JPS6389981A - Train position detecting device - Google Patents

Abstract

PURPOSE:To detect the range section of the stop position of a train by an ordinary image processing device and camera, by arranging plural markers within the range of the stop position of the train, and performing position detection by utilizing the fact that the marker is hidden by the advancement of the train. CONSTITUTION:The camera 20 picks up the image of a periphery including a train number displaying part 100 being mounted on the rear of the train 10, and transmits the image to the image processing device 40. The image is classified according to the advancement of the train to three cases: a case that neither of the markers 21 and 22 and detected, the case that either the markers is detected, and the case that both markers are detected, and the image is written in a picture memory in the device 40, however, the periphery of the marker is enclosed by a white frame, and the inside of the frame is painted out in black, and it is binarized at a binarization circuit, and only a black area is taken out, thereby, the presence of the frame can be confirmed. At the device 40, the position of the train can be discriminated by a marker discrimination circuit which compares a shape with a data stored in advance, and decides that the error of areas is within a prescribed range, through a marker readout circuit which designates and extracts the area where the marker exists.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Application Field) The present invention provides a train position detection device that images a stopped train with a camera and determines the classification of the train's stopping position based on the number of recognizable signs. Regarding.

(Conventional technology) Because trains operate based on a schedule, trains are equipped with train number indicators that display symbols to distinguish between the service number on the schedule, the name of the company to which it belongs, the depot, etc. We are prepared.

On the other hand, the transportation command center that monitors and controls train operations is
The train number is read from the train actually being operated and is compared with the service number on the train schedule to monitor whether the train is operating according to the schedule and to control the points at junctions and junctions.

By the way, conventionally, in order to read the train number, a method has been used in which a train number setting device and an overboard are provided on the train, and a wayside device and a train number reading device are provided as ground equipment. However, since this method requires onboard equipment, the route configuration is limited to the same electric railway company, and it has been difficult to implement it on lines that share mutual access with other companies' lines.

In order to solve this problem, we do not install any onboard equipment, and instead input the screen captured by a camera of the surrounding area including the train number display on the front and rear of the train, and detect the subtitles of the train number display from that screen. However, there is a method of recognizing train numbers using image processing technology.

Below, a method for determining the approximate range on the input screen of the train number display section will be explained using an example using a train passing detection sensor.

First, we will discuss an image processing method based on variations in the stopping positions of trains on platforms.

An image processing type vehicle number identification device that images a train number display attached to the front or rear of a train with a camera, imports this captured screen into an image processing device, and recognizes the train number characters through image processing. From the standpoint of the image processing device, if the train number display section is always present at a constant position on the image, it is easy to take out the display section.

Therefore, all you need to do is to use a camera to capture images while the train is running.
This requires a camera with a high-speed (1/1,000) shutter and an irradiation device to compensate for the lack of illuminance during imaging.

However, cameras and irradiation devices currently available on the market do not provide satisfactory performance, so advanced technological development is required, equipment costs are high, and practical application is difficult. Therefore, if it is decided to take an image with a camera while the train is stopped, a commercially available camera can be used and an irradiation device is not required in the tunnel, so it is easy to implement.

However, on the other hand, the position of the train number display section on the screen differs depending on the variation in the train stopping position, which complicates image processing and can lead to misidentification.

Therefore, if the train stop position classification is detected in advance by another means, the positions that exist on the screen of the train number display section can be limited, and image processing becomes easier.

Here, a conventional train position classification method using a train passing sensor will be explained with reference to FIG.

In FIG. 6, two train passing sensors 50 (A, B
) and respective reflectors 51 are provided, and the train stopping range 52 is
Arrange the inside so that it is roughly divided into three parts. The camera 20 is installed at a position where it can cover the train stopping range 30.

−4= When the train passing sensor 50 is not present, the reflector 51
Since the signal sent out is returned, it is detected that there is no train.

Now, if the train 10 enters the platform 11 from the direction of the arrow, the signal transmitted by the train passing sensor 50 will be blocked by the train 10 and will not return. The vehicle is divided into j sections shown in FIGS. 6(a) to 6(c) depending on the stopping position.

That is, the train number display section 100 is located on the right side when viewed from the arrow in FIG. 6(a), and approximately in the center in FIG. 6(b).
In Figure (c), the image is taken on the left side, so the train number display section 1
The position of 00 can be narrowed down to three parts.

In this way, the range is limited by the train stop position classification, using the advantage of being able to obtain a target image at a fixed position, as in the case of imaging while a train is running.

(Problems to be Solved by the Invention) As mentioned in Part 2, since train passing sensors and reflectors are used to divide train stopping positions, in addition to equipment costs, wiring work, setting of installation locations, and images are required. The input interface to the processing device uses a photoelectric sensor.

Therefore, in order to remove dirt caused by sticky dust generated by brakes when a train stops, etc., maintenance must be increased to compensate for this, and there are problems with the service life of relays in sensors that operate more than 200 times a day. The use of sensors is a major obstacle to practical application.

The present invention is an image processing type train position detection device that overcomes the difficulties of conventional devices and detects train stop position range divisions using a camera and image processing device used in a normal image processing device. Its purpose is to provide.

(Means for Solving the Problems) The present invention provides for installing a camera at a position within the range of the stop position of a train stopping at a platform where it can image the surrounding area including the train number display section attached to the front or rear of the train. For example, two signs installed within the range of the train's stopping position are recognized when the train enters the platform, the images of the signs are blocked by the incoming train, and a certain period of time after the train arrives at the platform. , there are three cases: cases in which both signs can be detected, cases in which only one sign can be detected, and cases in which no sign can be detected. The area where the sign exists is specified and extracted by executing a conversion circuit that converts it into binary values of white and black by converting it with a binarization threshold depending on the density of the sign and making the sign part stand out. The train position is distinguished by a sign determination circuit that compares the shape with pre-stored data through a sign extraction circuit and determines that the area is within a certain error range.

(Function) Since the images captured by the camera near the sign are binarized in advance, the position is detected by taking advantage of the fact that the sign is obscured by the train when the train approaches.

(Embodiment) The structure and operation of an embodiment of the present invention are shown in FIGS. 1 to 4.

FIG. 1 is a block diagram of an embodiment of the present invention.

An image processing device 40 that detects the train position, and a camera 2 that captures an image of the train number display section 100 of the train 10 that is stopped at the platform.
0, and signs 21 and 22 indicating signs A and B for distinguishing at which position within the camera imaging range 30 the train number display section 100 is located.

The program of the image processing device 40 for detecting the train stop position by section detects that the train 10 has entered the platform 11 and requests the next process after the train 10 has stopped on the platform, as shown in FIG. Shown in (a). Figure 4 (b) shows a block diagram for determining the stop position classification after the train has stopped.
).

First, FIG. 4(a) will be explained in detail.

As shown in FIG. 4(a), when the train 10 enters the platform 11, it is sufficient to detect that the sign A 21 is hidden behind the train 10 and is no longer visible. The input image input circuit 61 writes the image into the image memory. The image written in the image memory is retrieved, and the sign recognition processing block diagram shown in FIG. 4(c) shown later is called to determine whether recognition is possible. If the sign recognition processing block determines in block 62 that the sign A has changed from recognizable to unrecognizable, the time block for calling the train stop position classification processing block is executed after the time when the train 10 is expected to stop at the platform 11. The circuit 63 is activated to complete this block processing group. Furthermore, if the sign 21 can be recognized, the image input circuit 61 is executed again.

When a certain period of time has elapsed since the signs A and 21 were hidden behind the train 10 and could no longer be detected, the image input circuit 61 shown in FIG.
The camera-captured images 31 shown in FIGS. (a) to (c) are written into the image memory. It is determined in block 62 whether the sign A can be recognized from the screen captured in the image memory by calling the above-mentioned sign recognition processing block diagram in FIG. Train 10 is at the position shown in Figure 1 (C).
This fact is transmitted to the train number recognition process, which is not described in the present invention.

In addition, if recognition is not possible, block 62' is used to mark B.2.
2 can be recognized by calling the above-mentioned block diagram of the sign recognition process shown in FIG. is determined to exist. If recognition is not possible, the answer is NO, and it is determined that the train 10 was present at the position shown in FIG. 1(a), and the information is transmitted to the next train number recognition process where no description is made.

Therefore, the position on the screen of the train number display section 100 consisting of FIG. 2(C), (b), and (a) can be divided into three sections.

Next, in the block diagram of the sign recognition process shown in FIG. 4(c), block 71 is first activated, and the filtering circuit smoothes the density and removes noise of small pixels of about 1 to 2 dots, and then block 72 is executed. In order to make it easier to take out the sign, if we surround the sign in advance with a white frame as shown in Figure 2(c) and fill the inside of the frame with pure black, the binarization circuit will convert the sign to 2 at a threshold value close to pure black. Convert it into a digitized value and extract only the black area including the marker part.

Then, in block 73, only the area where the sign exists is extracted, and the next block 74 is activated.

Block 74 compares the areas of the shapes in order to investigate the similarity between the extracted black area and the shape stored in advance, and if it is within the tolerance, it is recognized as a sign and exits with Yes. If the error is not within the allowable range, the process exits to the No exit and returns to the called plotter.

In this way, the identification process of the marker part is executed.

In this example, the camera is placed within the range of the train stop position.
There is no need for any special equipment other than placing the sign at a position where the sign can be imaged and attaching a plurality of signs within the range of the train stop position, so there is little resistance when installing the sign.

In addition, since train stop positions can be classified more precisely, it is possible to sufficiently narrow down the position in response to changes in the position of the train number display section on the screen due to variations in train stop positions. It is possible to support the execution of character extraction processing that is almost the same as when an image is captured.

The following means can be considered as other embodiments of the present invention.

1) As shown in Figure 5 (a), the signs C 23 are installed consecutively in the width direction, divided into predetermined widths, and the signs C23 that form a pattern can be used to more precisely classify train stopping positions. Can be divided.

2) The shape of the mark may be arbitrary, and a plurality of marks may be combined.

3) As shown in Figure 5(b), the signs 24 indicating the train 10 are installed continuously over the entire width above the train roof, and the sign A 21 for detecting the approach of the train 10 is placed on the other side. Sign D-2 with a continuous pattern divided into predetermined widths
4, the pattern 26 that overlaps with the perpendicular line 25 extending upward from the right side of the train 10 is obtained, and by using it for the train position classification of the train 10, the train approach detection and train stop position detection are each made into independent signs. is also possible.

= 12 = [Effects of the Invention] A photoelectric sensor can be cited as a method for accurately detecting a train stop position. However, it is susceptible to dirt and is generally not suitable for use on railways due to its reliability.

However, there are no alternative methods, and although the electromagnetic induction method is highly reliable, its detection accuracy is low and it is not suitable.

From this point of view, the image processing type is suitable as a method that can obtain accuracy equivalent to that of a photoelectric sensor and more precisely detect classifications of train stop positions.

Thus, according to the present invention, one advantage is that the camera and image processing device used to identify train numbers through image processing can be used as is, and two, by simply attaching only the signs, it is possible to use other devices. It is easy to implement because it does not require any incidental equipment.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention;
Figure 2 is a train stop position division diagram, Figure 3 is a camera image capture screen diagram, Figure 4 is a program block diagram in the image processing device, Figure 5 is a sign diagram of other embodiments, and Figure 6 is a conventional It is an explanatory diagram of train stop position classification. 10...Train, 11...Platform, 20...Camera, 21...Sign A122...Sign B130...Camera imaging range, 40...Image processing device, 50...Train passing Sensor, 51...Reflector, 100...Train number display section. Applicant's agent Mr. Sato Figure 1 (a) (b) Figure 5

Claims (1)

[Claims] A camera mounted at a position within the stopping position range of a train stopping at a platform so as to be able to image the surrounding area including a display section such as a train number display section or a destination display section attached to the front or rear side of the train. , a plurality of signs installed at equal intervals within the range of the train's stopping position, and means for capturing into an image memory an image taken by a camera that determines whether the train enters the platform and blocks the signs and becomes invisible; A means for setting a binarization threshold value from the captured image according to the density of the mark stored in advance in a data area, converting information in the image memory to a binarized image, and determining the area and shape where the mark is present. A means for creating a sign area by leaving only the area where the sign exists and deleting the others from a binarized image based on the area where the sign exists, which is stored in the data area in advance; The area ratio of the shape is determined and if it is within the allowable error, it is recognized as a sign.A sign judgment process is performed periodically, and if the sign judgment process does not recognize it as a sign, it is assumed that it is blocked by a train and the train arrives at the platform. Image processing formed from means for recognizing a plurality of signs after a lapse of time by a sign determination process, and detecting the train position in units of sign intervals depending on whether the sign can be recognized at all or when all the signs can be recognized. A train position detection device comprising: a device;