JP6090190B2 - Mobile management device - Google Patents

Description

  The present invention relates to a technique for determining the type of a moving body such as a train (for example, a train, a train, or a linear motor car) that travels along a track such as a track.

  Conventionally, railway companies have set a target stop position at a station platform to stop a train. The train mentioned here is a train, train, linear motor car, or the like that travels along a track such as a track. The target stop position of the train at the station platform is defined so that passengers can get on and off the train smoothly at the station platform. The target stop position defines a target reference position on the station platform surface, for example, and has an allowable range of several tens of centimeters (for example, about 20 to 30 cm as a whole) before and after the target reference position. For example, the position of the rear end (or front end) of a train on the station platform is the position of the train. The train driver stops the train at the station platform so that the rear end (or front end) of the train is located at the target stop position.

  A train where the train's target stop position (range where the rear end or front end of the train is located) is painted on the station platform, and train operators and station staff stop at the station platform based on this paint Was visually checked to see if it stopped at the target stop position.

  In addition, for the purpose of reducing the workload of train drivers and station staff, etc., the train stop position of a train stopped at the station platform is detected, and this train stop position is the target stop position specified for the station platform. There has been proposed a technique for outputting a determination result for determining whether or not there is (see Patent Document 1). Patent document 1 is a structure which detects a train stop position by measuring the distance (distance from a sensor to a coupler) to the coupler provided in the front surface of the head vehicle.

JP 2012-240519 A

  However, the configuration of Patent Document 1 is a configuration that measures the distance to a coupler provided in front of the leading vehicle in order to detect the train stop position at the station platform regardless of the type of train. That is, Patent Document 1 does not determine the type of train when detecting the train stop position.

  Moreover, in the structure which measures the distance to the coupler provided in the front surface of the head vehicle and detects a train stop position like patent document 1, the sensor which measures the distance to the coupler provided in the front surface of the head vehicle Will be installed under the home. Under the platform, there is a large amount of rust, mud, etc. scattered by contact between the train wheels and the rails. That is, under the platform is an environment in which scattered rust, mud, and the like easily adhere to the light projecting lens and the light receiving lens of the sensor. For this reason, when the sensor is installed under the platform, the frequency of maintenance (cleaning work) for wiping off rust, mud, and the like adhering to the light projecting lens and the light receiving lens increases. In addition, since this maintenance is performed under the platform of the worker, to ensure the safety of the worker, it is necessary to select a time zone from midnight to early morning when the train operation is stopped. In addition, it was difficult to secure workers.

  Recently, in order to ensure the safety of passengers at the station platform, it has been promoted to install a platform door along the end of the station platform on the track side. The home door has an openable / closable door (or fence) at a position facing a train door whose train stop position is the target stop position. When the sensor described in Patent Literature 1 is attached above the station platform, the platform door becomes an obstacle, and the distance to the rear end (or front end) of the train on the station platform surface cannot be detected. That is, the train stop position cannot be detected.

  An object of the present invention is to provide a technique for processing a distance image obtained by imaging a moving body traveling along a track at an angle from above and determining the type of the moving body.

  Another object of the present invention is to provide a technique for determining whether or not the moving body is located within a predetermined allowable range from a distance image obtained by imaging the moving body traveling along the track at an angle from above. There is to do.

  In order to achieve the above object, the mobile management apparatus of the present invention is configured as follows.

  The distance image input unit receives a distance image obtained by imaging the back surface or the front surface of the moving body that travels along the track at an angle from above. The distance image can be captured by, for example, a known TOF (Time Of Flight) camera, a distance measuring sensor using laser light, or a stereo camera.

  The distance image processing unit processes the distance image input to the distance image input unit, and for each space divided into three or more so as not to overlap in the height direction of the moving body traveling along the track, the distance image processing unit The representative distance to the moving object is detected.

  The determination value calculation unit calculates a determination value based on a distance difference between representative distances detected by the distance image processing unit for a space related to the combination for each preset space combination. A plurality of combinations of spaces are set.

The mobile body type determination unit uses a plurality of determination values calculated by the determination value calculation unit for each combination of preset spaces, and uses a plurality of determination values calculated by the distance image input unit. Determine the type.

  For example, the determination value calculation unit calculates a determination value for each type of moving object in each space combination. The mobile body type determination unit determines, for each mobile body type, whether or not the determination value calculated for the mobile body of the type is appropriate for the mobile body of the type in each space combination, In any combination of spaces, the type of the moving body determined that the calculated determination value is appropriate is determined to be the type of the moving body captured in the distance image input to the distance image input unit. .

  In addition, by additionally providing the position determination unit shown below, the movable body is determined in advance from a distance image obtained by imaging the back surface or the front surface of the mobile body traveling along the track at an angle from above. It can be determined whether or not it is located within.

  The position determination unit determines whether the moving body is within a predetermined allowable range based on the representative distance detected by the distance image processing unit in any space and the type of the moving body determined by the moving body type determination unit. Determine. The position determination unit stores, for example, a correction value to be added to the representative distance of a specific space for each type of moving object, and corresponds to the determined type for the representative distance detected in the specific space. The position to which the correction value is added is calculated as the position of the moving body. The position determination unit determines whether the calculated position of the moving body is within a predetermined allowable range. Therefore, even if the position to be the position of the moving body is not captured in the distance image input to the distance image input unit, the position of the moving body is appropriately calculated (detected), and the position of the moving body is determined in advance. It can be determined whether it is within the allowable range.

  According to the present invention, it is possible to process a distance image obtained by imaging a moving body traveling along a track at an angle from above and determine the type of the moving body.

It is the schematic which shows the stop position determination system which determines whether the stop position of the train which stopped at the station platform is appropriate. It is a block diagram which shows the structure of the principal part of a train position determination apparatus. It is a figure explaining 1st space and 2nd space. It is a figure which shows a train classification table. It is a flowchart which shows operation | movement of a train position determination apparatus. It is a flowchart which shows a train position determination process. It is a flowchart which shows a train classification determination process. It is a flowchart which shows a stop determination process.

  Embodiments of the present invention will be described below.

  FIG. 1 is a schematic diagram illustrating a stop position determination system that determines whether or not a stop position of a train stopped at a station platform is appropriate. The stop position determination system according to this example includes a train position determination device 1, a TOF (Time Of Flight) camera 2, a video camera 3, and an ultrasonic sensor 4.

  This stop position determination system determines the type of train that has entered the station platform. Moreover, the determination result which determined whether the position of the train is within the range of the predetermined target stop position is output. This determination result is displayed by a display or a lamp installed at a position that can be easily seen by train drivers, station staff, and the like. As a result, it is possible to easily check whether the train position is the target stop position by the train driver, station staff, or the like. The target stop position here has an allowable range of several tens of centimeters around the target reference position defined on the station platform surface (for example, about 20 to 30 cm as a whole). The position of the rear end (or front end) of the train on the station platform is the position of the train.

  The train position determination device 1 determines the type of the train that has entered the station platform, and calculates (detects) the position of the train according to the type determined here. And the train position determination apparatus 1 determines whether the position of the train which has entered the station platform is the target stop position (within the allowable range), and outputs the determination result. This train position determination apparatus 1 corresponds to a mobile body management apparatus according to an embodiment of the present invention. Moreover, the train position determination apparatus 1 is corresponded to the apparatus which performs the mobile body management method concerning embodiment of this invention. Moreover, the train position determination apparatus 1 has a computer that executes a moving body management program according to the embodiment of the present invention.

  A distance image captured by the TOF camera 2, a visible image captured by the video camera 3, and a detection output of the ultrasonic sensor 4 are input to the train position determination device 1. Details of the train position determination device 1 will be described later.

  The TOF camera 2 is attached above the station platform. The TOF camera 2 is attached at an angle that captures the back of the train located near the target stop position. As is well known, the TOF camera 2 includes a light source that irradiates an imaging area with infrared light, and an imaging device (n × m pixel imaging device) in which n × m light receiving elements are arranged in a matrix. The TOF camera 2 captures the distance image of the imaging area and the received light intensity image at the same timing. The TOF camera 2 captures a distance image of this imaging area by measuring the time (flight time) from irradiating the imaging area with infrared light to receiving the reflected light for each pixel. Further, the TOF camera 2 captures a received light intensity image in the imaging area with the received light amount of each pixel. The TOF camera 2 inputs a distance image (hereinafter referred to as a distance frame image) obtained by imaging the imaging area to the train position determination device 1. The frame rate of the TOF camera 2 is, for example, 10 frames / second.

  In this example, the TOF camera 2 is described as not receiving the captured light reception intensity image of the imaging area to the train position determination device 1, but the TOF camera 2 inputs the received light intensity image of the imaging area and captures it. For each pixel of the received light intensity image and the distance image, it may be determined whether the pixel is a noise pixel affected by noise. For example, if the received light intensity is less than a predetermined threshold, the pixel is determined as a noise pixel. Since the surface of the train is made of metal, this threshold can be set high to some extent. In this way, it is possible to reduce the processing load on the noise pixel in the processing described later, and to improve the reliability of the processing result.

  Similarly to the TOF camera 2 described above, the video camera 3 is also attached above the station platform. Further, the video camera 3 may be attached at an angle for imaging the back of the train located near the target stop position, or may be attached at an angle for imaging the side of the train located at the station platform. The frame rate of the video camera 3 is, for example, 30 frames / second. The video camera 3 inputs a visible image (hereinafter referred to as a visible frame image) obtained by imaging the imaging area to the train position determination device 1.

  The angle of the TOF camera 2 and the angle of the video camera 3 are not the same.

  The ultrasonic sensor 4 is attached to a railroad sleeper or the like, and detects the presence or absence of a train located in the detection area. The ultrasonic sensor 4 inputs a binary train detection signal indicating the presence or absence of a train to the train position determination device 1.

  FIG. 2 is a block diagram illustrating a configuration of a main part of the train position determination device. The train position determination apparatus 1 includes a distance image input unit 11, a distance image processing unit 12, a visible image input unit 13, a visible image processing unit 14, a detection signal input unit 15, a calculation unit 16, and an output unit 17. And comprising.

  A distance frame image captured by the connected TOF camera 2 is input to the distance image input unit 11. The distance image processing unit 12 processes the distance frame image input by the distance image input unit 11. The distance image processing unit 12 determines the three axes (x axis, y axis, and z axis) of the distance frame image input to the distance image input unit 11 according to the traveling direction of the train (X axis, A coordinate conversion process for converting into a distance frame image on the Y axis and the Z axis) is performed. The x-axis, which is the three axes of the distance frame image, is the width direction of the image sensor, the y-axis is the height direction of the image sensor, and the z-axis is a direction orthogonal to the image plane of the image sensor. Further, the three axes determined according to the traveling direction of the train, the X-axis is the width direction of the train, the Y-axis is the height direction of the train, and the Z-axis is the traveling direction of the train. The angle α formed by the x-axis and the X-axis, the angle β formed by the y-axis and the Y-axis, and the angle γ formed by the z-axis and the Z-axis are measured when the TOF camera 2 is installed. The distance image processing unit 12 performs coordinate conversion of the input distance frame image by rotating the angle α around the x axis, rotating the angle β around the y axis, and further rotating the angle γ around the z axis. A distance frame image of three axes (X axis, Y axis, and Z axis) determined according to the traveling direction of the train is obtained. The reference position of the distance frame image is the center of the imaging lens of the TOF camera 2.

  The distance image processing unit 12 uses the distance frame image (three-axis distance frame image determined according to the traveling direction of the train) that has been subjected to the above-described coordinate conversion processing to train in the first space set in advance. The representative distance A up to, the representative distance B to the train in the preset second space, and the representative distance C to the train in the preset third space are detected.

  FIG. 3 is a diagram illustrating the first space, the second space, and the third space. 3A is a rear view of the train, and FIGS. 3B, 3C, and 3D are side views of the vicinity of the back of the train. 3B, 3C, and 3D show different types of trains. As shown in FIG. 3, the first space is set to include a space where the upper end of the train is located. The second space is set below the first space. Further, the third space is set below the second space. Thus, the first space, the second space, and the third space are arranged in the height direction of the train. In addition, the first space, the second space, and the third space do not overlap with other spaces.

  The first space, the second space, and the third space have a TOF camera on the back of the train located near the target stop position without obstructing a structure such as a platform door installed in the station platform. 2 is a space that can be imaged. On the other hand, the fourth space shown in FIG. 3 is a space in which a structure such as a platform door installed in the station platform becomes an obstacle, and the back of the train located near the target stop position cannot be imaged by the TOF camera 2. It may be.

  As shown in FIGS. 3B to 3D, the position of the train on the station platform is the position of the rear end (or front end) of the train on the station platform surface. The rear end (or front end) of the train on the station platform surface belongs to the fourth space. Also, as shown in FIGS. 3B to 3D, the shape of the back surface varies depending on the type of train, so even if the position of the train on the station platform is the same, the distance to the train in the first space (Representative distance A), the distance to the train in the second space (representative distance B), and the distance to the train in the third space (representative distance C) vary depending on the type of train.

  Details of processing for detecting the representative distance A, the representative distance B, and the representative distance C will be described later.

  A visible frame image captured by the connected video camera 3 is input to the visible image input unit 13. The visible image processing unit 14 generates a difference image (so-called inter-frame difference image) between two temporally continuous visible frame images input to the visible image input unit 13 and determines whether the train is stopped. To do.

  A train detection signal indicating the presence / absence of a train is input from the connected ultrasonic sensor 4 to the detection signal input unit 15.

  The arithmetic unit 16 uses the representative distance A, the representative distance B, and the representative distance C input from the distance image processing unit 12 to enter the station platform (the train imaged in the input distance image). The type of is determined. Moreover, the calculating part 16 determines whether the position of the train in a station platform is a target stop position according to the classification of the train determined here.

  The calculating part 16 has a memory (not shown), and stores the train type table shown in FIG. 4 in this memory. This memory corresponds to the reference distance storage unit referred to in the present invention. As shown in FIG. 4, the train type table stores the reference distance (AA_TP1, etc.) for each of the first space, the second space, and the third space described above for each train type. Yes. This reference distance is a representative distance of each space detected by processing a distance image obtained by imaging the back of the train with the TOF camera 2 for each type of train. The distance image obtained from this reference distance only needs to capture the back of the train in each of the first space, the second space, and the third space described above (representative distance of each space (reference distance). The distance between the TOF camera 2 and the train is not determined. Further, the distance between the TOF camera 2 and the train in the distance image in which the reference distance is detected may be different for each type of train. In the train type table, correction values (Q1 and the like) are registered for each type of train. This correction value is a value used to calculate the position of the train on the station platform from one of the detected representative distance A, representative distance B, or representative distance C (from the representative distance A in this example).

  The distance image processing unit 12 and the calculation unit 16 correspond to a computer that executes the moving object management program according to the embodiment of the present invention.

  The output unit 17 outputs whether or not the position of the train on the station platform is the target stop position according to the determination result determined by the calculation unit 16. The determination result output from the output unit 17 is used for lighting control of an indicator lamp installed at a position that can be easily seen by a train driver, station staff, or the like. Therefore, a train operator, a station staff, or the like can confirm whether the position of the train on the station platform is the target stop position by confirming whether the indicator lamp is turned on or off.

  Moreover, the calculating part 16 has a function which determines whether the opening / closing of the platform door installed in the station platform is permitted. The output unit 17 outputs an open / close permission signal indicating permission / prohibition of opening / closing of the home door to a device (not shown) that controls opening / closing of the home door. Furthermore, the calculating part 16 controls operation | movement of each part of train position determination apparatus 1 main body. Furthermore, the train position determination apparatus 1 also has a function of instructing the TOF camera 2 and the video camera 3 to start imaging and stop imaging.

  Hereinafter, operation | movement of the train position determination apparatus 1 concerning this example is demonstrated. FIG. 5 is a flowchart showing the operation of the train position determination device.

  The train position determination device 1 waits for the train to enter the station platform (s1). At this time, the train position determination device 1 instructs the TOF camera 2 and the video camera 3 to stop imaging. The train position determination device 1 determines the approach of the train to the station platform based on the train detection signal of the ultrasonic sensor 4 input to the detection signal input unit 15. Specifically, when the train detection signal of the ultrasonic sensor 4 input to the detection signal input unit 15 is switched from a signal indicating no train to a signal indicating the presence of a train, the train position determination device 1 It is determined that a train has entered.

  When determining that the train has entered the station platform in s1, the train position determination device 1 instructs the TOF camera 2 and the video camera 3 to start imaging (s2). As a result, the TOF camera 2 starts capturing a distance image, and the video camera 3 starts capturing a visible image. The TOF camera 2 sequentially inputs the captured distance frame images to the distance image input unit 11. In addition, the video camera 3 sequentially inputs the captured visible frame images to the visible image input unit 13. The frame rate of the TOF camera 2 and the frame rate of the video camera 3 may be the same or different.

  The train position determination device 1 starts a train position determination process for determining whether or not the train position is the target stop position (s3). Moreover, the train position determination apparatus 1 starts a stop determination process for determining whether or not the train has stopped (s4). Details of the train position determination process starting at s3 and the stop determination process starting at s4 will be described later. The order of processing relating to s3 and s4 may be reversed.

  If the train position at that time is the target stop position and the train is stopped, the train position determination device 1 outputs an opening / closing permission signal for permitting opening / closing of the platform door from the output unit 17 (s5, s6). , S7). On the other hand, when the position of the train at that time is not the target stop position, or when the train is not stopped, the train position determination device 1 outputs an opening / closing permission signal for prohibiting the opening / closing of the home door from the output unit 17 ( s5, s6, s8).

  The train position determination apparatus 1 repeats the processes related to s5 to s8 described above until it is determined that the train has started from the station platform (s9). The train position determination device 1 determines the departure of the train from the station platform based on the train detection signal of the ultrasonic sensor 4 input to the detection signal input unit 15. Specifically, when the train detection signal of the ultrasonic sensor 4 input to the detection signal input unit 15 is switched from a signal indicating that there is a train to a signal indicating that there is no train, the train position determination device 1 It is determined that the train has started.

  If the train position determination device 1 determines that the train has started from the station platform in s9, the train position determination device 1 sets the system to an initial state (s10), ends the processing, and returns to s1. In s10, the TOF camera 2 and the video camera 3 are instructed to stop imaging. Moreover, the train position determination process started in s3 and the stop determination process started in s4 are stopped.

  Next, the train position determination process starting at s3 will be described. FIG. 6 is a flowchart showing the train position determination process. This train position determination process is a process using the distance frame image input to the distance image input unit 11.

  The distance image processing unit 12 takes in a distance frame image to be processed (s21). The distance image processing unit 12 captures a distance frame image that is latest in time and unprocessed among the distance frame images input to the distance image input unit 11 as a distance frame image to be processed.

  The distance image processing unit 12 performs a coordinate conversion process of converting the distance frame image to be processed into a distance frame image of three axes (X axis, Y axis, and Z axis) determined according to the traveling direction of the train ( s22).

  The distance image processing unit 12 detects the representative distance A in the first space shown in FIG. 3 from the distance frame image subjected to the coordinate conversion process (s23).

  In s23, the minimum distance in the Z-axis direction is set as the representative distance A among the pixels located in the first space in the distance frame image subjected to the coordinate conversion process and not determined as noise pixels in the process described later. Temporarily detect. The distance image processing unit 12 calculates an average value of distances in the Z-axis direction for a predetermined number of pixels (for example, about 10 pixels) positioned around the temporarily detected representative distance A pixels. Also in the calculation of the average value, the distance of pixels that have already been determined as noise pixels is not used. If the absolute value of the difference between the average value calculated here and the temporarily detected representative distance A is not within a predetermined range (for example, about 10 to 20 mm), the distance image processing unit 12 The pixel corresponding to the detected representative distance A is determined as a noise pixel, and the above-described temporary detection of the representative distance A is performed again. On the other hand, if the absolute value of the difference between the average value calculated here and the temporarily detected representative distance A is within a predetermined range, the distance image processing unit 12 determines the temporarily detected representative distance A as the current value. The representative distance A in the process is fixed.

  Thereby, it is possible to prevent erroneous detection of the distance measured by the pixel receiving the reflected light reflected by dust, dust, raindrops, etc. flying in the space as the representative distance A.

  In the above example, using the average value of the distances in the Z-axis direction among a predetermined number of pixels located around the pixels corresponding to the temporarily detected representative distance A, the pixels of the temporarily detected representative distance A are noise pixels. However, this provisional detection was performed using the mean square value, intermediate value, etc. of the distance in the Z-axis direction for a predetermined number of pixels located around the temporarily detected representative distance A pixels. It may be determined whether the pixel at the representative distance A is a noise pixel.

  The distance image processing unit 12 extracts all the pixels located in the second space shown in FIG. 3 from the distance frame image subjected to the coordinate conversion process, and detects the representative distance B (s24). Further, the distance image processing unit 12 extracts all the pixels located in the third space shown in FIG. 3 from the distance frame image subjected to the coordinate conversion process, and detects the representative distance C (s25).

  The processing relating to s24 and s25 is substantially the same as s23 described above, except that the target space is not the first space but the second space or the third space.

  In addition, if the light reception intensity image imaged with the TOF camera 2 is also input to the train position determination device 1, pixels whose light reception intensity is less than a predetermined threshold value are processed before the process of s <b> 23 described above. By adding the step of determining as a noise pixel, it is possible to reduce the processing load on the above s23, s24, and s25, and to improve the reliability of the processing result.

  The distance image processing unit 12 inputs the representative distance A detected in s23, the representative distance B detected in s24, and the representative distance C detected in s25 to the calculation unit 16.

  The computing unit 16 calculates a determination value used for determining the type of train using the representative distance A, the representative distance B, and the representative distance C input from the distance image processing unit 12 (s26). In s26, a determination value is calculated for each type of train. In this example, a first determination value related to the combination of the first space and the second space and a second determination value related to the combination of the second space and the third space are calculated. That is, in this example, the determination value relating to the combination of the first space and the third space is not calculated.

  A combination of the first space and the second space, and a combination of the second space and the third space are combinations of preset spaces referred to in the present invention. Further, the number of combinations of spaces for calculating the determination value may be plural, and may be three or more.

  The determination value calculated in s26 is the distance difference that is the difference between the representative distances detected in the space related to the combination and the reference distance of the space related to the corresponding combination in the type of train to be calculated (in the train type table shown in FIG. 4). This is a further difference from the distance difference that is the difference of the registered reference distance).

Specifically, when the space for the combination is the first space and the second space,
Determination value of train type AA = (representative distance A−representative distance B) − (AA_TP1-AA_TP2)
Determination value of train type BB = (representative distance A−representative distance B) − (BB_TP1−BB_TP2)
Determination value of train type CC = (Representative distance A−Representative distance B) − (CC_TP1−CC_TP2)
It is. Moreover, when the space concerning a combination is the second space and the third space,
Determination value of train type AA = (representative distance B−representative distance C) − (AA_TP2−AA_TP3)
Determination value of train type BB = (representative distance B−representative distance C) − (BB_TP2−BB_TP3)
Determination value of train type CC = (representative distance B−representative distance C) − (CC_TP2−CC_TP3)
It is.

  AA_TP1, AA_TP2, AA_TP3, BB_TP1, BB_TP2, BB_TP3, CC_TP1, CC_TP2, and CC_TP3 in the above equation are registered in the train type table shown in FIG.

In this example, the determination value is not calculated, but when the space for the combination is the first space and the third space,
Determination value of train type AA = (representative distance A−representative distance C) − (AA_TP1-AA_TP3)
Determination value of train type BB = (representative distance A−representative distance C) − (BB_TP1−BB_TP3)
Determination value of train type CC = (representative distance A−representative distance C) − (CC_TP1-CC_TP3)
It is.

  The processing according to s26 is the configuration according to the determination value calculation unit referred to in the present invention.

  The calculation unit 16 performs a train type determination process for determining the type of train captured in the distance frame image processed this time using the determination value calculated in s26 (s27).

  In the above example, the determination values are calculated for three types of trains AA, BB, and CC. As described above, the determination value is calculated for each type of train to be determined. In addition, a plurality of determination values are calculated for each type of train (a determination value is calculated for each predetermined combination of spaces).

  FIG. 7 is a flowchart showing a train type determination process. In this train type determination process, for each type of train to be determined, it is determined whether there is a possibility that the train imaged in the distance frame image processed this time is a train of that type. The type of the train imaged in the distance frame image processed this time is finally determined from the types determined to be present.

  The calculating part 16 determines whether each of the determination value (determination value calculated by s26) concerning the target classification train is appropriate (s41). The determination concerning s41 is determined to be appropriate if the determination value is larger than a predetermined lower limit value and smaller than the upper limit value, and is determined to be inappropriate in other cases. In s41, even if the determination value is the upper limit value or the lower limit value, it may be determined to be appropriate. The lower limit value and the upper limit value may have the same absolute value (for example, the lower limit value is −10 mm, the upper limit value is 10 mm), or the absolute values are different (for example, the lower limit value is −5 mm, The upper limit may be 10 mm).

  If all the determination values calculated in s26 for the target type train are appropriate, the calculation unit 16 determines that the train captured in the distance frame image processed this time may be the type of train. (S42, s43). On the other hand, if at least one of the determination values calculated in s26 for the type of train is inappropriate, the calculation unit 16 may have the train captured in the distance frame image processed this time as that type of train. (S42, s44).

  The calculating part 16 repeats the process of above-mentioned s41-s44 about all types of train to determine (s45).

  The calculation unit 16 determines that the train type cannot be determined if there is no train type determined to be possible in s43 (s46, s48). For example, when the back of the train is not captured in the distance frame image captured by the TOF camera 2, the train type cannot be determined.

  On the other hand, if there is a train type determined to be possible in s43, the calculation unit 16 calculates the sum of the squares of each determination value for each type determined to be possible, and the calculation is performed here. The type of train having the smallest sum of squares of the respective determination values is determined as the type of train imaged in the distance frame image processed this time (s46, s47).

  By providing the processing according to s47, it is possible to sufficiently ensure the determination accuracy of the train type even if there is a type of train having a similar back surface shape (or front surface shape). Moreover, in s47, you may use the sum total of the absolute value of each determination value instead of the sum total of the square of each determination value.

  Returning to FIG. 6, if the train type can be determined in s27, the calculation unit 16 estimates the train position (s28, s29). If the train type cannot be determined in s27, the calculation unit 16 returns to s21. In s29, the distance (representative distance A + correction value) obtained by adding the correction value associated with the train type determined in s29 from the representative distance A is estimated as the train position at the station platform. The train position estimated in s29 is based on the installation position of the TOF camera 2.

  The calculating part 16 determines whether the train position in the station platform estimated in s29 is the target stop position prescribed | regulated previously to the station platform (s30). The calculation unit 16 stores the distance from the installation position of the TOF camera 2 to the target stop position. The calculating part 16 performs the determination concerning s30 by comparing the train position estimated by s29, and the memorize | stored target stop position.

  The output unit 17 outputs the determination result determined by the calculation unit 16 in s30 (s31). The determination result output from the output unit 17 is used for lighting control of an indicator lamp installed at a position that can be easily seen by train operators, station staff, and the like. Therefore, a train operator, a station staff, or the like can confirm whether the position of the train on the station platform is the target stop position by confirming whether the indicator lamp is turned on or off. Further, as apparent from the above description, the train position determination device 1 repeatedly executes the train position determination process shown in FIG. 6 even when the train is not stopped. The driver can be recognized. That is, navigation of the position where the train is stopped can be performed for the train operator.

  Next, the stop determination process starting at s4 will be described. FIG. 8 is a flowchart showing the stop determination process. This stop determination process is a process using the visible frame image input to the visible image input unit 13.

  The visible image processing unit 14 generates a difference image (so-called inter-frame difference image) of a pair of visible frame images continuous in time input to the visible image input unit 13 (s51). The inter-frame difference image is an image representing a change between a pair of visible frame images, as is well known.

  The visible image processing unit 14 determines whether or not the inter-frame difference image whose degree of change is smaller than a predetermined threshold value continues for a predetermined number of frames (for example, 5 frames) (s52). If the visible image processing unit 14 determines that the predetermined number of frames have continued in s52, it determines that the train has stopped (s53). On the other hand, if the visible image processing unit 14 determines that the predetermined number of frames are not continued in s52, it determines that the train is not stopped (s54).

  Thus, the stop position determination system according to this example can determine the type of train. Further, since the position of the train is detected according to the type of train determined here, the position of the train can be detected with high accuracy.

  Further, as described above, since the TOF camera 2 is mounted above the station platform, it is possible to prevent dust and dirt from adhering to the imaging lens of the TOF camera 2. In addition, maintenance (cleaning work) for wiping off dust and dirt adhering to the imaging lens of the TOF camera 2 can be performed regardless of the train operation status.

  In addition, when the train is not stopped at the target stop position, the opening / closing of the platform door is prohibited, so that passenger safety can be further enhanced.

  In addition, the TOF camera 2 in the above example may be replaced with another device that can obtain a distance image. For example, as disclosed in Patent Literature 1, a laser beam is scanned and a reflected light is detected to obtain a distance image of an area scanned with the laser beam, and two imaging units having parallax are provided. It may be replaced with a stereo camera or the like.

  Moreover, the approach determination of the train to the station platform may be configured to be detected from the received light intensity image captured by the TOF camera 2 or the change between frames in the distance image, or may be captured by the video camera 3. The visible light image may be detected from a change between frames. In this way, the ultrasonic sensor 4 can be dispensed with. However, in this case, the TOF camera 2 or the video camera 3 that captures an image used for the determination of the train approach to the station platform is configured to always perform imaging and process the captured image.

  In addition, the stop determination process illustrated in FIG. 8 may be performed by generating a difference image (a so-called inter-frame difference image) of a pair of time-series distance frame images captured by the TOF camera 2. . In this case, the video camera 3 can be dispensed with.

  Moreover, in the above example, the case where the number of spaces for detecting the representative distance is three is taken as an example, but may be four or more. In the above example, two combinations of spaces for calculating determination values are used. However, three or more combinations of spaces for calculating determination values may be used. Further, by increasing the space for detecting the representative distance and increasing the total number of combinations of spaces for calculating the determination value, it is possible to improve the determination accuracy of the train type.

  In addition, as described above, the TOF camera 2 may be attached at an angle that captures the front of the train, not the back of the train. The correction value registered in the train type table may be a value for the representative distance B or a value for the representative distance C.

1-train position determination device 2-TOF camera 3-video camera 4-ultrasonic sensor 11-distance image input unit 12-distance image processing unit 13-visible image input unit 14-visible image processing unit 15-detection signal input unit 16 -Calculation unit 17-Output unit

Claims (6)

A distance image input unit for inputting a distance image obtained by imaging the back surface or the front surface of the moving body traveling along the track at an angle from above;
The distance image input to the distance image input unit is processed, and for each space divided into three or more so as not to overlap with the height direction of the moving body traveling along the track, the distance to the moving body in the space is determined. A distance image processing unit for detecting a representative distance;
A determination value calculating unit that calculates a determination value based on a distance difference between representative distances detected by the distance image processing unit for each space combination set in advance;
A mobile body type that determines a type of a mobile body that is captured in a distance image input to the distance image input unit by using a plurality of determination values calculated by the determination value calculation unit for each preset combination of spaces A moving body management apparatus comprising: a determination unit; The determination value calculation unit calculates a determination value for each type of moving object in each space combination,
The mobile body type determination unit determines, for each mobile body type, whether the determination value calculated for the mobile body of the type is appropriate for the mobile body of the type in each space combination, Even in the combination of spaces, it is determined that the type of the moving body determined that the calculated determination value is appropriate is the type of the moving body captured in the distance image input to the distance image input unit. The mobile body management apparatus according to claim 1.   Based on the representative distance detected in any space by the distance image processing unit and the type of the moving object determined by the moving object type determining unit, it is determined whether or not the moving object is located within a predetermined allowable range. The mobile body management device according to claim 1, further comprising a position determination unit. A reference distance storage unit that stores a reference distance of each space for each type of mobile body,
The determination value calculation unit stores, for each combination of preset spaces, the distance difference between the representative distances detected by the distance image processing unit for the space related to the combination and the reference distance storage unit for the space related to the combination. The mobile body management apparatus according to claim 1, wherein a determination value based on a difference in reference distances to be calculated is calculated. The distance image that is input to the distance image input unit and picks up the back image of the moving body that travels along the track at an angle from above or the front surface is processed, and the height of the moving object that travels along the track is processed. A distance image processing step for detecting a representative distance to a moving object in each space divided into three or more spaces so as not to overlap;
A determination value calculation step for calculating a determination value based on a distance difference between representative distances detected in the distance image processing step for each space combination set in advance;
A mobile body type that determines a type of a mobile body captured in a distance image input to the distance image input unit using a plurality of determination values calculated in the determination value calculation step for each preset space combination A moving body management method comprising: a determination step. The distance image that is input to the distance image input unit and picks up the back image of the moving body that travels along the track at an angle from above or the front surface is processed, and the height of the moving object that travels along the track is processed. A distance image processing step for detecting a representative distance to a moving object in each space divided into three or more spaces so as not to overlap;
A determination value calculation step for calculating a determination value based on a distance difference between representative distances detected in the distance image processing step for each space combination set in advance;
A mobile body type that determines a type of a mobile body captured in a distance image input to the distance image input unit using a plurality of determination values calculated in the determination value calculation step for each preset space combination A moving body management program for causing a computer to execute a determination step.

Images