JPH0288358A - Obstruction detecting device for railway crossing - Google Patents

Abstract

PURPOSE:To enable the proper detection of all types of automobiles, reduce the deposit of brake powder, dusts and the like generated during the travel of a train and lessen erroneous detection by mounting cameras obliquely downward at a high position. CONSTITUTION:A plurality of cameras 51 to 60 are positioned toward crossing gates 11 to 14 at an intermediate high position between double track lines 21 and 22 within a railway crossing for taking the photo of a crossing area including the track lines 21 and 22. Each camera takes the photo of the same place and the image signal thereof is taken into an image processing device. Furthermore, two images of the same place is superimposed with program processing within the image processing device and a shade difference in the same picture element is obtained throughout picture elements. When the number of picture elements having a shade difference above an allowable range exceeds the predetermined value, an obstruction is judged to be existing and the lamps of illuminating signals 73 and 74 arranged along the lines 21 and 22 are made to light, thereby giving a warning to a motorman on board a train entering a railway crossing.

Description

[Detailed Description of the Invention] [Purpose of the Invention (Industrial Application Field) This invention uses a plurality of cameras installed at specific locations within a railroad crossing to monitor the safety of trains passing through the railroad crossing. The present invention relates to a level crossing obstacle detection device that inputs an image and detects the presence of an obstacle by image processing.

(Prior Art) As a railroad crossing hazard detection device, the one shown in FIG. 6G) is conventionally known. In this conventional example, the orbit 21
．． Level crossing breakers 11 to 14 are installed to protect the general road side that intersects with 22, and the general road is divided by white lines 142 to form incoming lanes.

In order to detect obstacles, a light emitter 81.81- and a light receiver 82.82' are installed on both sides of the track 21.22. When the light beam 83 is received by the opposite light receiver 82.82'', it is assumed that there is no obstacle, and the light emitting signals 73 and 74 installed along the line remain off, and when an obstacle is detected, the light beam 83 is received. 82, 82'' will no longer receive light, the light emitting signals 73, 74 are turned on to notify the train driver entering the level crossing and urge him to stop.

However, it is possible that the train may be detected as an obstacle while an oncoming train is passing through the level crossing, so in order to prevent it from becoming an obstacle, it is necessary to check whether there is any problem in passing through the level crossing on the track on which the own train is running. When traveling on the track 21, obstacles are detected only by the light emitter 81 and light receiver 82 placed oppositely along the track 21 of the own train, and the light emitting signal 74 is turned on.

(Problem to be Solved by the Invention) However, in such conventional level crossing obstacle detection devices, both the emitter and receiver are installed at relatively low positions (7451 degrees above the road surface), Due to dirt, flakes, brake powder, etc. adhering to the light-emitting and light-receiving surfaces of the device, the device will gradually remain in a detection state in which no light is received, and if cleaning intervals are not strictly followed, normal operation will soon become impossible to maintain. Ta.

In addition, the detection points are only on both sides of the track, and if the leading end or dominant end of the vehicle that is an obstacle is stopped with its tip or dominant end protruding onto the track, the light beam may be detected under the floor of the vehicle, especially in large vehicles. There was a problem that it could become impossible to detect obstacles through the light.

Furthermore, since it is a light beam type, it is an absolute requirement that it be installed in a location where direct sunlight does not enter the receiver throughout the year, which poses the problem of requiring extra labor and equipment to take measures for this purpose.

This invention was made in view of the problems of the conventional optical level crossing obstacle detection device, and by mounting the camera diagonally downward from a high place, it is possible to reliably detect all types of automatic crossing obstacles. To provide a level crossing obstacle detection device capable of reducing the adhesion of brake powder, dust, etc. that occurs during train running, and reducing false detections.

[Structure of the Invention] (Means for Solving One Problem) The level crossing obstacle detection device of the present invention is arranged at a plurality of locations along the track, and is configured to take overlapping images of the same location on the road surface inside the level crossing. Multiple cameras are used to capture the images taken by each camera, and the two sets of images dancing in the same place are superimposed on each other, and the difference in shading of the same pixel is compared with a preset tolerance value. An image processing device that counts the number of pixels that exceed a permissible value across all pixels, compares this count value with a preset standard value, and if it exceeds a preset value, assumes that there is a level crossing obstacle and outputs a contact point m. and a light-emitting signal installed along the railway line that lights up to notify the driver of a train entering the level crossing of a obstruction to the level crossing when this contact close output is received.

(Function) In the level crossing obstacle detection device of the present invention, a plurality of cameras are arranged in the direction of the level crossing breaker and each camera is placed at a high point midway between the double-track tracks in the level crossing to take images of the inside of the level crossing including the tracks. A camera captures an image of the same location, the video signal is input to an image processing device, and the program processing within the image processing device combines the two images taken of the same location to calculate the difference in shading of the same pixel to all pixels. Ask for the mouth. When the number of pixels with a gray level difference that exceeds the allowable range exceeds a specified value, it is assumed that there is an obstacle, and the lamps of the light-emitting signals installed along the railway are turned on, and the train entering the level crossing is operated f. Make them aware of danger.

(Example) Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is an overall configuration diagram of an embodiment of a level crossing obstacle detection device according to the present invention.

Tracks 21 and 22 on which trains run are level crossing breaker 1
1 to 14 are blocked from general roads. In this embodiment, the general road is composed of two lanes with double up and down tracks on tracks 21 and 22. Additionally, fences 41 are installed at areas other than railroad crossings to prevent people from entering.

A strut 31.32 is installed along the track between the tracks 21.22 and is tied through a support 33.

Using this support member 33, half of each of the cameras 51 to 60, the lights 61 to 64, and the illuminance detectors 65 and 66 are mounted in the direction of the track 121 and 22.

There are W control lights 71 and 72 and light-emitting signals 73 installed along the line.

In addition, in order to limit the lighting of the lights 61 to 64, in this embodiment, the existing track circuit conditions are
Train presence signals for a total of two sections, the '1M section and one section ahead in the direction of travel, are input via track relays 75 and 76.

White lines 42 indicate lane boundaries.

Figure 2(a) targets the monitoring area 40d.
It shows the range of images taken by the camera 52 at the center and the cameras 51 and 53 on its left and right sides. 101 to 103 Shows the image carrying areas of cameras 51 to 53, and the imaging area 10
2 is further divided into left and right common areas 104 and 105. One common area 104 is a right half image taken by the left camera 51 and a 1llII image taken by the center camera 52.
The images of the left half of the left half overlap, and the common area 105 is the right half imaged by the center camera 52 and the right half of the image taken by the right camera 53.
The left half image taken with is overlapping.

Figure 2(b) shows a place where the buildings and trees 1000 are reflected on the road surface in the common areas 104 and 105, and the central camera 52
An image 104=105- taken at is shown. This image is not much different from the right half image taken by the left camera 51 and the left half image taken by the right camera 53. The only slight difference is the size of the image, which is inversely proportional to the distance between the camera and the road surface, due to the difference in the mounting position a of the camera.

Figure 2 (C) shows that there is an obstacle 1 inside the railroad crossing as shown in Figure 3.
Image 10 captured by the left camera 51 when 06 was present
The right camera 53 is mounted so that the left side surface 1002 of the obstacle 106 is captured, and the right side camera 53 obtains an image 105'' shown in FIG. 2(e), showing the left side surface 1003 of the obstacle 106. FIG. 2 (■) is an image 104, 105 captured by the central camera 52, showing the roof 1001 of the automobile 106. In this way, the left camera 51 images the left side of the vehicle, the right camera 53 images the right side of the vehicle, and the center camera 52 images the front of the vehicle. Note that the same applies to the other monitoring areas 40b to 40d.

The fourth image) is a hardware configuration diagram for monitoring monitoring areas 4Qa and 4Qb. The camera 51 has an infrared filter 151 and a lens 25 with an AE (automatic aperture mechanism).
1. It consists of a CCD black and white camera body 351.

The infrared filter 151 is a filter that cuts wavelengths of 690 to 720111 or less and does not allow them to pass through, and cuts the visible light region of sunlight so that it is not affected. At the same time, the night lighting also cuts out visible light and emits longer wavelengths than infrared light so that it does not interfere with road traffic at railroad crossings or trains running, so that it is barely noticeable to the eyes. The AE lens 251 has a function of automatically adjusting the aperture of the amount of light entering the camera depending on the brightness of external light. The camera body 351 is a general-purpose CG[) camera.

Cameras 51-5 made with this configuration
5 are housed in camera cases 511 to 515, respectively.
Eaves 45 to prevent direct reception of external light in directions other than the imaging direction.
1 to 455 are provided.

The illuminance detector 65 is attached to the center of the support member 33, and the condenser lens 3 is placed inside the milky white semicircular globe 265.
65 collects external light and improves the light receiving sensitivity of Cd5165. The amplifier 465 contains circuit components having the configuration shown in FIG. Outputs (analog value).

The lights 61 and 62 have built-in halogen lamps 161 and 162, and light emitting lenses 261 and 262 are placed to set the maximum irradiation angle θ.
The light is generated only in the monitoring areas 4Qa and 40b, and the amount of light is increased. An infrared filter 361 is installed in front of the light emitting lens 261°262.
．． 362 is provided to allow only light with a long wavelength of 690 to 720 n- or more to pass through so that it does not come into direct contact with human eyes, in consideration of driving safety. Note that the other monitoring areas 40c and 40d have exactly the same configuration.

FIG. 4(C) is a hardware configuration diagram of the cameras 51 to 55 control unit 721, in which the calculation unit 1111 includes 1. 2 for image processing processor, 3 for video input module, 4 for image memory, 5 for image program memory, 6 for digital output module, 7 for program memory, 8 for data memory
, and a power-on module 9.

1 in the tlD section is 1 in the program command.
Extract and decode each step. The image processing processor 2 performs image processing calculations by fetching the image processing subroutine stored in the image processing program memory 5 and setting necessary parameters (numeric values). Here, the pixels of one screen are composed of U256x256 points horizontally, and one pixel is 6 to 8 bit data. Therefore, if one pixel is 6 bits, white is 28 bits.
-1-63, black is converted to a numerical value of 20-1-0 to represent 11 degrees of the pixel at each point.

Image memory 4 serves as a memory area for temporarily storing image information captured from the camera, and as a working area for temporarily storing mask patterns for calculation and intermediate results of calculation. . 5 cameras 51 connected to the video input module 3
~ When one of 55 is selected, connect the image signal line, detect the vertical synchronization signal on the camera side, and capture the image signal sequentially from the beginning of the screen until the white signal 1■ and the black signal 0.3V. Set the level for each pixel in 6 to 8 bits (64 to 25
(divided into 6 ways) into a digital signal (A→D conversion)
Then, by storing one pixel every four pixels in the image memory, all pixels, that is, for one screen, are sequentially stored.

The program memory 7 is a memory for storing programs other than image processing, and is a memory for storing programs other than image processing.
In order to sequentially write 11 images into the image memory 4, the cameras are sequentially selected, the video input module is given a capture instruction to 3, the image processing processor is given a calculation instruction to 2, and the image processing processor A series of processes are executed to take in the result of calculation by 2 and operate 6 in the digital output module.

The data memory 8 serves as a working area for temporarily storing parameters necessary for various calculations and intermediate results of calculations.

The digital output module 6 is an interface for operating external equipment by modifying the results processed by the arithmetic and control unit 1, and has a non-voltage contact output.

The power-on module 9 is a module that generates an interrupt when the power is turned on and executes a program starting from the address written in the interrupt destination address.

FIG. 4(b) shows the power supply for driving the AE of the AE-equipped lens (151 in the case of the camera 51) in the cameras 51 to 55, and the lamp 161 of the lighting 61, 62 for illuminating the night light etc. External terminals c, c: c, c are circuits for supplying the excitation power of the relays that flow the current to the relay fill 723a necessary to light up the .162.
has been established. This power supply circuit includes a transistor Tr+
~Tr3 and resistors R1 to R3.

The operation of the eight level crossing obstacle detection units configured as described above will be described next.

In FIG. 1, first, the track 'a' connected to the track relay 76 is
When a train approaches 22, the relay contacts in the track relay 76 shown in FIG.
4d is closed, and as a result, the alternating current voltage ACV is applied to the second relay plate 7.
It is supplied to one of the contacts 723b and 723c in 23.

On the other hand, in the illuminance detector 65, external light passes through the globe 265, is collected by the condensing lens 365, and is supplied to the Cd5165, and in the circuit shown in FIG. The power is amplified as described in , and the AE of cameras 51 to 55 and the lights 65 and 66 are turned on and off.
Relay coil 723ak for FF: supplies If source.

In Figure 4(b), when the surrounding brightness is dark, Cd816
As a result, the potential at point a1 rises and the base current of transistor rr+ flows in the direction of arrow I, causing transistors Tr2. The base current of Tr3 flows and a
3. The potential at point a4 rises, and the voltages at C and C rise.

Resistors R4 and R6 are variable resistors, one resistor R4 is provided to adjust the voltage level of AE, and the other resistor R6 is used for adjustment to detect the brightness of the boundary where the illumination is turned on. It is provided as a.

C#i! The child output is supplied to each AE of the cameras 51 to 55 in the camera cases 511 to 515, while the C terminal output is supplied to the second AE.
The relay coil 723a of the relay plate 723 is energized or demagnetized.

The relay coil 723 of the second relay plate 723 can
a disappears, contacts 723a and 723C remain open, and lamps 161 and 162 of illumination 61 and 62 do not light up. On the other hand, if the outside light is not dark enough to turn on the lights,
The relay coil 723a of the second relay plate 723 is excited, the contacts 723b and 723c close, and the lamps 161.162 of the lighting 61.62 are turned on.

The conditions for the lamps 161 and 162 to turn white are the orbital relay 7
6 and the track relay 7 of the track 22 where the light emitting signal 73 is present.
5 relay contact falls (This track relay 75 is always energized and its contact is hot, but when the two rails are electrically short-circuited between the train's wheels and axles, the energization is released and the contact is energized by gravity. When the contact falls and opens), the back contact (not shown) is used to excite the relay coils 724b and 724a of the third relay plate 724, and the contact 724d,
724C is closed and AC voltage ACV is supplied. That is, the lights 61, 62 indicate that the train is connected to the track relay 76.
Power is supplied only when the lamp is in either or both orbits of the lamp 75, thereby preventing the lamp life from being shorted due to continuous illumination.

Next, software processing will be explained based on FIG. 5.

In FIG. 4(C), when the power is turned on, the control unit 721 first sets an interrupt bit 9 in the power-on module in a hardware manner to generate an interrupt. When this interrupt occurs, the controller 1 detects the interrupt, decodes the program start address at the address where the υ1 include bit is set, and programs the program starting from this address. 7 into the memory and execute it. Between boxes, the program start address corresponds to the entrance, and the program is first executed from the start address of block B11. The program of this block Bll connects only the video signal line of the camera 51 to the video signal module 3, and sends the image signal of the camera 51 for one screen from the beginning of the screen in accordance with the synchronization signal of the camera 51 for one pixel. The images are captured one by one, converted into digital signals, and stored in image memory in four image areas P1 (referring to the memory area for one screen starting from address P1). This is an image of the imaging area 101 in FIG. 2). Then, the next block F312 is executed.

The program for push button 12 sets the left half to black and the right half to white, and presses the first button 811 to display the contents of image area P1 (
Pl), leaving only the common area 104, filling the left half with black (data O), temporarily storing it in the image area PM1, and executing the next block 813.

The program in block B13 connects only the video signal line of camera 52 to the video input module.
Similarly, the imaging signal of No. 2 is captured pixel by pixel and written into the image area P2 of No. 4 in the image memory. This is the second icon) II
This is an image of @ area 102. And the next block 81
Execute 4.

The program in block B14 masks the image area (R2) with a second pattern 814 that is black and white inverted from the first pattern 811, with the left half being white and the right half black, leaving only the common area 105, and masking the right half. It is filled in with black and temporarily stored in the image area PM21, and the next block 815 is executed.

The program in block 815 calculates the difference in shading between the content [PM1] of the image area PM1 and the content [PM21] of the image area PM2 (difference in numerical representation where white is 63 and black is O at 6 bits). ), take out the subroutine for calculating the difference from 5 into the image processing program memory, have the image processing processor execute it in 2, and
The difference is calculated for each pixel, the result is taken into the arithmetic control unit as 1, and the tolerance value [α
] (the content of address α is an allowable value) is counted, and the pixels are added to the image area PM1. Contents of PM2 [:P
Compare the accumulated results accumulated over the entire area of PM2 with the numerical value [β stored in advance in the data memory in 8. If it is within the number of β, it is considered the same and the block is The execution moves to block B21, and if the obstacle has been crossed, it is assumed that there is an obstacle, and the execution moves to block B21.

That is, as shown in Figure 2 (2), Akira 1000, which is a building and a grove of trees that appear on the road surface inside the railroad crossing, is not three-dimensional but flat, so it does not have any thickness. Therefore, the common image taken with camera 51, L
In the image of the rear 104 and the image of the common area 104 captured by the camera 52, the objects become smaller in inverse proportion to the distance between the camera and the road surface, but when superimposed, they match at the center line of the common area 104. If you add up the deviations that appear as you move away from the left and right, you will get a ginsel,
As a result, the error is small and they almost match.

On the other hand, when an obstacle (car) is present as shown in FIGS. 2(C) to (0), the height of the heavy object will affect the difference between the image 104n captured by the camera 51 and the image 104 conveyed by the camera 52. , the camera 51 obtains an image taken from the left side, and the camera 52 obtains an image taken from the front, so even if the two images are superimposed, they do not match.Therefore, the higher the height, the greater the discrepancy becomes. This allows for clear identification.

Note that, regarding the common area 105 as well, the images taken by the camera 52 and the images taken by the camera 53 produce exactly the same result in principle.

Now, the program in block 816 is to send the image processing processor to the image area P which is stored in the image memory at 4.
Contents of 2 [Take out P2 and perform a logical product (AND) with the first pattern 811, leave only the right half common area 105, fill in the image area PM22, block B
Execute step 17.

The program in block 817 connects only the video signal line of camera 53 to the video input module.
The camera 53 receives the imaging signal of No. 3 for one screen from the beginning of the screen.
One pixel is taken in in accordance with the synchronization signal, converted to a digital signal, and written to image area P3 of image memory 4.

This is an image of the imaging area 103 in FIG. 2(2).

Then, the next block 818 is executed.

The program of step 1 818 is the second pattern S1.
4 into the image processing processor, 2 into the image memory, the contents of the image area P3 which were taken out by 4 and written in the processing of block B17 [Mask P3 and leave only the common area 105, and fill the rest with black, and then write the image. The image area PM3 in 4 is stored in the memory, and the next block B19 is executed.

In the program of block B19, similarly to block B15, the image processor 2 retrieves the contents [PM3] of the image area PM3 in 4 from the image memory, and stores this [PM3] and the contents of the image area PM3 in the image memory 4 obtained in block 816. A subroutine for determining the difference from the content [8M22] of the image area PM22 is extracted and executed, the density difference is determined for each pixel, and the result is input to the arithmetic control section. Then, the pixels exceeding the tolerance value [α] stored in advance in the data memory in pixel area PM3 are counted. Contents of PM22 [P
M3], [8M22] are accumulated over the entire area, and the cumulative results are stored in advance in the data memory in 8 [β]
If it is within the number [β], it is assumed that they are the same, and the process moves to block 820, and if it exceeds it, it is assumed that there is an obstacle, and the process branches to block B21.

The program in block 820 includes digital outputs 7 and 1.
- 1 decodes the command word that instructs relay contact between 6 and 6 to the arithmetic control unit and outputs "0" to 6 to the digital output module.
Output a signal. Then, the first relay plate 72 of the fourth icon)
No. 2 relay coil 722a is demagnetized with no voltage,
Contact 722b is opened to turn off light emitting signal 73 to notify the train driver that it is safe, and block 822 is executed.

In the program of block 821, 1 decodes the instruction word that instructs the digital output module to close the relay contact 6, and outputs "1" to the digital output module 6.
Outputs the signal. Then, the relay coil 722a of the first relay plate 722 in the fourth image) is excited by the application of voltage, the contact 722b is attracted and closed, and the light emitting signal 7
3 is turned on to notify the train driver of the danger, and the block signal B22 is executed.

Block B22 is a dummy program, for example, D
o Play LOOP about 5000 times and put 1 file in the control section.
Block B11 is temporarily held until the digital output module 6 is activated, the contact 722b of the first relay plate 722 is activated, and the light emitting signal 73 is turned on or off.

Although the above operation has been explained only for the monitoring area 40a in the railroad crossing, the monitoring areas 40b to 40d also operate in exactly the same way.

In this way, in this embodiment, a plurality of cameras 51 to 60 are installed at the level crossing breaker 11 to 60 to image the inside of the level crossing, including the track 3121°22 at an intermediate height between the double track track 21 and the track 22 in the level crossing. Each camera is arranged in the direction of 14, and the same place is imaged, the video signal is taken into the image processing device 71.72, and the two images taken of the same place are captured by the program processing in the image processing device 71.72. Differences in 11 shades of the same pixel are obtained over all pixels by overlapping them. When the number of pixels with an I/D difference exceeding the allowable range exceeds the specified value, it is assumed that there is an obstacle and the lamps of the light-emitting signals @ machines 73 and 74 installed along the railroad are turned on and the train enters the railroad crossing. This will alert the train driver of the danger.

In addition, in this invention, the number of cameras to be installed can be increased or decreased depending on the width of one lane, and the cameras may be installed at positions that handle both lanes at once, and the number of cameras installed can be increased or decreased depending on the width of one lane. Nor is it limited to. Furthermore, in the above embodiment, the areas where two adjacent cameras capture images of the same place are divided into halves, but this is also a matter of determining the overlapping areas based on the distance between the cameras and the road surface. There is no limit to the size of the area.

[Effects of the Invention] As described above, according to the present invention, since the camera is mounted diagonally downward from a position higher than the height of the train as an input for detecting obstacles in the level crossing, imaging due to lumps and brake powder is prevented. Adhesion to surfaces can be significantly reduced, and obstacle detection can be performed stably without false detection, contributing to reduced maintenance.

In addition, images of the same location taken by multiple cameras are compared, and the number of pixels with a gray level difference that exceeds an allowable value is cut, and if the count value is greater than a specified value, it is determined that there is an obstacle. regardless of the size or type of vehicle being
Obstacles can be detected stably and reliability can be increased.

[Brief explanation of drawings]

FIG. 1 is an overall layout diagram of an embodiment of the present invention, and FIG. 2 (a
) to (e) are explanatory diagrams showing an example of imaging inside a railroad crossing by the camera used in the above embodiment, Figure 3 is a sketch showing the installation of the camera, lighting, and illuminance detector equipment in the above embodiment, and Figure 4 is a diagram) ~(C) is a hardware configuration diagram of the control unit of the above embodiment, FIG. 5 is a software configuration diagram of the control unit of the above embodiment, and 6th image), (υ is the configuration of the conventional obstacle detection device. It is a diagram. 11-14... Level crossing breakers 21-22... Track 42... White lines 51-60.
...Camera 61-64...Lighting 65.66...Illuminance detector 73.74...Control box 74.73...Light-emitting signal 75.76...Track relay 101-103...l1ii Statue area 104.105
...Common area 106...Obstacles

Claims (1)

[Claims] A plurality of cameras are arranged at multiple locations along the track to take overlapping images of the same location on the road surface inside the railroad crossing, and the above-mentioned 2. Overlay each set of images on top of each other,
The difference in density of the same pixel is compared with a preset tolerance value, and the number of pixels that exceed the tolerance value is counted across all pixels, and if this count value exceeds the preset standard value, is an image processing device that assumes there is a level crossing obstruction and outputs a closed contact signal, and a light emitting device installed along the railway line that lights up when this contact output is received to notify the driver of a train entering the level crossing that there is a level crossing obstruction. A railroad crossing obstacle detection device comprising: a traffic light;