JP2018086872A - Train information discrimination device - Google Patents

Abstract

A technique for determining a stop position of a train and determining train information by image processing is provided. As a facility on the train 90 side, a strip-like tape 30 serving as an identifier is attached to the platform side surface of the train 90. As equipment on the ground side, the train information determination system 10 includes a small camera 11, a stop position detection unit 12, and a home fence control unit 13. The camera 11 is installed on the home fence 99, and the angle of view and the like are set so that the strip-shaped tape 30 is included in the captured image. The stop position detection unit 12 detects the strip tape 30 by double circular operator processing (image processing) based on the image captured by the camera 11, and performs stop position detection processing and train information determination processing. . [Selection] Figure 1

Description

  The present invention relates to a train information discriminating apparatus, and more particularly to a train information discriminating apparatus using an image processing technique.

  As a technique for detecting a fixed position stop position of a train, a technique using a ground element (transponder) and a vehicle upper element is known (for example, see Patent Document 1). In the technique disclosed in Patent Literature 1, a ground element having a coil is installed before and after the position where the train stops, and the vehicle body is mounted on the vehicle side and the power wave transmitted from the vehicle element is received. The fixed position stop position is detected based on the output voltage difference between the two ground elements.

  In addition, as a technique for detecting train information, there is a technique for controlling a home fence using train attribute information (information such as a train number and the number of trains) obtained from a ground element (transponder). When there is no ground element, after the train stops, the platform fence at the position corresponding to the number of trains is opened and closed by operating the conductor.

JP 2001-151112 A

  By the way, in the above-described technology, a ground element is placed at a position where the train stops, a vehicle upper element is mounted on the vehicle side, and a position where the train stops is detected by exchanging power waves between the ground element and the vehicle upper element. ing. However, these ground and vehicle elements are expensive. Moreover, since it is necessary to mount a vehicle upper body on all traveling vehicles, there is a problem that it takes time for maintenance.

  This invention is made | formed in view of the above situations, Comprising: It is providing the technique which solves the said subject.

The train information discriminating apparatus of the present invention determines whether or not the identifier is included in an image captured by the image capturing unit that captures a band-shaped identifier attached to a train, and the image captured by the image capturing unit. And a stop position detector that determines a stop position of the train according to the position of the identifier in the image when detected.
The stop position detection unit may extract a feature amount of the identifier and determine a type of train according to the feature amount.
In addition, the stop position detection unit may detect the identifier using a double circular operator process.
Moreover, you may provide the home fence control part which controls the operation | movement of a home fence in one formation unit or a vehicle unit according to the detection result of the said stop position detection part.

  According to the present invention, a technique for determining a stop position of a train or determining train information can be realized by image processing.

It is a figure which shows schematic structure of the train information determination system based on this embodiment. It is a functional block diagram of a train information judging system concerning this embodiment. It is a figure explaining the technique of the fixed position determination process of the train based on this embodiment. It is a figure explaining the technique of the fixed position determination process of the train based on this embodiment. It is a figure explaining the outline | summary of the process using a double circular operator based on this embodiment. It is a flowchart explaining the process by the stop position detection part based on this embodiment.

  Next, modes for carrying out the present invention (hereinafter, simply referred to as “embodiments”) will be specifically described with reference to the drawings.

  In the present embodiment, a technique for detecting a stop position of a train and detecting train information by image processing without using a ground element and a vehicle upper element will be described.

  FIG. 1 is a diagram illustrating a schematic configuration of a train information determination system 10 according to the present embodiment, and illustrates a state viewed from the upper surface of a train 90 that enters a platform. FIG. 2 is a block diagram illustrating a schematic configuration of the train information determination system 10.

  As equipment on the train 90 side, a strip-like tape 30 serving as an identifier is attached to the platform side surface of the train 90. As equipment on the ground side, the train information determination system 10 includes a small camera 11, a stop position detection unit 12, and a home fence control unit 13.

  The camera 11 is installed on the home fence 99. The camera 11 has an angle of view and the like so that the striped tape 30 is included in the captured image. The stop position detection unit 12 includes an image processing unit 21 and an arithmetic processing unit 22.

  The image processing unit 21 performs stop position detection processing by image processing based on the image captured by the camera 11. The arithmetic processing unit 22 performs a train information discrimination process based on the detection result of the image processing. Although details will be described later, the stop position detection unit 12 determines the stop position and train information of the target train 90 by determining the position and size of the strip tape 30 on the side surface of the train 90. The home fence control unit 13 controls the opening / closing operation of the home fence 99 and a movable step (not shown) based on the processing result of the stop position detection unit 12.

  When the belt-like tape 30 detected by the stop position detection unit 12 is detected in the image continuously for the specified number of frames at the same position, the stop position detection unit 12 determines that “the train 90 has stopped”.

  3 and 4 are diagrams for explaining the technique of the fixed position determination process of the train 90. FIG. The fixed position stop determination process of the train 90 will be specifically described with reference to these drawings.

As shown in FIG. 3, the difference between the position of the white line (band-shaped tape 30) detected by the train 90 and the train stop position target is measured, and the determination of the train position is performed as follows.
(1) When the belt-like tape 30 is detected within 150 mm before and after the train stop target position, it is determined as “JUST position” and stopped at a fixed position.
(2) When the belt-like tape 30 is detected within 550 mm before and after the train stop target position, it is determined that the vehicle has stopped in the stop allowable range.
(3) If the strip tape 30 cannot be detected within 550 mm before and after the train stop target position, it is determined as “OVER position / SHORT position”, that is, “not stopped within the allowable range”, and the determination result is the home fence. Not transmitted to the control unit 13. That is, the home fence 99 and the movable step are not controlled.

  As shown in FIG. 4, the camera 11 is installed in a direction in which a stop allowable range centered on the stop target position is visible in the camera field of view. Here, about 1500 mm before and after the stop target position is set as the camera field of view range (shooting range).

  Next, determination of information on the train 90 (train information) will be described. The discrimination of the train information is performed by detecting the feature amount (size, color, etc.) of the strip-shaped tape 30. Train information includes the number of trains, number of doors, train type, and the like. By discriminating these train information based on the size of the strip tape 30, etc., desired information is acquired, and the control of the platform fence 99 is executed.

For example, the size of the belt-like tape 30 is specified and determined as follows.
(1) The number of knitting vehicles is defined by changing the length of the belt-like tape 30.
(2) The number of vehicle doors is defined by changing the thickness of the belt-like tape 30.
(3) The types of trains include “a train that stops and treats a customer”, “a train that stops but does not treat a customer (forward train, trial run train, etc.)”, and “pass train”.
(4) Among “stopped trains”, “trains handled as customers” and “trains not handled as customers” are distinguished by changing the color of the strip tape 30.
(5) If it is determined that the train is “stopped but not treated as a customer”, the determination result is not transmitted to the home fence control unit 13. That is, the home fence 99 and the like are not controlled.
(6) Regarding the determination of “passing train”, in the camera field of view, if the target strip tape 30 is not shown in consecutive frames, it is determined that the vehicle is traveling faster than the specified speed, and the passing train Judge.
(7) As one of the detection methods of the strip-shaped tape 30, the technique of Unexamined-Japanese-Patent No. 2013-131174 (known technique 1) is used. This method performs feature extraction by a double-circle operator. Robust strip shape detection in the real environment is possible with a small amount of calculation. Note that the publicly known technique 1 also discloses a bird's-eye view conversion process that eliminates / adjusts the influence of the angle of view set in the camera 11, and can also be used in this embodiment.

  FIG. 5 is a diagram for explaining the outline of processing using double circular operators (operator processing), and FIG. 5A shows an example of setting inner and outer circles constituting the double circular operator. FIG. 5B shows an example in which a double circular operator rides on the white line. Here, the width of the strip-shaped tape 30 is 50 mm, and the inner circle has a standard size of 9 pixels and the outer circle has a standard size of 15 pixels. Depending on the width of the strip tape 30, the size of the double circular operator is determined. In the example of FIG. 5B, it is determined that the center of the double circular operator is located at the approximate center of the width of the white line (band-shaped tape 30). At this time, one straight line is detected by Hough transforming the row of white line candidate points.

A specific example of the setting of the strip tape 30 is shown below.
(1) The number of knitting is defined by the length of the belt-like tape 30.
For example, when the number of knitting is 10, 8, 6, 6, or 4, the length of the strip tape 30 is determined according to the number of knitting. The length and the number of knitting may be proportional. When the strip tape 30 is measured, the position (coordinates) of the start point and the end point of the strip tape 30 are measured, so that the length can be easily measured.

(2) The number of doors is defined by the thickness (width) of the strip tape 30.
When the white line is thicker than 50 mm, the white line candidate point is detected as a line having a width of 2 pixels or more. As a result, a plurality of line segments can be detected at the time of line segment extraction by Hough transform.

(3) The stop train is identified by the color of the belt-like tape 30.
When the strip-shaped tape 30 is white, it is determined as “a train handled as a customer”, and when it is red, it is determined as “a train not handled as a customer (forward train, trial operation, etc.)”.

(4) Whether the train is a passing train is determined based on the recognition speed of the belt-like tape 30.
Within the field of view of the camera 11, after recognition of the strip tape 30, if the target strip tape 30 cannot be recognized in the next frame, or if the vehicle is traveling at a specified speed or more, it is determined as a passing train.

  FIG. 6 is a flowchart showing a processing flow of the stop position detection unit 12, and the processing of the stop position detection unit 12 will be described collectively based on the flowchart.

  The image photographed by the camera 11 installed on the home fence 99 is taken into the image processing unit 21 of the stop position detection unit 12 (S10). By this process, an image of the side surface of the train 90 is created as an original image.

  The image processing unit 21 performs overhead conversion processing on the original image (S12). For example, in the case where the video shot by the camera 11 is a video captured obliquely downward, the distance from the strip tape 30 and the width of the strip tape 30 cannot be accurately grasped even if the original image is used as it is. . Therefore, the image processing unit 21 creates a bird's-eye view image of the train side by converting the viewpoint to an image in which the shooting range is looked down from directly above. Specifically, the size is changed so that one pixel of the image becomes an actual 5 mm, for example.

  Subsequently, the image processing unit 21 performs a smoothing process on the overhead image (S14). Here, in order to suppress variation in luminance, gray scale conversion and smoothing using a median filter are performed. By this process, a smoothed bird's-eye view image is created.

  After smoothing, the image processing unit 21 performs the above-described operator processing on the smoothed bird's-eye view image (S16), and performs white line candidate point extraction processing (S18). As a result, a white line candidate point image of a binary image is obtained.

  The image processing unit 21 further performs a Hough transform process on the white line candidate point image (S20). Linear interpolation is performed, and a white line (band-like tape 30) is assumed to be a certain length or more as a line as a Hough conversion parameter. Interpolation is also performed on white line candidate points on the white line that has been closed, and a white line is detected (S22).

  Next, the image processing unit 21 calculates the coordinates of the white line (band-shaped tape 30) in the reference coordinate system (S24). Here, the reference coordinate system is a coordinate system with the camera 11 as a reference, and the coordinate system in the overhead image is an image coordinate system. The reference coordinate system and the image coordinate system are calculated by calibration processing using a chessboard reference point or the like when the camera 11 is installed. By this processing, the fulcrum and end point of the white line candidate point expressed in the reference coordinate system, that is, the positions of both end portions of the strip-shaped tape 30 are determined.

  When the position of the strip tape 30 is determined by the image processing unit 21, the arithmetic processing unit 22 performs a stop position deviation calculation process (S26), and calculates the distance between the position of the strip tape 30 and the stop target position. Based on the calculation result, the arithmetic processing unit 22 performs a fixed position determination process (S28). That is, as described above, the arithmetic processing unit 22 determines whether the position where the train 90 is stopped is “JUST position”, “stop allowable range”, “OVER position”, or “SHORT position”.

  When the determination result is “JUST position” or “stop allowable range”, the arithmetic processing unit 22 transmits the determination result to the home fence control unit 13 (S30).

  In the above flow, only the position of the belt-like tape 30 is detected, but the length of the belt-like tape 30 can be detected by determining the positions of both ends of the belt-like tape 30. Moreover, the color of the strip-shaped tape 30 can also be detected from the image before the frame scaling. Further, the width of the strip tape 30 can be detected from the relationship between the size of the double circular operator and the width (number of pixels) of the white line candidate points.

  As mentioned above, according to this embodiment, since the strip | belt shape (band | belt-shaped tape 30) stretched on the camera 11 and the train 90 is detected by an image processing technique, without using a ground element and a vehicle upper element, compared with the conventional equipment, It is inexpensive and the train 90 (vehicle) can be easily modified and adjusted. In addition, since the detection process of the strip tape 30 by image processing using a double circular operator is low in load, it is easy to secure time for other processes necessary to ensure safety. is there.

  The present invention has been described based on the embodiments. This embodiment is an exemplification, and it is understood by those skilled in the art that various modifications are possible for the combination of each of those components, and such modifications are also within the scope of the present invention.

  For example, in the above-described embodiment, the fixed position determination process is performed by detecting the belt-like tape 30 of the head train 90 (vehicle) with one camera 11. For example, a plurality of cameras 11 may be arranged, and the belt-like tape 30 may be detected in units of vehicles. By setting it as such a structure, the operation control of the home fence 99 can be changed flexibly in one formation unit or a vehicle unit.

DESCRIPTION OF SYMBOLS 10 Train information determination system 11 Camera 12 Stop position detection part 13 Home fence control part 21 Image processing part 22 Calculation processing part 30 Strip tape 90 Train 99 Home fence

Claims (4)

Photographing means for photographing the band-shaped identifier attached to the train;
It is determined whether or not the identifier is included in the image captured by the imaging unit, and when the identifier is detected, the stop position of the train is determined according to the position of the identifier in the image. A stop position detector for determining;
A train information discriminating apparatus comprising:   The train information discriminating apparatus according to claim 1, wherein the stop position detecting unit extracts a feature amount of the identifier and discriminates a type of train according to the feature amount.   The train information discriminating apparatus according to claim 1 or 2, wherein the stop position detecting unit detects the identifier using a double circular operator process.   The train according to any one of claims 1 to 3, further comprising a home fence control unit that controls the operation of the home fence in units of one train or a vehicle according to a detection result of the stop position detection unit. Information discrimination device.

Images