JP4041678B2 - Home fallen person detection method and apparatus - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a home fallen person detection device and a home fallen person detection method for detecting a fallen person from a platform at a station.
[0002]
[Prior art]
Conventionally, a fallen person detection mat is used as an apparatus for detecting a fallen person from a platform of a station. The fallen person detection mat consists of a mat laid on the side of the railroad track near the train door between the platform and the track. When a passenger falls on the mat from between the train and the platform, the impact is detected. Thus, the switch in the mat is operated. When the switch is operated, the related traffic signal is displayed as a stop, and the departure of the train is suppressed.
[0003]
[Problems to be solved by the invention]
However, if the above-mentioned faller detection mat is provided over the entire length of the platform to reliably detect a fallen person and cover the entire surface of the track, troubles may occur in sleeper replacement work, ballast tamping work, etc. New problems occur.
[0004]
The present invention has been made to solve the above problems, and an object of the present invention is to provide a home faller detection apparatus and method that can detect a fallen person over the entire length of the home and can be easily realized at low cost.
[0005]
[Means for Solving the Problems]
In order to achieve the above object, a home fallen person detection method according to the present invention shoots a predetermined detection area in a station platform with a stereo camera having a pair of left and right cameras, and the home is detected based on the captured image signal. The falling object is detected, the size of the falling object is obtained, and image processing for determining whether or not the falling object is a person from the size is performed.
[0006]
The home faller detection device according to the present invention includes a pair of left and right cameras, a stereo camera that captures a predetermined detection area in a station platform, and a fall from the home based on an image signal captured by the stereo camera. Image processing means for detecting an object, determining the size of the falling object, and determining whether the falling object is a person from the size is provided.
[0007]
[Action]
Therefore, according to the present invention, a predetermined detection area in the platform of the station is photographed by the stereo camera, the falling object from the platform is detected based on the photographed image signal, and the size of the falling object is obtained. Whether or not the fallen object is a person can be determined from the size. In order to determine the size of the fallen object, for example, the distance to the fallen object is obtained based on the known distance between the left and right cameras in the stereo camera and the focal length, and the height of the fallen object is determined by a predetermined calculation based on this distance. The size of the fallen object can be obtained based on this height and the number of pixels occupied by the fallen object. If this size is greater than or equal to a predetermined value, it is determined that the falling object is a person, and if it is smaller than the predetermined value, it can be determined that the falling object is a small object other than a person.
[0008]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
The home fallen person detection device according to the present embodiment uses a stereo camera to detect a fallen person from the home. Since one stereo camera has a predetermined detectable range (detection area), a plurality of stereo cameras are installed at a predetermined height along the home in order to monitor the total length of the home. Based on the image signal captured by the stereo camera, the distance from the camera to the falling object and the size of the falling object are detected while the falling object is falling, and the fallen person and other small objects are detected. I try to identify.
[0009]
FIG. 1 is a block diagram showing a home fallen person detecting device according to an embodiment of the present invention.
In FIG. 1, reference numeral 100 is a field device unit, and 200 is a monitoring processing unit.
In the field device unit 100, reference numeral 10 denotes a stereo camera that captures a predetermined detection area and has a certain dynamic range corresponding to a change in illuminance. In the present embodiment, the case where it is installed in an 11-car train-compatible home (about 240 m) is taken as an example, and the length of the detection area of one stereo camera 10 is about 40 m. 1-No. 6 stereo cameras 10 are installed. Reference numeral 11 denotes an E / O converter that converts an electrical image signal photographed by each stereo camera 10 into an optical image signal. Reference numeral 12 denotes an O / E converter that converts a light control signal sent from the monitoring processing unit 200 into an electrical control signal.
[0010]
Reference numeral 13 denotes an optical fiber line that transmits the optical image signal to the monitoring processing unit 200, and reference numeral 14 denotes an optical fiber line that transmits the optical control signal to the field device 100.
[0011]
In the monitoring unit 200, 15 is an O / E converter that converts the optical image signal into an electrical image signal, and 16 is an image signal sent from the field device unit 10 to detect a fallen person and detect a fallen person. This is an image processing unit that outputs the detection signal 18 and its image signal and also outputs a control signal for the stereo camera 10, and six image processing units 16 of 1 to 6 are provided corresponding to the six stereo cameras 10. ing. 17 is an E / O converter that converts a control signal from the image processing unit 16 into an optical control signal, 19 is a central monitoring unit that performs monitoring processing, and 20 is a central monitoring unit that detects a fallen person detection signal 18 from the image processing unit 16. A hub such as a change-over switch to be transmitted to 19.
[0012]
Reference numeral 21 denotes a fallen person detection relay that is operated based on the fallen person detection signal 18, and reference numeral 22 denotes a monitor TV that displays a captured image of the stereo camera 10 based on the faller detection signal 18. A power supply unit 23 converts an AC power supply voltage into a predetermined DC power supply voltage and supplies the converted power supply voltage to the field device unit 100 and the monitoring processing unit 200. For example, an uninterruptible power supply (UPS) is used.
[0013]
FIG. 2 shows an arrangement relationship between the stereo camera 10 and the home 30.
As shown in FIG. 2A, the six stereo cameras 10 are installed along the home 30 having a total length of 240 m. 1-No. For each of the 6 cameras, a detection area 40 (No. 1 to No. 6) 40 m ahead is set. Arrow a is the direction of train entry and is the same as the shooting direction of each camera. FIG. 2B shows an example of dimensions and installation height of the stereo camera 10. The stereo camera 10 has the dimensions shown in the figure, and is set to an installation height H _c (= 2.4 to 3.4 m) from the top surface of the home 30. The height H ₀ (falling height) from the upper surface of the home 30 to the reference surface 50a that is the head surface of the rail 50 is 1.1 m, and the height of the rail 50 is 0.1 m.
[0014]
FIG. 3 shows the relationship between the detection area 40, the home 30, and the stereo camera 10.
In FIG. 3, the detection area 40 has a length L (in this case, 40 m) and a width W that spans the two rails 50. Threshold values H ₁ , H ₂ and H ₃ as detection heights ₁ , ₂ and ₃ are set between the height H ₀ (1.1 m in this case) from the top surface of the home 30 to the reference surface 50a as shown in the figure. ing. Reference numeral 60 denotes a fallen person from the home 30. Among these thresholds, H ₁ is the value of a so-called dead zone, which is provided in order to prevent malfunction by the detection of such passengers umbrella above and swayed bags for passing the end of the home, H _3, except humans A value provided to determine that the fallen object is small, H _2, is set as an intermediate point when the time from H ₁ to H ₃ is measured.
[0015]
Next, a fallen person detection operation with the above configuration will be described.
FIG. 4 shows in principle the image processing performed based on the images taken by the left camera 10 _L and the right camera 10 _R constituting the stereo camera 10.
In FIG. 4, the left camera 10 _L and the right camera 10 _R are arranged with an interval B (0.4 m in FIG. 2B). Let f be the focal length of each camera, and z be the distance to the object 60X including the fallen person 60. In the stereo camera 10, since the two cameras 10 _L and 10 _R are shifted in parallel, the left and right captured images are shifted in parallel as shown in the figure. This amount of deviation is called parallax d, and is a value determined by the above B, f, and z. The parallax d can be obtained by image processing of the image processing unit 16 based on the left and right captured images.
[0016]
In the present embodiment, the distance z is obtained by the following equation (1) from d obtained by image processing and the above B and f.
z = B · f / d (1)
[0017]
Next, the height h of the object 60X is obtained from z. FIG. 5 shows the method.
In FIG. 5, the actual height h of the object 60X photographed at the position (X, Y) from the distance z (X, Y) to the object 60X and the distance distribution z ₀ of the reference surface 50a is expressed by the following equation. (2). .
h = H _c {1− (z / z ₀ )} (2)
As described above, the height h of the object 60X can be calculated based on the distance z from the stereo camera 10 to the object 60X.
[0018]
In the image processing unit 16, first, a moving object that moves in a predetermined direction (falling direction) from the captured image of the stereo camera 10 and becomes lower than the threshold value H ₁ , that is, a falling object is detected. Next, the number of occupied pixels on the captured image is obtained using the detected falling object as the object 60X, and the actual size of the object 60X is obtained based on the number of occupied pixels and the height h of the object 60X. . If this size is equal to or greater than a predetermined value, it is determined that the object 60X is a person (falling person 60) as shown in FIG. When the size is smaller than the predetermined value, the object 60X is a small object other than a human (eg, umbrella, bag, newspaper, bird, small animal, etc.) as shown in FIGS. It is determined that
[0019]
Furthermore, as a condition for determining whether or not the object 60X is the fallen person 60, when a person falls from the home 30 having the height H ₀ , it takes about 0.5 seconds to land. That is, longer falling objects than the time 0.5 seconds to pass through the threshold _{H 0 → H 1 → H 2} → H 3 in FIG. 3, it can be judged not to be a human comprising for example those lighter newspapers etc. . In this embodiment, the determination based on the time is performed. However, the technique of the present invention is not limited to the case of comparing the time itself, and includes all cases in which the determination is performed based on the falling speed. .
[0020]
Also, if the passenger as twirl protrude from the end of the home 30 a bag or the like to the line side, the determination process is started as the object 60X from when the bag or the like becomes lower than the threshold value H ₁ (H ₂ side) However, if it does not fall below a predetermined threshold, it will not be erroneously detected as a fallen person. Similarly, when a bird, newspaper, or the like descends and then rises in the middle, no false detection will occur.
Since each determination process described above is performed in the image processing unit 16 while the object 60X is falling, the faller 60 can be detected quickly.
[0021]
Incidentally, in FIG. 6 (c), the case where the object 60X is of a predetermined size smaller ones and thus to land to fall, since undetectable becomes lower than the threshold value H _3, if not the falling person 60 Can be determined.
[0022]
Next, the operation described above will be described with reference to FIG.
In FIG. 1-No. Each stereo camera 10 of No. 6 detects the object 60X for the detection area 40 in charge. The photographing operation of each stereo camera 10 is controlled by a control signal sent from each image processing unit 16 via the E / O converter 16, the optical fiber line 14, and the O / E converter 12. The image signal photographed by each stereo camera 10 is sent to the image processing unit 16 via the E / O converter 11, the optical fiber line 13, and the O / E converter 15, and the image processing unit 16 Processing is performed 6 frames per second to detect the object 60X. If it is determined that the object 60X is the fallen person 60, the faller detection signal 18 is sent to the central monitoring unit 19 via the hub 20 together with the image signal.
[0023]
In the central monitoring unit 19, when the fallen person detection signal 18 from the image processing unit 16 continues for 3 frames or more, the faller detection relay 21 is operated and the state of the scene by the photographed image from the stereo camera 10 is monitored by the monitor TV22. To display. The person in charge is immediately notified and measures are taken to stop the train from entering. Furthermore, as long as an image larger than a predetermined size (for example, the size of a person) is detected in the monitoring area, it is assumed that the faller has not yet been evacuated or rescued, and the alarm state is maintained. You may make it do. Further, instead of the image processing, the alarm state may be maintained when an image exists at a position higher than the lowest threshold value H ₃ .
[0024]
【The invention's effect】
As described above, according to the present invention, it is possible to reliably detect a person who falls from the home from another falling object. In addition, a fallen person can be detected quickly in the middle of the fall, and can also be detected from a stereo camera at a remote location.
In addition, unlike conventional faller detection mats, there is no need to lay beside the track, so there is no hindrance to the track repair work, etc., and the faller can be monitored over the entire length of the platform. A home faller detection device can be easily realized at low cost.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a home fallen person detecting device according to an embodiment of the present invention.
FIG. 2 is a configuration diagram showing the relationship between the arrangement of stereo cameras and the home.
FIG. 3 is a configuration diagram showing a detection area of a stereo camera.
FIG. 4 is a block diagram showing in principle the image processing of an image signal of a stereo camera.
FIG. 5 is a configuration diagram showing a method for obtaining the height of an object based on the distance from the stereo camera to the object.
FIG. 6 is a configuration diagram showing various falling objects from the home.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 10 Stereo camera 11, 17 E / O converter 12, 15 O / E converter 13, 14 Optical fiber line 16 Image processing part 18 Falling person detection signal 19 Central monitoring part 20 Hub 21 Falling person detection relay 22 Monitor TV
30 Home 40 detection area 50 rail 50a reference surface 60 fallen person 60X object H _1, H _2, H ₃ detects the height threshold

Claims (4)

A stereo camera having a pair of left and right cameras is used to photograph a predetermined detection area at the platform of the station, a falling object from the platform is detected based on the imaged image signal, and the size of the falling object is obtained. Then, it is characterized by performing image processing to determine whether the fallen object is a person ,
The image processing determines that an object that is lower than a predetermined height set downward from the top surface of the home is a fallen object, obtains a distance from the stereo camera to the detected fallen object, and sets the distance to this distance. Based on this, the height of the fallen object and the number of screens occupied by the fallen object are obtained, and the size of the fallen object is calculated from the height and the number of occupied pixels. And then
The home fallen person detection method , wherein the image processing performs the determination based on a time during which the falling object falls at a predetermined height .   2. The image processing according to claim 1, wherein the image processing sets one or more threshold values between the predetermined heights, and determines the presence of a falling object when the falling object becomes lower than the predetermined threshold. Home fallen person detection method. A stereo camera that has a pair of left and right cameras, and detects a falling object from the home based on a stereo camera that captures a predetermined detection area in the station platform, and an image signal captured by the stereo camera, and determines the size of the falling object And an image processing means for determining whether the fallen object is a person from its size ,
The image processing means determines that an object that is lower than a predetermined height set downward from the top surface of the home is a fallen object, and obtains a distance from the stereo camera to the detected fallen object. The height of the falling object and the number of occupied pixels of the falling object are calculated based on the above, and the size of the falling object is calculated from the height and the number of occupied pixels. It is determined that there is,
The home fallen person detecting device , wherein the image processing means performs the determination based on a time during which the falling object falls at a predetermined height .   The image processing means sets one or more threshold values between the predetermined height and a reference plane, and determines that the falling object is a person when the falling object becomes lower than the predetermined threshold value. The home faller detection device according to claim 3.

Images