is provided in front of the pyroelectric elements. The optical system is configured to allow infrared rays from a plurality of zones to enter the pyroelectric element, and these plural zones are integrated to form a detection range of the pyroelectric PIR sensor. The detection range forms fan-shaped detection spaces in the horizontal and vertical directions around the PIR sensor. The angle of the detection range is set to be narrower than the angle of view of the camera of the imaging unit 20. For example, by setting a detection range with the detection directionality in the horizontal direction narrowed down in the form of a beam along the outer wall of the building, it is possible to detect only the human body approaching the building and efficiently detect suspicious individuals who are trying to enter the building Can be detected.

  The space sensor unit 22 extracts an amount corresponding to the amplitude of the detection signal (light reception signal) output from the PIR sensor as the signal level of the detection signal, determines a threshold based on the signal level, and determines the presence or absence of a human body in the detection space. . The space sensor unit 22 outputs a signal indicating the presence or absence of a human body to the monitoring processing unit 24, and also outputs the signal level of the detection signal at that time to the monitoring processing unit 24 when it is determined that a human body is present. Note that the threshold value Z3 set in the space sensor unit 22 is set to such an extent that a case where it is clear that the light reception signal of the PIR sensor is not caused by a human body is excluded. That is, the space sensor unit 22 outputs to the monitoring processing unit 24 not only when it can be determined that it is a human body but also when it is suspected that it is a human body.

  FIG. 2 is a schematic diagram illustrating an example of the positional relationship between the monitoring space 50 photographed by the camera of the imaging unit 20 and the detection space 52 of the PIR sensor of the space sensor unit 22. FIG. 2A includes the position of the composite sensor 2, and shows the positional relationship of the composite sensor 2, the monitoring space 50, and the detection space 52 on a vertical plane along the shooting direction of the camera. ) Indicates the positional relationship on a horizontal plane. FIG. 2C shows a detection area 56 of the PIR sensor on the camera image 54 in the arrangement of FIGS. 2A and 2B.

  Although the imaging unit 20 and the spatial sensor unit 22 do not necessarily have to be at the same position, in the present embodiment, they are arranged at one place as a relatively small composite sensor 2, and the visual axis 60 of the camera of the imaging unit 20. And the origin of the central axis 62 of the detection range of the PIR sensor of the space sensor unit 22 substantially coincide with each other. On the other hand, the angle between the visual axis 60 of the camera and the central axis 62 of the PIR sensor can be changed. For example, the angle θ in the pan direction between the two can be changed. In this case, the position of the detection area 56 in the image 54 is fixed in the tilt direction (Y-axis direction), and the pan direction (X-axis) Direction). The orientation of the camera or PIR sensor with respect to the main body (housing) of the composite sensor 2 may be changeable. By making the angle changeable in this way, it is possible to suitably set the monitoring space 50 and the detection space 52 for the monitoring target and the installation position that can be variously different. Note that the angle of the tilt direction of the camera and the PIR sensor may be changeable.

  The monitoring processing unit 24 is configured using a microprocessor or the like, and performs processing on a monitoring image captured by the imaging unit 20 and a signal from the spatial sensor unit 22 according to a program to be executed. The monitoring processing unit 24 functions as an abnormality monitoring unit 40 and an area registration unit 42. The abnormality monitoring unit 40 operates the composite sensor 2 in the alert mode. Specifically, the monitoring processing unit 24 as the abnormality monitoring unit 40 analyzes the image acquired by the imaging unit 20, determines an intrusion abnormality based on the analysis result and a signal from the space sensor unit 22, and reports it. Performs monitoring processing to alarm. On the other hand, the area registration means 42 operates the composite sensor 2 in the registration mode. Specifically, the monitoring processing unit 24 as the area registration unit 42 is configured to detect the position of the detection area of each sensor on the image based on the signal from the spatial sensor unit 22 and the signal from the imaging unit 20 that accompany the walk test. An area registration process for specifying the relationship and registering information regarding the position of the detection area in the storage unit 26 is executed. In the present embodiment, as described above, the position in the pan direction of the detection space 52 in the monitoring space 50 can be changed, while the tilt direction is fixed. From the viewpoint of facilitating understanding of such points and the essence of the configuration, in this embodiment, the positional relationship in the Y-axis direction of the detection area is fixed in a preset state, and the area registration unit 42 determines the position of the detection area. Only the position of the detection area 56 of the PIR sensor in the image 54 in the X-axis direction is obtained as the relationship information. In the configuration in which the position is also changed in the tilt direction as described above, the monitoring processing unit 24 is configured to specify the position in the Y-axis direction in the same manner as the position in the X-axis direction.

  The storage unit 26 stores various programs executed by the monitoring processing unit 24 and data necessary for the programs. In particular, the storage unit 26 associates the position of the human image in the image of the imaging unit 20 with the signal level of the detection signal from the spatial sensor unit 22 for each position of the human body in the monitoring space in the processing by the area registration unit 42. The in-image intensity characteristic of the detected signal is stored. Further, as described above, the storage unit 26 stores the positional relationship information of the detection areas specified by the area registration unit 42.

  The display unit 28 is an image display unit such as a liquid crystal display, and is used by a worker to confirm the image of the imaging unit 20 in the registration mode.

  The operation unit 30 is a means for an operator to perform a mode switching operation or to input various settings, information, instructions, etc. to the monitoring processing unit 24 in the registration mode. For example, a keyboard, a pointing device, a button, etc. It is.

  The display unit 28 and the operation unit 30 may be configured by a portable information terminal or the like, and may be carried by an operator who is separated from the composite sensor 2 and performs a walk test in the registration mode. During the separation, the display unit 28 and the operation unit 30 are wirelessly connected to the composite sensor 2 body. Further, the display unit 28 and the operation unit 30 are configured separately from the composite sensor 2, and carry a terminal to be used as the display unit 28 and the operation unit 30 when an operator performs area registration work. The sensor 2 may be wirelessly connected.

  The communication unit 32 is a communication interface with the security device 4.

  Next, processing of the monitoring processing unit 24 in the alert mode and the registration mode will be described. FIG. 3 is a schematic process flow diagram for explaining the operation in the alert mode. In the alert mode, the monitoring processing unit 24 executes processing as an abnormality determination unit that determines the presence or absence of an intruder in the monitoring space based on the detection result of the image sensor or the human body sensor. When an image is input from the imaging unit 20, the monitoring processing unit 24 extracts a difference area between the image and the background image stored in the storage unit 26, and a significant change that forms a certain size occurs. It is monitored (S70). If no image change is detected, the input of the next image is awaited (in the case of “No” in S70). Note that image change detection may be performed by inter-frame difference processing instead of background difference processing.

  On the other hand, when an image change is detected (in the case of “Yes” in S70), an evaluation value E indicating the probability that the change area is a person is calculated (S72). The evaluation value E is calculated based on, for example, the area, width, height, and the like of the change area. Further, the evaluation value E may be calculated using other information such as a movement trajectory and a movement speed.

  The monitoring processing unit 24 assumes that the change area extracted in the process S70 is a human image, and determines whether the position in the image is within the detection area of the PIR sensor or other area (S74). ). Here, since the image hits the detection area of the image sensor, the detection area of the PIR sensor in the image is a combined detection area of the image sensor and the human body sensor, while the other image areas are only the image sensor. It is a single detection area.

  The monitoring processing unit 24 executes a different abnormality determination process depending on whether the human image extracted from the image exists in the composite detection area or the single detection area. When the position of the human image is a detection area of the PIR sensor, that is, a composite detection area, the monitoring processing unit 24 calculates the signal level (P value) of the detection signal of the PIR sensor input from the space sensor unit 22 and processing S72. The abnormality is determined based on the evaluated value (E value). Specifically, when the P value is equal to or greater than a certain reference value Th that affirms the possibility of the human body (“Yes” in S76), if the E value is equal to or greater than the reference value E0 (in S78). In the case of “Yes”, it is determined that the intrusion is abnormal (S80). On the other hand, if the P value is less than the threshold value Th (in the case of “No” in S76), if the E value is equal to or greater than the reference value E1 (in the case of “Yes” in S82), an intrusion abnormality is determined (S80). ).

  When the position of the human image is a single detection area where the image is detected alone, the monitoring processing unit 24 performs abnormality determination based only on the E value. Specifically, if the E value is greater than or equal to the reference value E2 (“Yes” in S84), it is determined that there is an intrusion abnormality (S80).

  If the monitoring processing unit 24 determines that there is an intrusion abnormality, the monitoring processing unit 24 transmits a signal notifying the intrusion abnormality and an image of a predetermined period at the abnormality detection timing to the monitoring center 10 via the security device 4 (S86). In this case, for example, the monitoring processing unit 24 can generate and transmit an image obtained by combining the detection area of the PIR sensor with the image captured by the imaging unit 20. Specifically, referring to FIG. 2C, the monitoring processing unit 24 displays a composite image in which the detection area 56 is displayed in an identifiable manner in the image 54 based on the position information of the detection area registered in the storage unit 26. Is generated. For example, the composite image can be an image in which a frame indicating the detection area 56 is superimposed. Thereby, the monitoring center 10 can visually recognize the detection area on the image, and can easily determine the cause of the alarm. The composite image may be generated by the monitoring center 10. In this case, the composite sensor 2 transmits the position information of the detection area together with the image.

  Further, the security device 4 may transfer the abnormality report from the composite sensor 2 to the operation display 6 and display the image on the operation display 6. The composite sensor 2 may transmit the current image in response to a request from the monitoring center 10 or the operation display 6 as well as when an abnormality is reported.

  If no intrusion abnormality is detected (“No” in S78, S82, S84), the monitoring processing unit 24 automatically continues the operation in the alert mode.

  In the above process, the reference values E0, E1, E2 related to the evaluation value E are set to E0 <E2 <E1, for example. When the spatial sensor unit 22 detects a human body in the configuration of the present embodiment in which the detection result by the image and the PIR sensor is integrated to determine intrusion abnormality (in the case of “Yes” in S76), Even if the evaluation value of the image sensor is low, there is validity to determine that there is an intrusion abnormality, and conversely, even though it is a composite detection area, the space sensor unit 22 does not detect a human body (“No” in S76). In this case, the validity of requesting a higher evaluation value of the image sensor exists than when the space sensor unit 22 detects a human body. Therefore, in the present embodiment, among the three reference values, E0 and E1 are set to have a relationship of E0 <E1.

  When the human image is in the single detection area (S74), the detection signal of the space sensor unit 22 cannot be used for intrusion abnormality determination, and the determination is performed using only the evaluation value E of the image sensor as described above. The reference value E2 in this case is higher than the reference value E0 in the case where the space sensor unit 22 affirms the existence of the human body from the viewpoint that the space sensor unit 22 does not affirm or deny the existence of the human body. If the space sensor unit 22 is set lower than the reference value E1 in the case of denying the presence of a human body, it can be expected that false / missing reports are suitably suppressed. Therefore, in this embodiment, E2 is set to an intermediate value between E0 and E1.

  In the above-described processing, when determining whether the position of the human image belongs to the composite detection area or the single detection area, all or most of the human image (for example, 70 to 80% or more area) is the composite detection area. If it is included, it is determined that it belongs to the composite detection area. This is because a signal level (P value) sufficient to affirm the possibility of the human body cannot be obtained if only a part of the human body enters the detection area of the PIR sensor. When the PIR sensor has high sensitivity, the area to which the PIR sensor belongs may be determined based on the fact that a majority region of the human image is included.

  As in the above-described warning mode, by making the abnormality determination logic different between the composite detection area and the single detection area, it is possible to suitably realize high-accuracy abnormality determination in which false / missing reports are suppressed. The abnormality determination logic is not limited to the above-described configuration, and can be variously configured according to the monitoring application and purpose.

  One example is a configuration that monitors intrusion only in the combined detection area. In this case, the image sensor is configured not to perform the process itself for detecting the human image outside the composite detection area, or the determination based on the evaluation value E (corresponding to the process S84 described above) is performed even if the human image is detected outside the composite detection area. Not configured. That is, the abnormality determination corresponding to the image area outside the composite detection area is not performed, and the image in the area is basically used for confirmation by the monitoring center 10 or the like when an abnormality is detected.

  Another example is a configuration in which the composite detection area is focused and monitoring of the single detection area is not so important. In this case, the reference value E2 can be set larger than the reference values E0 and E1, and priority can be given to suppressing false alarms over false alarms in the single detection area, while priority can be given to suppressing false alarms over false alarms in the composite detection area.

  Yet another example is a configuration in which the image sensor and the human body sensor share human detection in the composite detection area and the single detection area. In this case, the above-described determination processes S78 and S82 are omitted, and intrusion abnormality is determined based only on the P value in the combined detection area. On the other hand, in the single detection area, an intrusion abnormality is determined based only on the E value.

  For the suitable abnormality determination realized by the above-described warning mode, it is necessary to grasp the composite detection area and the single detection area in the monitoring image. These detection areas are set in the registration mode and registered in the storage unit 26. FIG. 4 is a schematic process flow diagram for explaining the operation in the registration mode. The worker switches the monitoring processing unit 24 to the registration mode, and performs a walk test in which it moves around in the monitoring space and checks whether it is detected by the space sensor unit 22. The monitoring processing unit 24 monitors the signal level (P value) of the detection signal of the PIR sensor during the walk test, and for each worker position where the P value is equal to or greater than a predetermined threshold Z2 in the monitoring space (in S100). In the case of “Yes”, the measurement data in which the position of the worker's human image in the photographed image at that time is associated with the P value is recorded in the storage unit 26 (S102).

  In processing S102, the monitoring processing unit 24 functions as a human image extracting unit. Extraction of a human image from an image is performed, for example, by obtaining an image change area caused by an operator from a background difference. Further, the position of the human image is, for example, the center of gravity of the region showing the human image in the image.

  The walk test can be continued until the worker performs an end operation, or the monitoring processing unit 24 can be configured to set a timer at the start and automatically end after a predetermined time. Monitoring of the P value (S100) and recording of measurement data (S102) are continued until the walk test is completed (in the case of “No” in S104).

  When the walk test ends (in the case of “Yes” in S104), the measurement data is accumulated in the storage unit. The set of measurement data represents a change characteristic (intra-image intensity characteristic) of the signal level of the detection signal in the image. The intensity characteristic in the image represents a change in the signal level of the detection signal in accordance with the position in the image, and is basically defined on the XY plane of the image. In the present embodiment, the detection in the image is performed. Corresponding to the fact that only the position of the area in the X-axis direction can be changed, the one-dimensional in-image intensity characteristic P (X) projected on the X-axis is handled.

  FIG. 5 is a schematic diagram for explaining processing for specifying a composite detection area in an image based on the in-image intensity characteristic P (X). In the figure, the horizontal axis is the X axis, the vertical axis is the P value, and the solid line portion of the characteristic curve 150 represents the in-image intensity characteristic P (X). Since the measurement data obtained by the walk test is not necessarily continuous with respect to the X axis, the in-image intensity characteristic P (X) is expressed as an approximate curve of each measurement data. When a plurality of measurement data is obtained corresponding to the same position on the X axis, the measurement data having the maximum P value is employed as the representative measurement data at the position.

  The monitoring processing unit 24 generates corrected measurement data by correcting the human image position according to the moving speed of the human body in the monitoring space with respect to measurement data having a P value equal to or higher than a reference level at which the human body is detected. Based on the data, a process as an area specifying unit for specifying the composite detection area is executed. Here, a value suitable as a reference level for human body detection is Z1. That is, Z1 is the level of the P value that is appropriate to be determined as a human body, and is the smallest value among the possible values for the threshold value Th of the reporting criteria, that is, Z1 ≦ Th. Specifically, the false alarm rate is set based on, for example, less than several percent. The monitoring processing unit 24 corrects the moving speed for the measurement data having a P value equal to or greater than Z1.

  The measurement data to be corrected may be all or part of the measurement data satisfying P (X) ≧ Z1. In the present embodiment, the monitoring processing unit 24 uses the measurement data with P (X) ≧ Z1 having the maximum P value as a representative point, and corrects the representative point. Based on the characteristic P (X), the monitoring processing unit 24 obtains the representative peak value Pmax that is the P value of the representative point and the representative peak position Xp that is the position on the image (S106).

  Subsequently, the monitoring processing unit 24 performs processing as a speed estimation unit that estimates the moving speed of the human body in the monitoring space based on the image, and processing as a direction estimation unit that estimates the moving direction of the human body based on the image. Is executed (S108). For example, the monitoring processing unit 24 obtains the change amount and change direction of the position of the human image between the image taken at the representative point measurement time and the previous image. Then, an estimated value of the moving speed is calculated from the difference between the change amount and the image capturing time. Further, the change direction is estimated as the movement direction at the time of measuring the representative point.

  The monitoring processor 24 corrects the representative peak position Xp based on the estimation result of the moving speed and the moving direction (S110). Specifically, the monitoring processing unit 24 corrects the position of the human image position constituting the representative point measurement data at a distance proportional to the moving speed in the direction opposite to the estimated moving direction, and generates corrected measurement data. When calculating the movement distance, an empirically obtained value can be used as the proportionality coefficient multiplied by the movement speed. When the response time of the PIR sensor used for the space sensor unit 22 is known, it is preferable to use the response time as a proportional coefficient. Further, a walk test for estimating the response time may be performed. For example, a straight line that crosses the detection range of the space sensor unit 22 is set, and the worker moves on the straight line at a constant speed and crosses the detection range. This operation is performed by changing the moving direction on the same straight line, and the in-image intensity characteristic P (X) at that time is obtained. For example, the operation is performed one or more times to detect the positional deviation in the X-axis direction between the in-image intensity characteristic P (X) and the in-image intensity characteristic P (X) of the forward path, and the difference between the positions and the work The response time of the PIR sensor can be obtained from the moving speed of the person.

  The position on the image of the representative point corrected in the process S110 is represented as a corrected peak position Xpc. Here, the actual size W of the composite detection area on the image is given by the angle of view of the camera of the imaging unit 20 and the angle of the detection range of the PIR sensor of the spatial sensor unit 22. The monitoring processing unit 24 determines the range where the width is W around the corrected peak position Xpc as the composite detection area (S112). When the composite detection area is determined, an area other than the composite detection area in the image is determined as the single detection area. The range on the X axis of the composite detection area and the single detection area in the image is registered in the storage unit 26 as information indicating the positional relationship of the detection areas of the sensors (S114). Both detection areas registered in this registration mode are used in the alert mode as described above.

  By performing such correction, it is possible to absorb the position shift due to the movement speed of the worker during the walk test, and to appropriately set the position of the detection area on the image even when the movement speed is high. Can do.

  Note that the method for obtaining the moving speed and the moving direction described above in step S108 is an example, and other methods may be used. For example, the moving speed can be estimated from the fluctuation cycle of the detection signal waveform output from the PIR sensor. Specifically, it is estimated that the moving speed is higher as the fluctuation period is shorter. Furthermore, not limited to the above example, the moving speed and moving direction may be acquired by a separate method / means.

  The above-described embodiment is based on the premise that the in-image intensity characteristic P (X) is maximized at the center of the composite detection area. Such a premise can be suitably established when the detection range of the PIR sensor is configured by a single zone in the pan direction, or when the sensitivity in the center direction of the detection range (the center of the pan direction width) is the highest.

  On the other hand, the measurement data subject to position correction is not limited to only one peak, but the measurement data satisfying P (X) ≧ Z1 is corrected at a plurality of points, and P (X) ≧ from the plurality of corrected measurement data. The corrected position of the range that becomes Z1 is estimated, and the combined detection area can be set around the corrected range of P (X) ≧ Z1. For example, all the measurement data satisfying P (X) ≧ Z1 can be corrected, and the composite detection area can be set based on the range including the set of the corrected positions X. In such a configuration, it is not necessary to assume that the in-image intensity characteristic P (X) is maximized at the center of the composite detection area.

  In addition, if the measurement data is acquired only when the worker is moving in a specific direction of either the positive direction or the negative direction of the X axis in the walk test, the monitoring processing unit 24 estimates the moving direction. There is no need, and only the moving speed can be estimated.

  When the above response time is known, the human image position for determining the moving speed and moving direction in step S108 is determined based on an image taken at a time retroactive to the response time of the representative point. Also good.

  FIG. 6 is a schematic diagram for explaining the process of specifying the composite detection area based on the range including the human image position of the correction measurement data satisfying P (X) ≧ Z1 described above. In the figure, the horizontal axis is the X axis, the vertical axis is the P value, and the solid line portion of the characteristic curve 200 is the intensity characteristic P (X) in the image obtained by correcting the position of the measurement data P (X) ≧ Z2. Represents.

  Based on the corrected characteristic P (X), the monitoring processing unit 24 extracts a detection position distribution range Ro including the human image position X having a P value equal to or greater than the threshold value Z1. Here, regarding the detection area corresponding to one spatial sensor unit 22, the set of X satisfying P (X) ≧ Z1 is not necessarily one continuous range, and may be divided into a plurality of ranges. In this regard, the detection area for one spatial sensor unit 22 should be essentially one range, and the threshold value Z1 is set to such an extent that it can be said to be a human body reliably as described above. , P (X) ≧ Z1, and the entire position R (≡ [Xmin1, Xmax1]) defined by the minimum value Xmin1 and the maximum value Xmax1 is the detection position distribution range that becomes the core of the detection area of the PIR sensor It is defined as Ro. The monitoring processor 24 can set a composite detection area having a width W so that the range Ro is located at the center.

  Furthermore, the monitoring processing unit 24 expands and corrects the detection position distribution range so as to include the human image position X that exists within a predetermined extension limit distance B from the detection position distribution range Ro and satisfies P (X) ≧ Z2. The correction distribution range Re may be obtained, and the composite detection area may be set so that the range Re is located at the center. Specifically, outside the lower limit of the range Ro by expansion correction, the range [Xmin1-B, Xmin1] is extended to the minimum value Xmin2 of the position X that satisfies P (X) ≧ Z2, and from the upper limit of the range Ro. The outside is extended to the maximum value Xmax2 of the position X that satisfies P (X) ≧ Z2 in the range [Xmax1, Xmax1 + B], and the range [Xmin2, Xmax2] is set as the correction distribution range Re. The threshold value Z2 is a P value level for which the possibility of being a human body cannot be denied, and is set slightly lower than the threshold value Z1, which is a level that can be determined with high accuracy from the human body. Further, the threshold value Z3 is set higher than the threshold value Z3, which is a level that often includes false alarm factors. Although the threshold used in the extended correction process is set to Z2 that is the same as the threshold in the process S100 when the original in-image intensity characteristic P (X) is acquired, the threshold may be different.

  The composite detection area is preferably set so as to include as much as possible the range reported by the spatial sensor unit 22 in the walk test. Therefore, the integration of the in-image intensity characteristic P (X) within the composite detection area is preferable. Maximizing the value can be a guideline for setting the composite detection area. In this respect, since the correction distribution range Re preferably captures a range where the value of the characteristic P (X) is large, the position of the composite detection area can be more suitably set with the range Re as the center.

  The larger the distance B, the better the range Re can capture the main part of the characteristic P (X), while the range Re is estimated to extend beyond the actual composite detection area. On the contrary, the accuracy and reliability of the position of the detection area can be lowered. Therefore, it is preferable to set the distance B so that the range Re does not easily reach the outside of the actual composite detection area. Specifically, the extended limit distance B increases as the difference (W−Ro) between the size W of the composite detection area and the size of the detection position distribution range Ro (represented as Ro here) increases. A smaller value can be set smaller. For example, B may be set to half of the difference (W−Ro). In this way, by performing the expansion correction by variably limiting the distance B according to the size of the detection position distribution range Ro, a P value equal to or greater than the threshold value is obtained for some reason at a position outside the detection area. However, the composite detection area can be appropriately set by excluding the position.

  The space sensor unit 22 can also be configured using another sensor that can detect a change in state caused by a human body existing in the space. For example, an ultrasonic sensor or a microwave sensor can also be used.

  2 Composite Sensor, 4 Security Device, 6 Operation Display, 8 Communication Network, 10 Monitoring Center, 20 Imaging Unit, 22 Spatial Sensor Unit, 24 Monitoring Processing Unit, 26 Storage Unit, 28 Display Unit, 30 Operation Unit, 32 Communication Part, 40 abnormality monitoring means, 42 area registration means, 50 monitoring space, 52 sensing space, 54 images, 56 sensing area, 60 visual axis, 62 central axis.

Claims (2)

A function for detecting a human body from an image obtained by imaging a monitoring space and a function for detecting a human body in the detection space according to a signal level of a detection signal obtained from a detection space set in a part of the monitoring space. In a composite sensor that determines the presence or absence of an intruder in the monitoring space based on the detection results of these human body detection functions,
A human image extraction unit for extracting a human image from the image;
A recording unit for recording measurement data in which the position of the human image on the image at each measurement timing and the signal level by the human body sensor are associated with each other with respect to the human body moving in the monitoring space;
A speed estimator for estimating a moving speed of the human body in the monitoring space;
A correction measurement data is generated by correcting the human image position according to the moving speed for the measurement data having the signal level equal to or higher than a predetermined level at which a human body is detected, and on the image based on the correction measurement data An area specifying unit for specifying a composite detection area corresponding to the detection space;
An abnormality determination unit that performs different abnormality determination processing depending on whether the human image extracted from the image exists in the composite detection area or the single detection area that is other than the area,
A composite sensor characterized by comprising: The composite sensor according to claim 1,
A direction estimating unit that estimates a moving direction of the human body in the monitoring space based on the image;
The area specifying unit performs position correction of a distance according to the moving speed in a direction opposite to the moving direction with respect to the human image position of the measurement data, and generates the corrected measurement data;
A combined sensor characterized by

Images