JP5082897B2 - Type identification unit, automatic ticket gate, and maintenance device - Google Patents

Description

  The present invention relates to a type discrimination unit that processes a captured image of a subject imaged by a camera and discriminates the type of the subject, and an automatic ticket gate and a maintenance device to which the type discrimination unit is applied. .

  Conventionally, in order to ensure the safety of users in the station premises, people such as drunk people, visually impaired persons, elderly people, etc. are detected as monitoring subjects from the users, and the detected monitoring subject is the station There has been proposed a system that issues an alarm when a predetermined dangerous area on the premises is entered (see Patent Document 1). The dangerous area mentioned here is, for example, a staircase or a platform in a station, and is a low risk area for a user who is not detected as a monitoring target (hereinafter referred to as a healthy person).

This patent document 1 detects a monitoring subject at a ticket gate. Specifically, an alcohol detection sensor is provided at the ticket gate, and the alcohol detection sensor detects a drunken person who emits alcohol from users passing through the ticket gate, or a magnetic sensor at the ticket gate. This magnetic sensor detects a magnetic tape affixed to a visually impaired person or an elderly person's cane from users passing through the ticket gate. Moreover, the captured image of the camera arrange | positioned in a station platform is processed, the action of the monitoring subject in a station platform is monitored, and a warning will be given if a monitoring subject enters a danger area. No alarm is given even if a healthy person enters the dangerous area. That is, the alarm is effectively functioned by not performing a useless alarm.
JP 2003-224844 A

  However, in the configuration of Patent Document 1, an alcohol detection sensor for detecting a drunk person and a magnetic sensor for detecting a magnetic tape attached to a cane of a visually impaired person or an elderly person must be provided. That is, for each type of user detected as a monitoring target person, a sensor for detecting the user belonging to that type must be provided. For this reason, in the structure of patent document 1, there existed a problem that the kind of required sensor increased.

  Moreover, in the structure of patent document 1, the apparatus for monitoring the monitoring subject's action in a station premises is also required. In addition, in order to ensure the accuracy of specifying the monitoring target person detected at the ticket gate from the users imaged by the cameras arranged in the station, a captured image of a place that is not a dangerous area is also required. That is, in order to track the detected person to be monitored, not only a camera that captures a dangerous area but also a camera that captures a place that is not a dangerous area is required, and the number of necessary cameras increases.

  Furthermore, as the number of cameras increases, the number of captured images to be processed increases. For this reason, an image processing apparatus with high processing capability capable of processing captured images at high speed is required.

  Therefore, the configuration of Patent Document 1 has a problem that the cost of system enlargement and system construction is high.

  An object of the present invention is to provide an inexpensive type discrimination unit that processes a captured image of a subject imaged by a camera and discriminates the type of the subject being imaged.

  It is another object of the present invention to provide an automatic ticket gate and a maintenance device that can secure the safety of the user with a simple configuration by applying the above type discrimination unit.

  The type discriminating unit of the present invention is configured as follows in order to solve the above problems and achieve the object.

In this type discrimination unit, the image processing means processes the captured image of the subject imaged by the camera and detects the positions of both knees of the subject with respect to the camera. Further, the knee behavior acquisition means processes the plurality of captured images by the image processing means, and acquires the behavior of both knees of the subject from the positions of both knees of the subject detected for each captured image. . This knee behavior acquisition means is a distance between the subject's one knee (for example, the right knee) with respect to the camera and a distance between the subject's other knee (for example, the left knee) with respect to the camera over time. The change in the difference between and is acquired as the behavior of both knees of the subject. Then, the type discriminating unit discriminates the type of the subject from the behavior of both knees of the subject acquired by the knee behavior acquiring unit.

  The types of subjects referred to here are, for example, healthy persons, drunk persons, persons with lower limb disorders, and pregnant women. In addition, there is a difference in foot movement during walking for each type of subject. The behavior of both knees differs depending on the difference in the movement of the feet during walking. For example, in the case of a healthy person, the left and right legs alternately come in front of the other leg, but the person with a lower limb disorder walks as if dragging a leg with a disability. Compared to the length of the non-disabled foot before the disabled foot or the length of the disabled foot before the disabled foot, the disabled foot is better than the unimpaired foot The length that comes out is short. As a result, the left and right feet become unbalanced. In addition, a drunk person walks while staggering from side to side because his feet are not firm. Therefore, the type of the subject can be determined from the behavior of both knees of the subject acquired by the knee behavior acquisition means.

  In this way, the captured image of the camera is processed to determine the type of the user being imaged, so that the configuration can be simplified and the cost can be reduced.

  In addition, by applying the above-described type discrimination unit, an automatic ticket gate and a maintenance device that can ensure the safety of the user with a simple configuration can be obtained. For example, when applied to an automatic ticket gate, the door opening and closing control means restricts the opening and closing of the door by the door opening and closing means according to the type of the passer who has entered the ticket gate passage determined by the type determination unit. do it. Further, when applied to the maintenance device, when the type of the subject determined by the type determining unit of the type determining unit is any of the predetermined types, the notification unit notifies that fact. What is necessary is just composition.

  According to the present invention, the captured image of the camera is processed to discriminate the type of the user being imaged, so that the configuration can be simplified and the cost can be reduced.

  Further, by applying to an automatic ticket gate or a maintenance device, the safety of the user can be ensured.

  Embodiments of the present invention will be described below.

  FIG. 1 is a diagram showing the configuration of the main part of a type discrimination unit according to an embodiment of the present invention. The type discrimination unit 1 includes a control unit 11, a captured image processing unit 12, a type discrimination data storage unit 13, and a discrimination result output unit 14. This type discriminating unit 1 discriminates whether the subject is a healthy person, a drunk person, a quasi-drunken person, a lower limb handicapped person, a pregnant woman, or a quasi-pregnant woman, and outputs the discrimination result. A quasi-drunken person is a subject who is not likely to be a drunk person but is more likely to be a drunk person than a non-drunken person. A semi-pregnant woman is a subject who cannot be determined to be a pregnant woman but is more likely to be a pregnant woman than a non-pregnant woman. The control unit 11 controls the operation of each part of the main body and executes a type discrimination process for discriminating the type of the subject described later.

A camera 5 is connected to the captured image processing unit 12. The camera 5 is a so-called stereo camera that images an imaging region from two directions. The captured image processing unit 12 captures a captured image of the camera 5 at an appropriate timing, and processes the captured captured image. Specifically, the captured image processing unit 12 processes the captured captured image, and detects the distance and direction from the camera 5 for each of the right knee and the left knee of the subject being imaged. The captured image processing unit 12 inputs the distance from the camera 5 and the direction detected for each of the subject's right knee and left knee to the control unit 11. The control unit 11 is based on the distance from the camera 5 input from the captured image processing unit 11 and the direction of each of the subject's right knee and left knee,
(1) Change in distance between subject's right knee and left knee over time (2) Size of left-right wobbling of subject's knees,
(3) The distance between the knees of the subject,
(4) The magnitude of vertical movement of the subject's knee,
Is obtained as the behavior of both knees of the subject. The process in which the control unit 11 acquires the behavior of both knees of the subject will be described later.

  The type discrimination data storage unit 13 stores discrimination data used to discriminate the type of the subject from the behavior of both knees of the subject. This determination data includes a wobble threshold, a vertical threshold, and a knee interval threshold. The wobbling threshold is a threshold for the amount of change in the left and right directions of both knees of the subject, and is used to determine whether or not the subject is a drunk person. The vertical threshold is a threshold for the amount of change in the vertical direction of both knees of the subject. The knee spacing threshold is a threshold for the distance between both knees of the subject. The upper and lower threshold values and the knee interval threshold value are used to determine whether or not the subject is a pregnant woman. The determination result output unit 14 outputs the type of the subject determined by the control unit 11.

  Here, the behavior of both knees of the subject will be described for each type of the subject. First, the behavior of both knees of a healthy person will be described. A healthy person as used herein refers to a person who has no physical impairment and is not a drunk or pregnant woman. FIG. 2 is a diagram illustrating an image of a captured image of a subject (here, a healthy person) captured by a camera. Time passes in the order of FIGS. 2A, 2B, 2C, 4D, E, and F. FIG. As shown in FIG. 2, the healthy person turns the left and right feet alternately and forwards (walking direction side) from the other foot. Therefore, in the healthy person, the distance between the camera 5 and the right knee and the distance between the camera 5 and the left knee change as shown in FIG. In FIG. 3A, the horizontal axis is the time axis, and the vertical axis indicates the distance between the camera 5 and the knee. FIG. 3B shows the difference between the distance of the right knee relative to the camera 5 and the distance of the left knee relative to the camera 5. In FIG. 3B, the horizontal axis is a time axis, and the vertical axis indicates the size of (distance between camera 5 and right knee−distance between camera 5 and left knee). The scales of the time axes in FIGS. 3A and 3B are different. In addition, since the healthy person walks almost straight, the fluctuation of both knees in the left-right direction is small. Moreover, since a healthy person does not walk with both legs extended, the space | interval of both knees is comparatively small.

  In addition, healthy persons have variations in the size of the range in which both knees move up and down during walking depending on the body shape and the heel. For this reason, the use of the size of the range in which both knees move up and down to determine whether or not the subject is a healthy person increases the erroneous determination.

  Next, the behavior of both knees of persons with lower limb disabilities will be described. People with lower limbs walk as if they have a disabled leg, so the disabled leg is not in front of the non-disabled leg, or the non-disabled leg is higher than the disabled leg. Compared with the length that appears before, the length of the foot with the obstacle is shorter than the length of the foot without the obstacle. In other words, the left and right feet move in an unbalanced manner because they walk repeatedly by putting their feet with no obstacles forward and aligning the feet with their obstacles with the unobstructed feet with their feet forward. become. For example, in the case of a handicapped person whose leg with a disability is not in front of the leg without a disability, the distance between the camera 5 and the right knee and the distance between the camera 5 and the left knee are shown in FIG. Changes as shown. FIG. 4 illustrates the case of a leg handicapped person whose right leg is a disabled leg. In the case of a leg handicapped person whose leg is a left leg, the right leg and the left leg shown in FIG. Become. FIG. 4B shows the difference between the distance of the right knee with respect to the camera 5 and the distance of the left knee with respect to the camera 5. In FIG. 4A, the horizontal axis is the time axis, and the vertical axis indicates the distance between the camera 5 and the knee. In FIG. 4B, the horizontal axis is a time axis, and the vertical axis indicates the size of (distance between camera 5 and right knee−distance between camera 5 and left knee). The time axis scales in FIGS. 4A and 4B are different.

  Some people with lower limb disabilities can walk straight depending on the degree of disability, while others may stagger from side to side. Therefore, there is variation in the magnitude of the wobbling of both knees in the left-right direction of the lower limb disabled person. In addition, people with lower limb disabilities differ in how they spread their legs depending on the degree of disability, and the distance between the knees also varies. Furthermore, people with lower limb disabilities vary in the size of the range in which both knees move up and down during walking depending on the degree of disability. For this reason, using the magnitude of wobbling of both knees in the left-right direction, the distance between both knees, and the size of the range in which both knees move up and down to determine whether or not the person has a lower limb disorder is a false judgment. It becomes a factor to increase.

  Next, the behavior of both knees of a drunk person will be described. The drunken people mentioned here include not only those who have lost their consciousness due to alcohol, but also those who have consciousness but cannot move their bodies according to their own will. Since the feet of drunken people are not solid, the movement of both feet during walking is irregular. For this reason, in the drunk person, the distance between the camera 5 and the right knee and the distance between the camera 5 and the left knee change as shown in FIG. FIG. 5B shows the difference between the distance of the right knee with respect to the camera 5 and the distance of the left knee with respect to the camera 5. In FIG. 5A, the horizontal axis is the time axis, and the vertical axis indicates the distance between the camera 5 and the knee. In FIG. 5B, the horizontal axis is a time axis, and the vertical axis indicates the size of (distance between camera 5 and right knee−distance between camera 5 and left knee). In the case of a drunk person, the behavior of both knees during walking is irregular. Therefore, as shown in FIG. 5, the line indicating the change in the distance between the camera 5 and both knees is a line with a high frequency component. The time axis scales in FIGS. 5A and 5B are different. In addition, since the drunk person cannot walk substantially straight, the wobbling of both knees in the left-right direction is large.

  In addition, a drunk person has variation in the space | interval of both knees according to a condition, and the magnitude | size of the range which both knees move up and down at the time of walking. For this reason, the use of the distance between both knees or the size of the range in which both knees move up and down to determine whether or not the person is a drunk causes increased misjudgment.

  Next, the behavior of both knees of a pregnant woman will be described. Pregnant women also have their left and right feet alternately forward (walking direction side) with respect to the other foot, as with healthy individuals. Therefore, in the pregnant woman, the distance between the camera 5 and the right knee and the distance between the camera 5 and the left knee change as shown in FIG. Further, the difference between the distance between the camera 5 and the right knee and the distance between the camera 5 and the left knee changes as shown in FIG. In addition, since pregnant women walk almost straight like normal people, the wobbling of both knees in the left-right direction is small. However, because pregnant women are heavy, they walk with their legs spread, unlike healthy people. For this reason, the distance between both knees is relatively large. Furthermore, since pregnant women are heavy, the range in which both knees move up and down during walking is small.

  Thus, there are differences in the behavior of both knees during walking depending on the type of subject (healthy person, drunk person, lower limb disabled person, pregnant woman).

  Note that the above-described wobbling threshold is set to a value that suppresses the occurrence probability of misjudgment that discriminates that a drunk person is not a drunk person. In other words, the wandering threshold value is set to a value that allows a certain degree of occurrence probability of misclassification that identifies a person who is not a drunk person as a drunk person. In addition, the upper and lower threshold values and the knee interval threshold value are set to values that suppress the occurrence probability of misjudgment that determines that a pregnant woman is not a pregnant woman. In other words, the upper and lower threshold values and the knee interval threshold value are set to values that allow a certain degree of occurrence probability of misjudgment for discriminating a non-pregnant person as a pregnant woman.

  Next, the operation of this type discrimination unit 1 will be described. FIG. 6 is a flowchart showing the operation of this type discrimination unit. In the type discrimination unit 1, the captured image processing unit 12 captures a captured image of the camera 5 (s1). The captured image processing unit 12 processes the captured image captured in s1, and determines whether or not the subject (person) is captured in this captured image (s2). In s2, for example, a difference image between the background image of the imaging region captured in the absence of the subject in advance and the captured image captured in s1 is generated, and whether or not the subject is captured in this difference image is determined. The determination is made by a known method such as pattern matching. When the captured image processing unit 12 determines that the subject is not captured in s2, the process returns to s1. The captured image processing unit 12 repeats the processes related to s1 and s2 at the first time T1 interval until it is determined that the subject is imaged in s2.

  When the captured image processing unit 12 determines that the subject is imaged in s2, the captured image processing unit 12 detects the positions of both knees of the subject with respect to the camera 5 (s3). In s3, the distance and direction from the camera 5 are detected for each of the subject's right knee and left knee.

  Note that the camera 5 is a stereo camera as described above, and the captured image processing unit 12 captures two captured images obtained by the camera 5 capturing images from two directions. In s2, the captured image processing unit 12 performs processing on one captured image captured from the camera 5, and determines whether or not the subject is captured. The captured image processing unit 12 also detects two captured images (stereo) captured from the camera 5 in s3 in order to detect the distance and orientation from the camera 5 for each of the subject's right knee and left knee. Image).

  When the captured image processing unit 12 detects the positions of both knees of the subject with respect to the camera 5 in s3, the captured image of the camera 5 is captured (s4), and the captured image captured in s4 is processed to detect the subject to the camera 5. The positions of both knees of the person are detected (s5). The captured image processing unit 12 repeats the processes related to s4 and s5 N times (s6). The captured image processing unit 12 repeats the processing related to s4 and s5 at the second time T2 interval. The process concerning s4 is the same as the process concerning s1 mentioned above, and the process concerning s5 is the same as the process concerning s3 mentioned above.

  The above-described s1 and s2 are processes for determining whether or not the subject is being imaged, that is, whether or not the subject is in the imaging region of the camera 5. On the other hand, the process concerning s4-s6 is a process for acquiring the behavior of both knees of the subject. It is preferable to set the second time T2 interval for repeating the processes relating to s4 and s5 to about several tens of ms in order to appropriately acquire the behavior of both knees of the subject. On the other hand, the first time T1 interval in which the processes relating to s1 and s2 are repeated is a process for determining whether or not the subject exists in the imaging region of the camera 5, and is set to about several hundred ms. But there is no particular problem. That is, there is no particular problem even if the first time T1 is set longer than the second time T2, and in this way, the processing load when the subject is not imaged can be suppressed. . Therefore, the first time T1 may be the same as the second time T2, but is preferably longer than the second time.

  The captured image processing unit 12 inputs the detected positions of both knees of the subject to the control unit 11 every time the positions of both knees of the subject are detected in s5. The control unit 11 acquires the behavior of both knees of the subject based on the positions of both knees of the subject that are repeatedly input N times from the captured image processing unit 12 (s7). In s7, the distance with respect to the camera 5 is detected for each of the subject's right knee and left knee by detecting the distance to the camera 5 for each of the subject's right knee and left knee. (See FIGS. 3A, 4A, and 5A). Further, the control unit 11 obtains each of the subject's right knee and left knee from the distance change with the camera 5 with the passage of time, and the subject's right knee and the left knee with the passage of time. A change in the difference in distance from the camera 5 is obtained (see FIGS. 3B, 4B, and 5B).

  Further, the control unit 11 detects the amount of wobbling in the left-right direction of both knees of the subject, the distance between both knees, and the size of the vertical movement of both knees. For each of the subject's right knee and left knee, the magnitude of wobbling in the lateral direction of both knees is acquired from the azimuth change with respect to the camera 5 over time. Specifically, when the subject is walking substantially straight, the amplitude of the azimuth change of the subject's knee with respect to the camera 5 over time is small. On the other hand, when the subject is walking while wobbling in the left-right direction, the amplitude of the azimuth change of the subject's knee with respect to the camera 5 over time is large. The control unit 11 detects this amplitude for each of the subject's right knee and left knee, and acquires the larger one as the amount of wobbling in the left-right direction of the subject's knee.

  For example, in the case of the subject A whose walking direction is the direction toward the camera 5 as shown in FIG. 7A, if the subject A is walking substantially straight, The azimuth of the knee of the person A is substantially constant, and the line indicating the azimuth change with respect to the camera 5 is the line A shown in FIG. Further, even when the walking direction of the subject B shown in FIG. 7A is not the direction toward the camera 5, if the subject B is walking substantially straight, the camera 5 The line indicating the change in the orientation of the knee of the subject B with respect to is as shown in the line B shown in FIG. On the other hand, when the subjects A and B shown in FIG. 7A are walking while wobbling from side to side, the lines indicating the azimuth changes of the knees of the subjects A and B with respect to the camera 5 are shown in FIG. Lines A and B shown in FIG. The control part 11 acquires the amplitude of the line which shows the azimuth | direction change of a subject's knee as a magnitude | size of the fluctuation | variation to the left-right direction.

  7B and 7C show a change in orientation with respect to the camera 5 with respect to one knee of the subject. 7B and 7C, the horizontal axis is the time axis, and the vertical axis is the azimuth of the knee with respect to the camera 5.

  Further, the distance between both knees of the subject can be obtained by using the distance and orientation from the camera 5 detected by the captured image processing unit 12 for the right knee and the left knee of the subject. The control unit 11 acquires the average value of the distance between both knees of the subject obtained for each captured image as the distance between both knees of the subject.

  The control unit 11 detects the height of both knees of the subject for each captured image captured by the captured image processing unit 12 from the camera 5. The height of both knees of the subject can be obtained by using the distance and orientation from the camera 5 detected by the captured image processing unit 12 for the right knee and the left knee of the subject. For each of the subject's right knee and left knee, the control unit 11 sets the interval between the minimum value and the maximum value of the knee height obtained for each captured image as the magnitude of vertical movement, Is obtained as the magnitude of the vertical movement of the subject's knee.

The control unit 11 acquired as the behavior of both knees of the subject in s7,
(1) Change in distance between subject's right knee and left knee over time (2) Size of left-right wobbling of subject's knees,
(3) The distance between the knees of the subject,
(4) The magnitude of vertical movement of the subject's knee,
Is used to perform type discrimination processing for discriminating the type of the subject (s8).

  FIG. 8 is a flowchart showing this type discrimination processing. First, the control part 11 determines whether the high frequency component has overlapped with the line which shows the distance change of a subject's right knee and left knee with progress of time (s21). If the control unit 11 determines that the high-frequency components are overlapping, the control unit 11 determines whether or not the magnitude of the subject's fluctuation is less than the fluctuation threshold stored in the type determination data storage unit 13 (s22). . If it is not less than the wobble threshold, the control unit 11 determines that the type of the subject is a drunk person (s23). On the other hand, if it is less than the wobbling threshold, the type of the subject is determined to be a semi-drunken person (s24). Here, although it cannot be determined that the person is a drunk person, the subject who is more likely to be a drunk person than the possibility that the person is not a drunk person is determined to be a semi-drunken person.

  In addition, it is good also as a structure which eliminates the process concerning s22 and s24, and discriminate | determines the subject whose high frequency component has overlapped on the line which shows the difference of the right knee and the left knee with progress of time as a drunk person in s23.

  When determining that the high frequency components are not overlapped at s21, the control unit 11 determines whether or not both knees are in front of each other with good balance (s25). In s25, the control unit 11 determines that the right knee and the left knee are not alternately in front of the other knee, or that one knee is in front of the other knee, and the other knee is When the length that comes out before the knee is short, it is determined that both knees are not in front of each other in good balance. In other words, if the length that one knee is ahead of the other knee is approximately the same as the length that the other knee is ahead of the other knee, both knees should be in front of each other in a balanced manner. It is determined that If the control unit 25 determines that both knees are not in front of each other in a well-balanced manner, the control unit 25 determines that the type of the subject is a leg impaired person (s26).

  If the control unit 11 determines in s25 that the right knee and the left knee are alternately in front of the other knee, the size of the subject's wobbling is determined by the type discrimination data storage unit 13. It is determined whether it is less than the wandering threshold stored in (S27). s27 is the same process as s22 mentioned above. If it is not less than the wobble threshold, the control unit 11 determines that the type of the subject is a quasi-drunken person (s28). If the control unit 11 determines in s27 that the distance between both knees of the subject is less than the wandering threshold value, it determines whether or not the distance between both knees of the subject is less than the knee interval threshold value stored in the type discrimination data storage unit 13. (S29). When the control unit 11 determines that it is less than the knee interval threshold value in s29, the control unit 11 determines that the subject is a healthy person (s30).

  Furthermore, if the control part 11 determines with it not being less than a knee space | interval threshold value in s29, it will determine whether the magnitude | size of the subject's knee vertical motion is less than an up-and-down threshold value (s31). If it determines with the control part 11 being less than an up-and-down threshold value, it will discriminate | determine the type of a subject to be a pregnant woman (s32). If it is determined that it is not less than the upper or lower threshold, the type of the subject is determined to be a semi-pregnant woman (s33). Here, the subject who cannot be determined to be a pregnant woman but is more likely to be a pregnant woman than the possibility of not being a pregnant woman is determined to be a semi-pregnant woman.

  In addition, it is good also as a structure which eliminates the process concerning s31 and s33, and discriminate | determines that the subject determined not to be less than a knee space | interval threshold value by s29 as a pregnant woman by s32.

  When the type discrimination unit 1 discriminates the type of the subject in s8, the discrimination result is output from the discrimination result output unit 14 (s9).

  As described above, the type discrimination unit 1 is configured so that the subject can be any one of a healthy person, a lower limb disabled person, a drunk person, a quasi-drunken person, a pregnant woman, and a quasi-pregnant lady. It is possible to determine whether it is a type and output the determination result. Accordingly, since various types of sensors are not required to discriminate the type of the subject, the configuration of the apparatus main body can be simplified and the cost can be reduced.

In the above description, the positions of both knees of the subject detected in s3 are not used for the type determination of the subject, but the type determination of the subject may also be performed using this. . On the contrary, the processing relating to s3 may be eliminated. Furthermore, in the above embodiment, the camera 5 is described as a stereo camera, but the present invention is not limited to using a stereo camera. Any type of camera may be used as long as it can detect the behavior of both knees of the subject from the captured image. That is, the camera 5 may be a monocular camera or a compound eye camera as long as it is a 3D camera capable of obtaining a stereoscopic image. For example, a monocular camera capable of irradiating a subject (in this case, a subject) with modulated light, imaging the reflected light, and measuring the distance to the subject using the phase of the reflected light is used as the camera 5. It may be used as

  Next, an automatic ticket gate to which the above-described type discrimination unit 1 is applied will be described. FIG. 9 is a diagram showing a configuration of a main part of the automatic ticket checker, and FIG. 10 is a schematic diagram showing an appearance of the automatic ticket checker. This automatic ticket gate 20 is installed at a ticket gate of a station, and a user who does not have an appropriate ticket (such as a ticket or commuter pass) passes through the ticket gate and enters the station, Limit participation from The automatic ticket gate 20 includes a control unit 21, a ticket processing unit 22, a user detection unit 23, a display unit 24, a door opening / closing unit 25, and a notification unit 26. A discrimination unit 1 and a camera 5 are provided.

  The control unit 21 controls the operation of each part of the automatic ticket gate 1 main body. The ticket processing unit 22 receives a ticket held by a user passing through the ticket gate, and reads and writes the ticket information recorded on the received ticket. The ticket includes a magnetic ticket that records the ticket information as magnetic data, a non-contact IC card that has a wireless communication function and records the ticket information in a built-in memory, and the like. The ticket processing unit 22 has a magnetic head that reads and writes the ticket information recorded on the magnetic ticket, and has a wireless communication unit that performs wireless communication with a non-contact IC card. Yes. Since these configurations are known, a description thereof is omitted here.

  The user detection unit 23 detects the position of the user in the ticket gate passage by detecting the presence or absence of the user by using a plurality of sensors provided along the ticket gate passage. The display unit 24 displays a guidance message or the like for a user traveling through the ticket gate passage on a display (not shown) provided near the exit side of the ticket gate passage. Moreover, the door opening / closing part 25 restricts the passage of the user in the ticket gate passage by opening and closing the door 25a provided near the exit side of the ticket gate passage. If the user who has entered the ticket gate does not have an appropriate boarding ticket, the notification unit 26 notifies that fact. Since the user detection unit 23, the display unit 24, the door opening / closing unit 25, and the notification unit 26 are also publicly known, description thereof is omitted here.

  Furthermore, the camera 5 is attached to a position where both the knees of the user entering the ticket gate can be imaged before the entrance of the ticket gate. The camera 5 is a stereo camera and is connected to the type discrimination unit 1. The discrimination result output unit 14 of the type discrimination unit 1 is connected to the control unit 21. The type discriminating unit 1 sets a user who enters the ticket gate as a subject, and inputs the type discriminated for the subject to the control unit 21. The control unit 21 controls opening / closing of the door 25 a by the door opening / closing unit 25 according to the type of the user input from the type determination unit 1.

  Note that the mounting position of the camera 5 is not limited to the position shown in FIG. 10, and may be located anywhere as long as it is a position where both knees of the user entering the ticket gate can be imaged before the entrance of the ticket gate. It may be.

  Next, the operation of the automatic ticket gate 1 will be described. FIG. 11 is a flowchart showing the operation of the automatic ticket gate. In the automatic ticket gate 20, the type determination unit 1 performs the above-described processing, determines the type of the user who enters the ticket gate, and inputs the determination result to the control unit 21.

  The automatic ticket gate 20 is waiting for the ticket processing unit 22 to accept a ticket held by a user who has entered the ticket gate (s41). When receiving the ticket, the ticket processing unit 22 reads the ticket information recorded on the ticket (s42). Based on the ticket information read by the ticket processing unit 22, the control unit 21 determines whether or not the ticket gate passage for this user is allowed to pass (s43). In s43, whether or not the ticket gate is allowed to pass is determined depending on whether or not the user has an appropriate ticket. When determining that the passage of the ticket gate passage is permitted in s43, the control unit 21 instructs the door opening / closing unit 25 to open the door 25a (s44). Thereby, the door opening / closing part 25 opens the door 25a, and permits the user who has entered the ticket gate passage this time to pass through the ticket gate passage.

  The ticket processing unit 22 writes entry information, entry information, and the like on a ticket possessed by a user who has been permitted to pass through the ticket gate, and collects exhausted tickets.

  On the other hand, when determining that the passage of the ticket gate passage is not permitted in s44, the control unit 21 instructs the notification unit 26 to start notification (s45). As a result, the notifying unit 26 starts notifying and the user who does not have the proper ticket for the user or the user who does not have the proper ticket for the station staff in the vicinity is the ticket gate. Notify that you have entered the passage. In addition, the control unit 21 determines whether or not the user type input from the type determination unit 1 for this user is a healthy person (s46). In s46, the type of the user input immediately before is set as the type of the user who entered the ticket gate.

  When determining that the person is a healthy person in s46, the control unit 21 instructs the door opening / closing unit 25 to close the door 25a (s47). Thereby, the door opening / closing part 25 closes the door 25a and prohibits the user who has entered the ticket gate passage this time from passing through the ticket gate passage. On the other hand, when the control unit 21 determines that the person is not a healthy person in s46, that is, the type of the user determined by the type determination unit 1 is one of a drunk person, a semi-drunken person, a lower limb disabled person, a pregnant woman, and a semi-pregnant woman. If there is, the door opening / closing part 25 is instructed to open the door 25a (s48). Thereby, the door opening / closing part 25 opens the door 25a.

  As described above, the automatic ticket gate 1 performs notification by the notification unit 26 when a non-healthy user enters the ticket gate passage without having a proper ticket, but does not close the door 25a. Open. Therefore, it is possible to prevent a non-healthy user who has entered the ticket gate passage without having a proper boarding ticket from colliding with the closed door 25a and falling down, thereby improving the safety of the user. In addition, since the notification by the notification unit 26 notifies the station staff in the vicinity that the user has entered the ticket gate passage without having an appropriate ticket, the station staff can be appropriately handled. it can.

  Here, when a non-healthy user enters the ticket gate without having a proper ticket, the door 25a is opened, but the speed of closing the door 25a is lower than usual. You may do it.

  Next, a maintenance device to which the above-described type discrimination unit 1 is applied will be described. FIG. 12 is a diagram showing a configuration of a main part of the maintenance device. The maintenance device 30 is a device for monitoring a user who has entered a predetermined monitoring area and ensuring the safety of the user. The maintenance device 30 includes a control unit 31, a captured image output unit 32, and a notification unit 33, and includes the above-described type determination unit 1 and camera 5.

  The camera 5 is installed so that the monitoring area becomes the imaging area. The monitoring area is a place used by a large number of people, and if it is a healthy person, there is no problem with its safety, but it is a place where the risk is high for persons with lower limb disabilities, drunk people, pregnant women, and the like. For example, as shown in FIG. 13, the camera 5 is installed such that a station platform, a staircase, or the like becomes a monitoring area. The type determination unit 1 uses a person entering the monitoring area captured by the camera 5 as a subject, and inputs the type determined for the subject to the control unit 31. The control unit 31 controls the operation of each part of the maintenance device 30 main body. The captured image output unit 32 outputs a captured image of the camera 5 in accordance with an instruction from the control unit 31. Specifically, when the type determination unit 1 determines that the subject who has entered the monitoring area is not a healthy person and is of another type, the control unit 31 controls the camera 5 with respect to the captured image output unit 32. The output of the captured image is instructed. That is, when a non-healthy person enters the monitoring area, the captured image output unit 32 outputs the captured image of the camera 5. The captured image output from the captured image output unit 32 is input to a display device (not shown) provided in a security room or the like and displayed on the display device. Further, the notification unit 33 performs notification based on a notification instruction from the control unit 31. Specifically, the control unit 31 instructs the notification unit 33 to notify when the type determination unit 1 determines that the subject who has entered the monitoring area is not a healthy person and is of another type. . That is, when a non-healthy person enters the monitoring area, the notification unit 33 notifies the user.

  FIG. 14 is a flowchart showing the operation of the maintenance apparatus. In the maintenance device 30, the type determination unit 1 performs the above-described processing, determines the type of a user who enters the monitoring area, and inputs the determination result to the control unit 31. When the type of the user who has entered the monitoring area input from the type determination unit 1 is not a healthy person, the control unit 31 is any one of a drunken person, a quasi-drunken person, a leg-impaired person, a pregnant woman, and a quasi-pregnant lady. ) (S51), the captured image output unit 32 is instructed to output a captured image of the camera 5 (s52). Thereby, the captured image of the monitoring area | region where the user who is not a healthy person approached is displayed on the display apparatus of a security room. Therefore, the staff in the security room can appropriately monitor the behavior of the non-healthy user who has entered the monitoring area.

  Further, the control unit 31 instructs the notification unit 33 to start notification (s53). Thereby, the alerting | reporting part 33 starts alerting | reporting and can notify that the user who is not a healthy person approached into the monitoring area | region to the station staff in the surroundings of the monitoring area | region. Therefore, the station staff can quickly respond to non-healthy users who have entered the monitoring area.

  In the above description, the display device that displays the captured image of the camera 5 output from the captured image output unit 32 is provided in the security room. You may install the display apparatus which displays the output image of the camera 5 output.

  Thus, according to this maintenance device 30, the safety of a non-healthy user (drunken person, semi-drunken person, lower limb disabled person, pregnant woman, semi-pregnant woman) who has entered the monitoring area can be enhanced.

  In the above embodiment, the automatic ticket checker 20 and the maintenance device 30 to which the type discrimination unit 1 according to the present invention is applied are shown, but the type discrimination unit 1 can also be applied to other devices.

It is a figure which shows the structure of the principal part of a type discrimination | determination unit. It is a figure which shows the motion of both legs at the time of a healthy person's walk. It is a figure which shows the behavior of a camera and both knees in a healthy person. It is a figure which shows the behavior of a camera and both knees in a lower limb disabled person. It is a figure which shows the behavior of a camera and both knees in a drunk person. It is a flowchart which shows operation | movement of a type discrimination | determination unit. It is a figure explaining the fluctuation | variation of both knees of a subject. It is a flowchart which shows the type discrimination | determination process of a type discrimination | determination unit. It is a figure which shows the structure of the principal part of an automatic ticket gate. It is the schematic which shows the external appearance of an automatic ticket gate. It is a flowchart which shows operation | movement of an automatic ticket gate. It is a figure which shows the structure of the principal part of a maintenance apparatus. It is a figure explaining the monitoring area | region of a maintenance apparatus. It is a flowchart which shows operation | movement of a maintenance apparatus.

Explanation of symbols

1-type discrimination unit 5-camera 11-control unit 12-captured image processing unit 13-type discrimination data storage unit 14-discrimination result output unit 20-automatic ticket gate 21-control unit 22-ticket processing unit 23-use Person detection unit 24-Display unit 25-Door opening / closing unit 25a-Door 26-Notification unit 30-Maintenance device 31-Control unit 32-Captured image output unit 33-Notification unit

Claims (7)

Image processing means for processing a captured image of a subject imaged by a camera and detecting positions of both knees of the subject with respect to the camera;
The image processing means processes a plurality of captured images, and from the positions of both knees of the subject detected for each captured image, knee behavior acquisition means for acquiring the behavior of both knees of the subject;
From the behavior of both knees of the subject acquired by the knee behavior acquisition means, a type discrimination means for discriminating the type of the subject , and
The knee behavior acquisition means changes the difference between the distance of the subject's knee with respect to the camera and the distance of the subject's other knee with respect to the camera over time. A type discrimination unit acquired as the behavior of both knees of the examiner . The knee behavior acquisition means, as the behavior of both knees of the subject, even you get a change in the lateral direction of the knees of the subject, type determination unit according to claim 1. The knee behavior acquisition means, as the behavior of both knees of the subject, the vertical variation of the knees of the subject also you get, type determination unit according to claim 1 or 2,. The knee behavior acquisition means, as the behavior of both knees of the subject, the distance between both knees of the subject also you get, type determination unit according to any of claims 1-3. The image processing unit processes the captured image obtained by capturing an image of a subject from two directions, that to detect the position of the knees of the subject, type determination unit according to any of claims 1-4. In an automatic ticket gate equipped with door opening and closing means for opening and closing the door provided on the exit side of the ticket gate passageway,
The type discrimination unit according to any one of claims 1 to 5 ,
This type discrimination unit discriminates the type of the user entering the ticket gate passage imaged by the camera,
An automatic ticket gate comprising door opening / closing control means for restricting opening / closing of the door by the door opening / closing means according to a type of a user entering the ticket gate passage determined by the type determination unit. The type discrimination unit according to any one of claims 1 to 5 ,
This type discrimination unit discriminates the type of the subject entering the monitoring area imaged by the camera,
A maintenance apparatus provided with an informing means for informing, when the type of the subject determined by the type determining unit is any of predetermined types.

Images