JP3085755B2 - Obstacle detection device for railway vehicles - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an obstacle detecting device for a railway vehicle for detecting an obstacle present on the course of a running railway vehicle.

[0002]

2. Description of the Related Art In the operation of railway vehicles such as trains and trains,
In order to ensure the safety of the vehicle, it is conceivable to take an image of the front of the traveling vehicle and determine the presence or absence of an obstacle on the course from the captured image. In this case, the problem is the distance La to the obstacle. When the braking distance of the vehicle is L,
If L <La, collision can be avoided. However, if L> La, a collision will occur even if an obstruction can be detected. Therefore, it is meaningless if the obstacle is not detected beyond L (braking distance) ahead of the vehicle. Furthermore, considering a margin ΔL for the driver to make a decision, (L + Δ
L) It is necessary to detect obstacles beyond the distance. The following two methods can be considered as a method of determining the position of the rail in front of (L + ΔL) in the screen of the captured image.

(1) The distance between two rails is constant.
By utilizing this feature, the number of pixels between the rails in front of (L + ΔL) is calculated in advance from the angle of view of the camera.
The position of the (L + ΔL) forward rail in the screen is obtained.
However, this method is effective only when the track is traveling straight, and cannot calculate an accurate position when the track is curved. Then, the following method can be considered. (2) The distance is calculated by triangulation using two cameras, and the position of the rail in front of (L + ΔL) is determined.

[0004] If the position of the rail in front of (L + ΔL) is known on the screen, an area (obstruction search area) for inspecting the presence / absence of an obstacle is set centering on the position, and image processing is performed in the area. Thus, the presence or absence of an obstacle is determined.

[0005]

In the methods (1) and (2), it is necessary to identify and recognize the position of the rail on the screen by performing image processing and the like. Moreover,
The position in front of (L + ΔL) is obtained by the methods (1) and (2). However, in (1), it is impossible to determine the position when the track is curved, and in (2), the distance between the cameras cannot be greater than the lateral width of the vehicle body. There is a problem in accuracy because the distance between cameras cannot be sufficiently obtained.
Furthermore, since two cameras are used, there is a problem that the entire system is complicated, the time required for calculation is long, and the cost is high.

The present invention has been made in view of the above points, and it is possible to accurately determine the presence or absence of an obstacle at a distance (L + ΔL) beyond the braking distance and notify the driver of the obstacle. It is an object of the present invention to provide a railway vehicle obstacle detection device that can be configured at low cost without complicating the structure.

[0007]

According to a first aspect of the present invention, there is provided an obstacle detecting device for a railway vehicle, which is provided with an image input device installed so as to capture an image in front of a running railway vehicle, and an image captured by the image input device. An image memory for storing the current image, position detecting means for detecting the current position of the traveling vehicle, and how the position detecting means detects how a rail looks in an image taken by the image input device. Calculated in advance according to track conditions as route data corresponding to the vehicle position
Determination and route data memory for way to collect and store, whether on the basis of the position detection signal from the line data and the position detecting means stored in the memory is obstacle in the images stored in the image memory And a notifying means for notifying the driver of the presence or absence of an obstacle in accordance with the result of the determination by the determining means. The obstacle detection device for a railway vehicle according to the second invention is an image input device installed to capture the front of a running railway vehicle, an image memory that stores an image captured by the image input device, Position detecting means for detecting the current position of the traveling vehicle, and how the rail looks in an image taken by the image input device as route data corresponding to the vehicle position detected by the position detecting means Pre- gauge
And route data memory for storing are acquired by calculating values by way conditions, route de stored in the line data memory
Data based on the position signal detected by the position detecting means.
Data reading means for reading the route data
A predetermined distance before based on the route data read from the memory
Calculate the center point of the left and right rails and
Calculate the visible position and set the obstacle search area at this position.
Search area setting means for determining
Obstacle search area set by search area setting means
And a notifying means for notifying the driver of the presence or absence of an obstacle according to the result of the determination by the determining means. I do. The obstacle detecting device for a railway vehicle according to the third invention is an image input device installed to photograph the front of a traveling railway vehicle, an image memory storing an image photographed by the image input device, The image stored in this image memory is
Extraction to extract the rail portion as a binary image by binarization
Means, position detecting means for detecting the current position of the traveling vehicle, and how the rail looks in an image taken by the image input device, corresponding to the vehicle position detected by the position detecting means. As route data, it depends on track conditions in advance .
And route data memory for storing are acquired by Ri calculated value, this
The route data stored in the route data memory of
Reading based on the position signal detected by the position detecting means
From the data readout means and the route data memory
Based on the read route data, the left and right
Calculates the center point of the rail and the position visible on the screen
Calculating means, and reading from the route data memory
Left within a predetermined distance range ahead based on the route data
A rail image generating means for generating a right rail image;
Image generated by rail image generating means and binarization described above
The amount of misalignment is calculated by taking the correlation with the
The rail ahead of the specified distance that can be seen in the above screen depending on the amount of deviation
Correction means for correcting the center point, and correction by the correction means.
Set the obstacle search area at the center of the corrected rail
Search area setting means, and the search from the image memory.
The image of the obstacle search area set by the area setting means
The image data is read, and a determination unit for determining whether there is an obstacle in the image data , and a notification unit for notifying a driver of the presence or absence of an obstacle in accordance with the determination result of the determination unit are provided. .

[0008]

[Route] Route data obtained by converting the laying state of a railway into data is stored in a route data memory in advance, and the current position (for example, distance from a reference point) of a running vehicle is detected by a position detecting means.
Is detected and the storage content of the route data memory is read, whereby the laying state of the track in front of the vehicle can be confirmed. That is, when three-dimensional coordinates with the current position as the origin are set, the coordinates of the position p of the rail ahead of an arbitrary distance R can be uniquely determined. Where this point p is photographed by an image input device installed to photograph the front of the traveling vehicle, for example, an infrared camera, it can be found by a mathematical coordinate transformation calculation. Therefore, it is possible to know where the rail ahead of an arbitrary distance R can be seen in the screen. Therefore, an obstacle search area of a finite size is set centering on the position of the rail in the screen when R = (L + ΔL), and image processing is performed in the area to determine the presence or absence of an obstacle. Judgment is made, and this judgment result is notified to the driver.

[0009] By using the route data as described above, even if the front line is straight or curved, (L
+ ΔL) The position of the front rail in the screen can be accurately calculated. The distance (L + ΔL) is arbitrary, and can be set to a braking distance according to the speed of the vehicle. Further, by performing image processing or the like, there is no need to identify and recognize the position of the rail in the screen, so that the calculation time can be greatly reduced.

[0010]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows a configuration of an obstacle detecting device for a railway vehicle according to one embodiment of the present invention, which is provided with an infrared camera 11 as an imaging means. This infrared camera 11
Is mounted at the head position of the vehicle 10 as shown in FIG. 2 and photographs the front of the vehicle. The video signal captured by the camera 11 is converted into digital data by an A / D (analog / digital) converter 12. And this A
An output signal from the / D converter 12 is written to the image memory 13 based on a command from the arithmetic unit 14.

The arithmetic unit 14 is supplied with a kilometer (distance from a reference point) signal from a kilometer sensor 15 for detecting the current position of the vehicle, and route data stored in a route data memory 16 in advance. You. Above kilometer sensor 15
Is attached to the wheel of the vehicle as shown in FIG. 2 and outputs distance data corresponding to the number of revolutions, for example. An example of the route data memory 16 is shown in FIG.
As shown in (a) and (b), route data indicating a track shape corresponding to about a kilometer is stored. In the example shown in FIG. 3, the radius from 0 to 100 is about 50 m in radius and 90 ° in angle.
Is a right (R) curve, followed by a straight line of 400 m, and from 500 km to 700 km is a left (L) curve with a radius of 100 m and an intersection angle of 60 °, followed by a straight line of 100 m. In FIG. 3, BTC (Biginning
of Transition Curve) is the starting point of the transition curve, BCC (Bigi
nning of Circular Curve) is the starting point of the circular curve, ECC (EN
D of Circular Curve) is the end point of the circular curve, ETC (END of
Transition Curve) indicates the end point of the transition curve. The route data shown in FIG. 3 is for the horizontal direction,
Similarly, in the vertical route data, a track gradient corresponding to, for example, about km is recorded in the route data memory 16.

The arithmetic unit 14 shown in FIG. 1 is roughly divided into a binarizing means 14a, an obstacle search area setting means 1
4b and an obstacle presence / absence determining means 14c. After obtaining the image data from the image memory 13, the binarizing means 14 a performs binarization using various image processing algorithms and records the binary image in the image memory 13. The obstacle search area setting means 14b is a region (an obstacle search area) for inspecting the presence / absence of obstacles using the kilometer signal from the kilometer sensor 15, the route data from the memory 16, and the binary image from the image memory 13. ) Is set in the screen. The obstacle presence / absence determination unit 14c determines the presence / absence of an obstacle by performing various image processing in the above-described area (obstacle search area). This determination result is output as a notification from the output device 17 to the driver.

FIG. 4 shows a flow of the operation in the arithmetic unit 14. In step 101, a kilometer corresponding to the current position of the vehicle is set. Next step 102
Then, the output of the infrared camera 11 is digitized by the A / D converter 12 and stored in the image memory 13. Step 10
In step 3, the image data stored in the image memory 13 is binarized by various known image processing algorithms, the rail portion is extracted as a binary image, and the binary image is recorded in the image memory 13.

In step 104, a kilometer corresponding to the current position of the vehicle is read from the kilometer sensor 15. In step 105, the route data corresponding to this kilometer is read from the route data memory 16. In step 106, an obstacle search area is set in the screen using the binary image and the route data according to a method described later. Step 10
At 7, the presence or absence of an obstacle is determined by performing various image processing described later in the obstacle search area. Then, in the next step 108, it is determined whether or not an obstacle is detected in the above step 107. If an obstacle is detected, the process proceeds to step 109, where an output display for notifying the driver of the presence of the obstacle is made. Is executed. At step 110, it is determined whether or not the operation has been completed. If the operation has been completed, the present process ends.

Next, a method of setting the obstacle search area in step 106 will be described. FIG. 5 shows an example of the processing flow of the above method. Step 201 corresponds to steps 104 and 105 described above. By this operation, the laying state of the track ahead in the traveling direction can be known. Therefore, in step 202, 3 as shown in FIG.
The dimensional coordinate system 20 is set, and the coordinates (x, y, z) of the center point p21 of the left and right rails, for example, 1500 m ahead are calculated. Here, 1500 m is an example of the sum (L + ΔL) of the braking distance L and the margin ΔL for the driver to make a decision, and can be set to a value corresponding to the speed of the vehicle. In step 203, when the point p21 is photographed by the infrared camera 11, a position (m, n) 24 at which the point p can be seen in the image 23 is calculated as shown in FIG.

FIG. 7 shows a digital image stored in the image memory 13 after photographing by the infrared camera 11.
The coordinates (m, n) 24 are given by mathematical coordinate conversion calculation using the camera mounting position, angle, and camera angle of view, which can be uniquely determined in the three-dimensional coordinate system 20, as parameters. In step 204, an obstacle search area 25 is set around the position (m, n) 24 where the point p can be seen in the screen. The size of the obstacle search area is arbitrary, but if it is too large, it is meaningless, and if it is too small, the area to be inspected is not included, which is a problem.

An example of a method for setting the size of the obstacle search area will be described below. Considering the construction limit of the vehicle, it is considered that there is no problem if there are no obstacles in the area of 3.8 m x 3.8 m on the track. So 3.8m ahead of 1500m
The size of the area of mx 3.8 m in the screen is calculated. This is the number of pixels in the horizontal and vertical directions,
Determined by the horizontal and vertical angles of view of the camera. Assuming that the number of pixels is 640 × 486 and the angle of view is 3.5 ° × 2.7 °, the area of 3.8 m × 3.8 m ahead of 1500 m has a size of 27 × 27 (pixels) in the screen. . Therefore, as shown in FIG. 7, 27 × 2 around the coordinates (m, n) 24
An obstacle search area 25 having a size of 7 (pixel) is set. In the present embodiment, the coordinates (m,
Although n) 24 is calculated every time, it is also conceivable that the coordinates (m, n) 24 obtained in the above step 203 are stored in association with about kilometers and used as route data. That is, the route data may be stored in any form as long as the coordinates (m, n) 24 corresponding to kilometers are known.

[0018] The method shown in FIG. 5, it is possible in the case, such as negligible influence of the vibration of the vehicle. Actually, the camera mounting position and the angle with respect to the vehicle change due to the vibration, and the image is displaced. That is, an image different from the image when the vehicle is stationary is captured. Step 2 above
Position (m, n) at which point p found in 03 can be seen in image 23
Reference numeral 24 denotes a position on the screen when the vehicle is stationary. When the camera mounting position and the angle are greatly changed due to vibration, the position (m, n) 24 shifts in the screen, which causes a problem. FIG. 8 shows a processing flow of a method for eliminating the influence of vibration.

Step 201 in FIG. 8 has the same contents as step 201 in FIG. In step 301, a rail image 30 shown in FIG. 9 in which left and right rails are drawn using the route data stored in the route data memory 16 is generated in the image memory 13. The rail image 30 is, for example, a line 2 in which a rail is drawn with a bit “1” on the background of a bit “0”.
It is a value image.

An example of a method for generating the rail image 30 will be described below. In step 202 in FIG. 5, not only the point ahead of 1500 m but also the point
Calculation is performed every m, and the coordinates in the screen are obtained in step 203. Since these are the coordinates of the center of the rail, the coordinates of the left and right rails are obtained by adding the number of pixels between the rails according to the distance. Connect the coordinates with a straight line 200-1
A left and right rail image 500 m ahead is generated. It is not absolutely necessary to calculate 200 to 1500 m every 100 m, and it is needless to say that the calculation may be changed as needed. Also, as described above, the coordinates in the screen of the left and right rails for each 100 m ahead of 200 to 1500 m calculated in step 301 are stored in association with the kilometers,
Use as route data is also conceivable. In addition, any method can be used for storing the route data as long as a rail image 30 corresponding to about a kilometer can be generated.

In step 302, the shift amounts ρm and ρn are obtained by correlating the rail image 30 with the binary image 31 obtained by extracting the rail portion shown in FIG. The binary image 31 is shown in FIG.
As shown in the figure, for example, the rail 34a has a bit "1".
This is a binary image as represented by. Above displacement ρm3
As shown in FIG. 11, 5m is the number of pixels that are shifted in the horizontal direction (m-axis direction) between the actual rail 34 and the drawn rail 33, and the shift amount ρn35n is shifted in the vertical direction (n-axis direction). The number of pixels.

As shown in FIG. 9, a window 32 is set in the rail image 30 so as to surround the drawn rail 33. A correlation value in the window 32 is obtained by shifting one image up, down, left and right from a state in which the two images are completely overlapped. Specifically, a logical product image of the rail image 30 and the binary image 31 is obtained, and the sum of the bits “1” in the window 32 is set as the correlation value. The shift amount in the horizontal direction when the correlation value is maximum is regarded as the shift amount ρm, and the shift amount in the vertical direction is regarded as the shift amount pn. The method of setting the window for taking the correlation is not necessarily required to be the above-described method, and the shift amounts ρm, ρn
Any window can be used as long as it can be obtained accurately.

Next, at step 303, the deviation ρm35
The obstacle search area 25 is set by correcting m and pn35n. As shown in FIG. 12, the position (m, n) 2 where the center point p21 of the rail 1500 m ahead can be seen in the image 23.
4 is corrected by the shift amounts ρm35m and ρn35n (m
An obstacle search area 25 having a size of 27 × 27 pixels is set around (+ ρm, n + ρn) 36. That is,
An accurate obstacle search area 25 can be set by obtaining a shift amount in the screen and correcting the shift amount.

FIG. 13 shows an example of the obstacle presence / absence determination processing in step 107 in FIG. This apparatus uses an infrared camera 11 as an imaging means. When an obstacle such as a car is photographed by the infrared camera 11, a high-temperature portion such as an engine is reflected brightly. Utilizing this feature, if there is a high-luminance portion in the obstacle search area 25, it is determined that there is an obstacle. In step 401, image data in the obstacle search area 25 is extracted from the digital image 23 stored in the image memory 13. In step 402, for example, the image data is binarized with an appropriate threshold,
A high-luminance part (a part whose luminance is higher than the background) is extracted. In this case, for example, the high-luminance portion is bit “1”, and the other portions are “0”.

In step 403, step 402
It is determined whether or not a high-luminance portion has been extracted from the binary image generated by. As an example of the determination method, if a bit “1” exists, it is determined that there is an obstacle, and the process proceeds to step 404Y. Step 404 if there is no obstacle
Go to N. In this method, the presence or absence of an obstacle is determined based on the presence or absence of a high-luminance portion in the obstacle search area 25. The binarization method in step 402 uses any method that can extract a high-luminance portion. May be a method,
In addition, there is no reason that the method of determining the presence or absence of a high-luminance portion in step 403 must necessarily be the above method.

The obstacle presence / absence determination result 404Y or 404N obtained by the above processing is output to step 108 in FIG. 4, and if an obstacle exists, the driver is notified of the existence of the obstacle in step 109.

[0027]

As described above in detail, according to the obstacle detecting device for a railway vehicle according to the present invention, the use of the route data makes it possible to display the rail ahead of an arbitrary distance regardless of the state of the line laying in front of the vehicle. You can calculate where you can see it within. Therefore, the area to be inspected for obstacles can be determined accurately and in a short time, so that the presence or absence of obstacles in the area beyond the braking distance can be accurately determined and reported to the driver, and high safety can be achieved. Is guaranteed. In addition, there is no need to install two cameras, and there are no problems such as complication of the system, extension of required calculation time, and high cost.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an obstacle detection device for a railway vehicle according to an embodiment of the present invention.

FIG. 2 is an explanatory view of a vehicle on which the obstacle detection device is mounted.

FIG. 3 is a diagram showing an example of route data.

FIG. 4 is a flowchart illustrating a flow of processing of the obstacle detection device.

FIG. 5 is a flowchart for explaining the operation of an obstacle search area setting process in step 106 in FIG. 4;

FIG. 6 is an operation explanatory view in FIG. 5;

FIG. 7 is a view for explaining a method of setting an obstacle search area in FIG. 5;

FIG. 8 is a flowchart illustrating an example of processing for eliminating the influence of vibration in setting an obstacle search area.

FIG. 9 is a view for explaining a method of setting a window on a rail image when obtaining a correlation value.

FIG. 10 is a diagram showing a binary image in which a rail portion is extracted.

FIG. 11 is a diagram illustrating shift amounts ρm and ρn.

FIG. 12 is a view for explaining a method of setting an obstacle search area in FIG. 8;

FIG. 13 is a flowchart illustrating an obstacle presence / absence determination process in step 107 in FIG. 4;

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 10 ... Railway vehicle, 11 ... Infrared camera, 12 ... A / D converter, 13 ... Image memory, 14 ... Computing device, 14a ... 2
Value conversion means, 14b ... obstacle search area setting means, 14c
... Obstacle presence / absence determination means, 15 .about.km sensor, 16... Route data memory, 17... Output device, 20.
21: The center point p of the rail at the point p 1500 m ahead, 2
2 ... rail, 23 ... digital image stored in the image memory 13, 24 ... position where the point p can be seen in the digital image, 25 ... obstacle search area, 30 ... rail image, 3
1 ... binary image, 32 ... window for obtaining correlation value, 33
... Rail drawn using route data, 34 ... Actual rail, 34a ... Binary actual rail, 35m ... Displacement ρm, 35n ... Displacement pn, 36 ... Displacement ρm, ρ
Point corrected by n.

──────────────────────────────────────────────────続 き Continuing on the front page (72) Atsushi Nakamoto 4-22, Kannonshinmachi, Nishi-ku, Hiroshima City, Hiroshima Prefecture Inside the Hiroshima Research Laboratory, Mitsubishi Heavy Industries, Ltd. (72) Inventor Hiroshi Yamashita Hiroshi Kannonshinmachi 4, Nishi-ku, Hiroshima City, Hiroshima Prefecture Hiroshima Research Institute, Mitsubishi Heavy Industries, Ltd. (72) Inventor Kuniaki Wakusawa 4-622 Kannon Shinmachi, Nishi-ku, Hiroshima City, Hiroshima Prefecture, Hiroshima Research Institute, Mitsubishi Heavy Industries, Ltd. (72) Minami Shirakihara, Hiroshima Prefecture 5007 Itozakicho, Mihara-shi Mihara Works, Mitsubishi Heavy Industries, Ltd. JP-A-59-47663 (JP, A) JP-A-4-95889 (JP, A) JP-A-2-151563 (JP, A) JP-A-5-116626 (JP A) (58) investigated the field (Int.Cl. ^7, DB name) G01S 17/00 - 17/95 B61D 37/00

Claims (3)

(57) [Claims] 1. An image input device installed to take an image of the front of a running railway vehicle, an image memory storing an image taken by the image input device, and detecting a current position of the running vehicle. A position detecting means, and how a rail looks in an image taken by the image input device is calculated and stored in advance as a route data corresponding to a vehicle position detected by the position detecting means using a track condition. A route data memory to be stored, and a determination unit for determining whether there is an obstacle in an image stored in the image memory based on the route data stored in the memory and a position detection signal from the position detection unit And a notifying means for notifying a driver of the presence or absence of an obstacle in accordance with a result of the judgment by the judging means. 2. An image input device installed to take an image of the front of a traveling railway vehicle, an image memory storing an image captured by the image input device, and detecting a current position of the traveling vehicle. A position detecting means, and how a rail looks in an image taken by the image input device is calculated and stored in advance as a route data corresponding to a vehicle position detected by the position detecting means using a track condition. The route data memory to be saved and the route data memory
The stored route data is detected by the position detecting means.
Data reading means to read based on the output position signal
Stage and route data read from the route data memory
The center point of the left and right rails ahead of the specified distance based on
And calculate the position that can be seen on the screen.
A search area setting means for setting an obstacle search area;
From the image memory by the search area setting means.
Image data of the searched obstacle search area, and
An obstacle detecting device for a railway vehicle, comprising: determining means for determining whether there is an obstacle in data; and notifying means for notifying a driver of the presence or absence of an obstacle in accordance with the determination result of the determining means. . 3. An image input device installed to take an image of the front of a running railway vehicle, an image memory for storing an image taken by the image input device, and an image memo.
Binarize the image stored in the
Extraction means for extracting as an image, position detection means for detecting the current position of the running vehicle, and how the rail looks in the image taken by the image input device is detected by the position detection means and route data memory for storing are acquired by calculating values in advance by the trajectory condition as line data corresponding to the vehicle position, stored in the line data memory
Route data that has been detected by the position detection means.
Data reading means for reading based on the setting signal;
Based on the route data read from the route data memory
Calculating the center point of the left and right rails ahead of the specified distance
Means for calculating the position visible on the screen, and the route data
Based on the route data read from the
Rails that generate left and right rail images within a fixed distance range
Image generating means, and the image generated by the rail image generating means.
Correlation between the extracted image and the binarized rail image
Is found in the screen above
Correction means for correcting a point at the center of the rail ahead of a predetermined distance,
The rail center position corrected by this correction means
Search area setting means for setting an object search area;
Set by the search area setting means from the image memory
The image data of the obstacle search area is read out, and
A determining means for determining whether there is an obstacle in the vehicle, and an informing means for notifying a driver of the presence or absence of an obstacle in accordance with a result of the determination by the determining means. .