JP2020179798A - Turnout detection device and turnout detection method - Google Patents

Abstract

To provide a device and method capable of detecting a turnout without being affected by a signaler existing in a railway facility and lighting in a station yard.SOLUTION: A turnout detection device can surely photograph a rail track 5 illuminated by infrared light illumination means 3 by an infrared camera 2 even in darkness without being affected by a signaler existing in a railway facility and lighting in a station yard, and detect the rail track 5 and a turnout 50 by image processing because of including the infrared light illumination means 3 for illuminating the front of a vehicle by infrared light, the infrared camera 2 for photographing the rail track 5 illuminated by the infrared light illumination means 3, an image input part 41b for inputting an image photographed by the infrared camera 2, a rail track detection part 41c for detecting the rail track 5 from the image inputted by the image input part 41b, and a turnout detection part 41d for detecting the turnout 50 from the image by using a position of the rail track 5 on the image detected by the rail track detection part 41c.SELECTED DRAWING: Figure 3

Description

The present invention relates to a turnout detection device and a turnout detection method.

In a railway line, the track structure that divides one track into two is called a turnout. An example of a turnout is shown in FIG. As shown in FIG. 1, the turnout 50 is a mechanism for branching the line 5 into a plurality of lines, and has a point portion 51, a lead portion 52, and a crossing portion 53.
The point portion 51 is a portion where the tongue rail 50a for guiding the vehicle to the traveling course side is installed.

The lead portion 52 is a portion where a lead rail 50b connecting the tongue rail 50a and the crossing portion 53 is installed.
The crossing portion 53 is a portion where the railroad tracks 5 intersect, and a guardrail 50c is arranged.

The direction in which the vehicle branches with respect to the turnout 50 is referred to as facing, and the direction in which the vehicle merges with the track 5 is referred to as backward. In FIG. 1, the opposite direction is indicated by an arrow a, the back direction from the main line side is indicated by an arrow b, and the back direction from the branch line side is indicated by an arrow c.

Further, in the tongue rail 50a for changing the opening / closing direction of the turnout 50, the left and right tongue rails 50a do not move independently, and if one of the left and right is open, the other is closed. Is.
When the vehicle enters the turnout 50 backwards, if the opening and closing directions of the turnout 50 are different, the vehicle may not be able to pass through the turnout 50 and may derail.

JP-A-2018-39288

Various proposals have been made for the detection of turnouts, but in Patent Document 1, since image processing is used, there is a problem that it is easily affected by traffic lights in railway facilities and lighting in station premises. was there.

The present invention has been made in view of the above-mentioned prior art, and by installing an infrared camera in a vehicle and performing image processing on the acquired image data, the branching device is automatically recognized and driving to the vehicle driver. It provides support.

The turnout detection device according to claim 1 of the present invention, which solves the above-mentioned problems, is installed on the roof of the vehicle and has a photographing means for photographing the railroad track in front of the vehicle and an image processing for image-processing the image photographed by the photographing means. A turnout detection device including means, wherein the photographing means is an infrared camera corresponding to a wavelength of infrared light, and the image processing means inputs an image captured by the infrared camera. A branch that detects the turnout from the image by using a unit, a line detection unit that detects the line from the image input by the image input unit, and a position on the image of the line detected by the line detection unit. It is characterized by including a device detection unit.

In the turnout detection device according to claim 2 of the present invention, which solves the above problems, an infrared light illuminating means for illuminating a railroad track in front of the vehicle with infrared light is further installed on the roof of the vehicle. It is a feature.

In the turnout detection device according to claim 3 of the present invention, which solves the above problems, the image processing means further detects the position on the image of the line detected by the line detection unit and the turnout detection unit. It is characterized by including a turnout open / close recognition unit that recognizes the opening / closing of the turnout from the image by utilizing the position on the image of the turnout.

In the turnout detection device according to claim 4 of the present invention, which solves the above problems, the image processing means further detects the position on the image of the line detected by the line detection unit and the turnout detection unit. The type of the turnout and the opening / closing of the turnout are recognized from the image using the position on the image of the turnout, the approach direction of the turnout is known, and the type of the turnout and the opening / closing of the turnout are known. It is characterized by including an entry possibility determination unit for determining whether or not to enter the turnout based on the above.

The turnout detection method according to claim 5 of the present invention, which solves the above problems, captures a railroad track in front of the vehicle with an infrared camera installed on the roof of the vehicle, which is a photographing means corresponding to the wavelength of infrared light. A turnout detection method that processes an image taken by the infrared camera, wherein the image input step of inputting the image taken by the infrared camera and the image input in the image input step are used as described above. It is characterized by including a line detection step of detecting a line and a turnout detection step of detecting the turnout from the image by using the position on the image of the line detected in the line detection step.

The turnout detection method according to claim 6 of the present invention that solves the above problems is characterized in that the track in front of the vehicle is illuminated with infrared light by the infrared light illumination means installed on the roof of the vehicle. ..

The turnout detection method according to claim 7 of the present invention that solves the above problems is a position on the image of the line detected by the line detection step and a position on the image of the turnout detected by the turnout detection step. It is characterized by including a turnout opening / closing recognition step of recognizing the opening / closing of the turnout from the image by using the above.

The turnout detection method according to claim 8 of the present invention that solves the above problems is a position on the image of the line detected by the line detection step and a position on the image of the turnout detected by the turnout detection step. The type of the turnout and the opening / closing of the turnout are recognized from the image, the approach direction of the turnout is known, and the turnout is entered based on the type of the turnout and the opening / closing of the turnout. It is characterized by including an entry possibility determination step for determining the entry possibility of the above.

The turnout detection method according to claim 9 of the present invention that solves the above problems is a position on the image of the line detected by the line detection step and a position on the image of the turnout detected by the turnout detection step. The type of the turnout and the opening / closing of the turnout are recognized from the image, the approach direction of the turnout is known, and the turnout is entered based on the type of the turnout and the opening / closing of the turnout. A selection in which a first step including the approachability determination step and the line detection step and the turnout detection step or a second step including the line detection step and the approachability determination step is selected. It is characterized by having a process.

According to the present invention, even in the dark, it is possible to photograph a line with an infrared camera and reliably detect the line by image processing.
That is, since the present invention uses an infrared camera, it is possible to reliably detect the turnout without installing a dedicated sensor in each turnout. As a result, cost reduction can be achieved.

In particular, by using the infrared light illumination means, the influence of the traffic lights in the railway equipment and the illumination in the station yard is eliminated, and the detection accuracy of the railroad track and the turnout is improved.
Further, in addition to opening and closing the turnout, it is possible to determine whether or not to enter the turnout by recognizing the type of turnout.

By using a telephoto lens, the applicable distance can be set farther than that of a laser, so that the speed limit to the vehicle can be relaxed.

It is a schematic diagram which shows an example of a turnout. It is a schematic diagram which shows the structure of the turnout detection apparatus which concerns on 1st and 2nd Embodiment of this invention. It is a block diagram which shows the personal computer of the turnout detection apparatus which concerns on 1st Embodiment of this invention. It is a flowchart which shows the turnout detection method which concerns on 1st Embodiment of this invention. FIG. 5A is an explanatory diagram showing an example of a captured image, and FIG. 5B is a graph showing a luminance distribution in a region. It is explanatory drawing which shows the relationship between the line center coordinates and a line shape model. It is explanatory drawing which shows the turnout detection area. It is explanatory drawing which shows the type of a turnout. It is explanatory drawing which shows the detection example of a turnout. It is a block diagram which shows the personal computer of the turnout detection apparatus which concerns on 2nd Embodiment of this invention. It is a flowchart which shows the turnout detection method which concerns on 2nd Embodiment of this invention. FIG. 12A is an explanatory diagram showing a turnout in a state where the right line side is open, and FIG. 12B is a graph showing a luminance distribution at that time. FIG. 13A is an explanatory diagram showing a turnout in a state where the left line side is open, and FIG. 13B is a graph showing a luminance distribution at that time. It is explanatory drawing which shows the state which divided the left line side which the turnout is closed into four regions by the entry possibility judgment position and the detected line position. It is explanatory drawing which shows the state which divided the right line side which the turnout is closed into four regions by the entry possibility judgment position and the detected line position. It is a schematic diagram which shows the structure of the turnout detection apparatus which concerns on 3rd Example of this invention.

Hereinafter, the present invention will be specifically described with reference to the examples shown in the drawings.

A first embodiment of the turnout detection device and the turnout detection method according to the present invention will be described with reference to FIGS. 2 to 9.
As shown in FIG. 2, the turnout detection device according to the present embodiment includes an infrared camera 2 as a photographing means installed on the roof of an electric railway vehicle (hereinafter, simply referred to as “vehicle”) 1 and infrared. It is configured to include an infrared lighting device 3 which is an optical lighting means, and a personal computer (hereinafter, referred to as a measurement PC) 4 installed inside the vehicle 1.

The infrared camera 2 is an area camera corresponding to the band of infrared light, is installed in front of the roof of the vehicle 1, and photographs the track 5 on the traveling path side (in front of the vehicle 1) while the vehicle 1 is traveling. ..
Since the infrared camera 2 is used, it is possible to take a picture of the line even in the dark, and it is possible to reliably detect the line 5 by image processing.

Further, the infrared camera 2 is selectively equipped with a telephoto lens according to conditions such as the traveling speed of the vehicle 1.
By using a telephoto lens, the applicable distance can be set farther than that of a laser, so that the speed limit to the vehicle 1 can be relaxed.

The image taken by the infrared camera 2 is, for example, an image as shown in FIG. 5A.
The infrared illumination device 3 is an infrared illumination means for irradiating a line 5 in a range photographed by the infrared camera 2 with infrared light.

By photographing the line 5 illuminated by the infrared illumination device 3 with the infrared camera 2, there is an advantage that the influence of the traffic light in the railway facility, the illumination in the station yard, and the like is reduced.
As shown in FIG. 3, the measurement PC 4 includes an image processing unit 41 as an image processing means, a monitor 42 as a warning means, and a speaker 43 as a warning means.

The image processing unit 41 is a unit that analyzes an image taken by the infrared camera 2 to perform line detection, turnout detection, and turnout open / close recognition, and includes equipment data unit 41a, image input unit 41b, and line detection. A unit 41c, a turnout detection unit 41d, a turnout open / close recognition unit 41e, and a storage unit 41f are provided. In this embodiment, the image processing unit 41 is realized by installing software, but it can also be realized by hardware.

The equipment data unit 41a includes data on at least the installation height [mm] (hereinafter, camera height) of the infrared camera 2 and the installation angle [deg] (hereinafter, camera installation angle) of the infrared camera 2, and the reference brightness. This is the part where the distribution and the line shape model are input in advance.

The image input unit 41b is a portion for inputting image data captured by the infrared camera 2 in the image input process. This image data is sent to the line detection unit 41c and stored in the storage unit 41f.

<Track detection>
The camera height [mm], camera installation angle [deg], reference luminance distribution and line shape model are input from the equipment data unit 41a to the line detection unit 41c, and image data is input from the image input unit 41b.
In the track detection step, the track detection unit 41c detects the position (u [pix], V [pix]) of the traveling track 5 from the captured image which is the image data input from the image input unit 41c.

FIG. 5A shows an example of a captured image input from the image input unit 41c.
As shown in FIG. 5B, the horizontal distribution of the luminance value for one vertical position line on the image is an extremely low luminance value near the line surface.

From this feature, the difference between the brightness value of the track surface and the center of the gauge is used as the feature amount. That is, the difference between the brightness value of the line 5 and the center of the gauge is used as the feature quantity, and the position of the extremely low brightness value is referred to as a candidate for a plurality of lines 5 (hereinafter referred to as a line candidate point) based on the brightness value shown in FIG. 5 (b). ), And the centers of the left and right track candidate points are the candidate points of the gauge center coordinates C, respectively.

In addition, there is a strong correlation between the gauge center coordinates C in the image and the shape of the line 5.
For example, as shown in FIG. 6A, when the shape of the line 5 is curved to the right with respect to the traveling direction, the gauge center coordinates C are usually left and right in the left-right direction of the image (left and right with respect to the traveling direction). Direction. Hereinafter referred to as the horizontal axis direction) located on the right side.
Further, as shown in FIG. 6C, when the shape of the track 5 is a shape extending linearly toward the front in the vehicle traveling direction, the gauge center coordinate C is usually located at the center in the horizontal axis direction.

Further, as shown in FIG. 6E, when the shape of the track 5 is curved to the left with respect to the traveling direction, the gauge center coordinate C is usually located on the left side in the horizontal axis direction.
From this feature, as shown in FIGS. 6 (b), (d), and (f), a line shape model in which the shape of the line 5 changes depending on the gauge center coordinates C in the image is used. Feature evaluation points F are continuously arranged on the line shape model.

For example, for the gauge center coordinate C shown in FIG. 6 (a), the track shape model curved to the right shown in FIG. 6 (b), and for the gauge center coordinate C shown in FIG. 6 (c), FIG. The straight line shape model shown in (d) and the left-curved line shape model shown in FIG. 6 (f) are determined for the gauge center coordinates C shown in FIG. 6 (e).

Then, in the region A at an arbitrary V coordinate (a certain distance from the vehicle) of the image, a plurality of candidate points of the gauge center coordinates C are listed by raster scanning, and a track shape model corresponding to each candidate point is prepared.
The above-mentioned features are compared at the feature evaluation points F of each line shape model, and the candidate point of the gauge center coordinate C having the highest likelihood is detected as the line position.

<Switch detection>
In the turnout detection step, the turnout detection unit 41d detects the turnout 50 based on the line position obtained by the line detection unit 41c.
The types of branching of the turnout 50 are "branch to the left side" in FIG. 8 (a), "branch to the right side" in FIG. 8 (b), "merge from the left side" in FIG. 8 (c), and FIG. 8 (d). There are four types of "merging from the right side".

As shown in FIG. 8, of the two lead rails 50b on the left and right of the turnout 50, one of the lead rails 50b always exists in the gauge of the traveling track 5, regardless of the type of branching.
Therefore, the line detection unit 41c sets a turnout detection area (shown with hatching in FIGS. 7 and 8) G in a certain range around the designated turnout detection distance B, and within the turnout detection area G. The lead rail 50b is detected with.

That is, since the lead rail 50b has a shape relatively close to a straight line, a straight line D having various inclinations is searched for in the turnout detection area G as shown in FIG.
If the difference between the average luminance value in the straight line D and the average luminance value of the detected line 5 is within the threshold value among the plurality of searched straight lines D, the lead rail 50b is detected as a part of the turnout 50.

<Recognition of turnout opening / closing>
In the turnout open / close recognition process, the turnout open / close recognition unit 41e sets the lead rail 50b of the turnout 50 and the detection line based on the line position obtained by the line detection unit 41c and the turnout position obtained by the turnout detection unit 41d. By calculating the continuity of the brightness values between 5, the opening and closing of the turnout 50 is recognized.
Here, the lead rail 50b and the tongue rail 50a are schematically drawn as separate members in FIG. 1, but are generally an integral part.

Therefore, calculating the continuity of the brightness value between the lead rail 50b of the turnout 50 and the detection line 5, that is, the brightness value between the tongue rail 50a integrated with the lead rail 50b of the turnout 50 and the detection line 5. Is equivalent to calculating the continuity of.
As described above, the left and right tongue rails 50a for changing the opening / closing direction of the turnout 50 are in a state in which one of the left and right is open and the other is in a closed state.

Therefore, when opening and closing the turnout 50, it is recognized whether the tongue rail 50a is closed on the left or right side. Therefore, if either the left or right side of the tongue rail 50a is closed, the turnout 50 is recognized as closed.
Then, the turnout open / close recognition unit 41e recognizes that the turnout 50 is closed, that is, either the left or right side of the tongue rail 50a of the turnout 50 is closed, and the monitor 42 and the monitor 42 as warning means. Output to the speaker 43.

Therefore, the vehicle driver recognizes that either the left or right tongue rail 50a of the turnout 50 is closed, which is output as a warning to the monitor 42 and the speaker 43, and the vehicle driver guides the traveling track to some extent by the turnout 50. Can be grasped. This provides driving support for the vehicle driver.

The turnout detection method implemented by the turnout detection device in the present embodiment having the above configuration will be described with reference to the flowchart shown in FIG.

First, in the track detection step, the shape of the track 5 changes depending on the gauge center coordinates C, with the difference between the brightness value of the track surface and the center of the gauge as a feature quantity from the captured image which is the image data input from the image input unit 41c. Using the track shape model, the feature quantities described above are compared at the feature quantity evaluation points F of each track shape model, and the candidate point of the gauge center coordinate C having the highest likelihood is set as the position of the traveling track 5 by the track detection unit 41c. Detect (step S1).

Next, in the turnout detection step, the average luminance value in the straight line D was detected among the plurality of straight lines D searched in the turnout detection area G based on the line position obtained by the line detection unit 41c. If the difference between the average luminance values of the lines 5 is within the threshold value, the lead rail 50b is detected by the branch detection unit 41d as a part of the turnout 50 (step S2).

Subsequently, when the turnout is detected in step S2, in the turnout open / close recognition step, the lead of the turnout 50 is read based on the line position obtained by the line detection unit 41c and the turnout position obtained by the turnout detection unit 41d. By calculating the continuity of the brightness value between the rail 50b and the detection line 5 by the turnout opening / closing recognition unit 41e, the opening / closing of the turnout 50 is recognized (step S3).

Then, in step S3, when it is recognized that the turnout 50 is closed, that is, when it is recognized that either the left or right side of the tongue rail 50a of the turnout 50 is closed, the recognition result is branched. Output from the device open / close recognition unit 41e to the monitor 42 and the speaker 43 as warning means (step S4), and the entire process is completed (step S5).
On the other hand, if the turnout is not detected in step S2, the entire process is terminated (step S5).

As described in detail above, the turnout detection device and the turnout detection method of this embodiment have the following effects.

That is, since the image processing unit 41 includes an image input unit 41b, a line detection unit 41c, a turnout detection unit 41, and a turnout open / close recognition unit 41e, the line does not require a dedicated sensor to be installed in each turnout. By carrying out the detection step (step S1) and the turnout detection step (step S2) in order, the line 5 and the turnout 50 can be detected, whereby the cost can be reduced.

Further, by carrying out the turnout open / close recognition step (step S3), when it is recognized that the turnout 50 is closed, it is output to the monitor 42 and the speaker 43 as warning means (step S4). This provides driving support to the vehicle driver.

Further, since the infrared camera 2 is used, it is possible to take a picture of the line even in the dark, and it is possible to reliably detect the line 5 by image processing.
In particular, by using the infrared light illumination device 3, the influence of the traffic signal in the railway facility, the illumination in the station yard, and the like is eliminated, and the detection accuracy of the line 5 and the turnout 50 is improved.

If the infrared camera 2 is equipped with a telephoto lens, the applicable distance can be set farther than that of the laser, so that the speed limit to the vehicle can be relaxed.

A second embodiment of the turnout detection device and the turnout detection method according to the present invention will be described with reference to FIGS. 10 to 15.
This embodiment has the same basic configuration shown in FIG. 1 as in the first embodiment described above.

In this embodiment, as shown in FIG. 10, a point where a turnout open / close recognition unit 41e in the measurement PC4 is replaced with a turnout admissibility determination unit 40g, and a result processing unit 41h are provided in the measurement PC4. It is characterized by the added points.
Therefore, in the present embodiment, the equipment data unit 41a, the image input unit 41b, the line detection unit 41c, the turnout detection unit 41d, the line detection unit 41e, and the storage unit 41f in the image processing unit 41 are designated by the same reference numerals. The explanation will be omitted.

The position on the image of the line detected by the line detection unit 41e and the position on the image of the turnout 50 detected by the turnout detection unit 41d are input to the turnout determination unit 41g to enter the turnout. Based on the position on the image of the line detected by the line detection unit 41e and the position on the image of the turnout 50 detected by the turnout detection unit 41d, the approachability determination unit 41g shows whether or not the turnout 50 can be entered. Judgment is made based on the type of branching shown in 1 and the opening / closing direction of the turnout.

Table 1 shows the state of the turnout, the approach direction to the turnout, the type of turnout, and the opening / closing of the turnout.
For example, if the approach direction of the turnout is "opposing", it is possible to enter regardless of the type of branching and the type of opening / closing of the turnout.

Also, when the approach direction to the turnout is "backward from the main line", the type of branch is "merging from the left side", and the opening and closing of the turnout is "open on the left track side", entry is not possible.
In addition, when the approach direction to the turnout is "backward from the main line", the type of branch is "merging from the left side", and the opening and closing of the turnout is "open on the right track side", it is judged that the turnout is possible. .. The direction of approach to the turnout is known.

Therefore, the approachability determination unit 41g recognizes the opening / closing of the turnout 50 and the type of branching in the approach determination step of entering the turnout 50.

<Recognition of opening and closing of turnout>
The approachability determination unit 41g recognizes which of the left and right lines 5 the tongue rail 50a of the turnout 50 is in contact with in the approachability determination step of the turnout.
FIG. 12 (a) shows a turnout when the tongue rail 50a is in contact with the line 5 on the left side, that is, when the right line side is open, at the turnout determination position E (indicated by a broken line in the figure) of the turnout 50. ), And the brightness distribution at that time is shown in FIG. 12 (b).

FIG. 13A shows the turnout when the tongue rail 50a is in contact with the right line 5 at the turnout 50 entry possibility determination position E, that is, the left line side is open, and the brightness at that time. The distribution is shown in FIG. 13 (b).
As shown in FIGS. 12 (b) and 13 (b), since the brightness distribution near the position where the line 5 exists has a downwardly convex luminance distribution, the luminance distribution near the open side line has a downwardly convex distribution. There are two (minimum values) q.

On the other hand, in the luminance distribution near the closed track, only one convex distribution q appears downward because the traveling track 5 and the tongue rail 50a are in contact with each other.
Using this feature, a position where the brightness value is the minimum value and is similar to the average brightness value of the detected line 5 is searched from the detected left and right line positions toward the center of the gauge.
Then, it is recognized that the track side on which the minimum value is detected is open on the gauge center side from the track position.

<Branch type recognition>
The approachability determination unit 41g pays attention to the line side where the turnout 50 is closed in the type recognition of the branch.
As shown in FIG. 14, the closed left track side is divided into four regions e1, e2, e3, and e4 (indicated by a broken line in the figure) according to the entry possibility determination position E and the detected track position H.

For example, a rectangular region e1 is set in the direction from 9 o'clock to 12 o'clock centering on the contact point where the tongue rail 50a is in contact with the traveling track 5, and the square area e1 is set in the direction from 12 o'clock to 3 o'clock centering on the contact point. A square region e2 is set, a square region e3 is set from 6 o'clock to 9 o'clock with the contact point as the center, and a square region e3 is set from 3 o'clock to 6 o'clock with the contact point as the center. Set e4. At the entry possibility determination position E, the direction of the sleepers orthogonal to the track is 3:00 to 9:00, and the direction along the track position H is 6:00 to 12:00.

Further, the closed right track side as shown in FIG. 15 is divided into four regions e5, e6, e7, and e8 (indicated by a broken line in the figure) according to the entry possibility determination position E and the detected track position H.
For example, a rectangular region e5 is set from 9 o'clock to 12 o'clock centering on the contact point where the tongue rail 50a is in contact with the traveling line 5, and from 12 o'clock to 3 o'clock centering on the contact point. A square region e6 is set, a square region e7 is set in the direction from 6 o'clock to 9 o'clock with the contact point as the center, and a square region e7 is set in the direction from 3 o'clock to 6 o'clock around the contact point. Set e8.

Then, a straight line having a brightness value similar to that of the detection line 5 is detected in each region e1, e2, e3, and e4.
Depending on the detected area, the type of branch is recognized, such as branching if it is on the upper side and merging if it is on the lower side. Further, depending on the detected region, if it is on the left side, the branch or merge is recognized as the left side, and if it is on the right side, the branch or merge is recognized as the right side.

For example, in FIG. 14, since a straight line having a brightness value similar to that of the detection line 5 is detected in the region e2 on the upper left side, it is recognized as “branch on the left side” as the type of branching.
Further, in FIG. 15, since a straight line having a brightness value similar to that of the detection line 5 is detected in the region e6 on the upper left side, it is recognized as "branch on the left side" as the type of branching.
Then, based on Table 1, the approachability determination unit 41g to the turnout determines whether or not to enter the turnout 50 according to the type of branching and the opening / closing direction of the turnout. The direction of approach to the turnout is known.

<Result of decision on whether or not to enter the turnout>
The result processing unit 41h outputs the result of determining whether or not to enter the turnout by the combination of the opening / closing of the turnout and the type of the turnout to the monitor 42 and the speaker 43 as a result.
For example, as shown in Table 1, when the approach direction to the turnout is "backward from the main line", the type of branch is "merging from the left side", and the opening and closing of the turnout is "open on the left track side". , Outputs a warning as no entry. On the other hand, when the approach direction to the turnout is "backward from the main line", the type of branch is "merging from the left side", and the opening and closing of the turnout is "open on the right track side", the result is considered to be accessible. Output. The direction of approach to the turnout is known.

On the other hand, the result processing unit 41h outputs the position of the turnout 50 detected by the turnout detection unit 41d to the monitor 42 and the speaker 43.
Therefore, the vehicle driver outputs a warning to the monitor 42 and the speaker 43, the result of the decision on whether or not to enter the turnout determined by the turnout determination unit 41g, and the turnout 50 detected by the turnout detection unit 41d. Depending on the position, attention is drawn. This provides driving support for the vehicle driver.

The turnout detection method implemented by the turnout detection device in the present embodiment having the above configuration will be described with reference to the flowchart shown in FIG.
First, the infrared camera 2 acquires images for the number of images (step T1).

Next, the image data acquired by the infrared camera 2 is input to the image input unit 41b (step T2).
Subsequently, in the selection step, either the first step or the second step is selected (step T3). The first step is a turnout detection process including steps T4, T5, and T6. Further, the second step is a process of determining whether or not to enter the turnout including steps T4, T7, and T8. When either the first step or the second step is selected, step T4 is carried out in common.

That is, when the first step is selected, in the track detection step, the track detection unit 41c detects the position of the traveling track 5 from the captured image which is the image data input from the image input unit 41c (step T4). .. That is, the track detection unit 41c detects the position of the traveling track 5 in the same manner as in step S1 of the first embodiment.
Subsequently, in the turnout detection step, the turnout 50 is detected by the turnout detection unit 41d based on the line position obtained by the line detection unit 41c (step T5), and the presence or absence of the turnout is output (step T6). That is, the branch detection unit 41d detects the position of the turnout 50 in the same manner as in step S2 of the first embodiment.

After that, the presence / absence of the turnout is processed by the result processing unit 41h to the monitor 42 and the speaker 43 as a result (step T9).
On the other hand, when the second step is selected, the position of the traveling track 5 is detected by the track detection unit 41c in the track detection step as in the case where the first step is selected (step T4).

Subsequently, in the step of determining whether or not to enter the turnout, the opening and closing of the turnout is recognized depending on whether the tongue rail 50a of the turnout 50 is in contact with the left or right line 5, and attention is paid to the line side where the turnout 50 is closed. The type of the turnout is recognized, and based on Table 1, the decision unit 41g for determining whether or not to enter the turnout determines whether or not to enter the turnout based on the type of the turnout and the opening / closing direction of the turnout (step T7). Outputs whether or not the turnout can be entered (step T8).
After that, whether or not the turnout can be entered is processed by the result processing unit 41h to the monitor 42 and the speaker 43 as a result (step T9).

As described in detail above, the turnout detection device and the turnout detection method of this embodiment have the following effects.

That is, since the image processing unit 41 includes an image input unit 41b, a line detection unit 41c, a turnout detection unit 41, and a turnout determination unit 41g, a dedicated sensor should be installed in each turnout. First, the images for the number of images are acquired by the infrared camera 2 (step T1), then the image data acquired by the infrared camera is input to the image processing unit 41b (step T2), and in the selection step. A step of selecting the first step or the second step (step T3), and when the first step is selected, a line detection step (step T4), a turnout detection step (step T5), and a step of outputting the presence / absence of a turnout. (Step T6) is carried out in order, and when the second step is selected, the line detection step (step T4), the step of determining whether or not to enter the turnout (step T7), and whether or not to enter the turnout are output. By sequentially carrying out the steps (step T8), the line 5 and the turnout 50 can be detected, and whether or not the turnout can be entered can be determined. As a result, the result processing unit 41h can detect the monitor 42 and the speaker. As a result, it is processed to 43 (step T9), and cost reduction can be achieved.

In particular, the entry possibility determination unit 41g to the turnout recognizes the type of branching in addition to opening and closing the turnout, so that it is possible to determine whether or not to enter the turnout. This provides driving support to the vehicle driver.
Further, since the infrared camera 2 is used, it is possible to take a picture of the line even in the dark, and it is possible to reliably detect the line 5 by image processing.

In particular, by using the infrared light illumination device 3, the influence of the traffic signal in the railway facility, the illumination in the station yard, and the like is eliminated, and the detection accuracy of the line 5 and the turnout 50 is improved.
If the infrared camera 2 is equipped with a telephoto lens, the applicable distance can be set farther than that of the laser, so that the speed limit to the vehicle can be relaxed.

A third embodiment of the turnout detection device and the turnout detection method according to the present invention will be described with reference to FIG.
In this embodiment, as shown in FIG. 16, a visible light camera 2a and a visible light illumination device 3a are used in place of the infrared camera 2 and the infrared light illumination means 3 in the second embodiment. Other configurations are the same as shown in FIG.
Therefore, the turnout detection device of this embodiment has the following effects.

That is, as in the second embodiment, the image processing unit 41 includes an image input unit 41b, a line detection unit 41c, a turnout detection unit 41, and an entry possibility determination unit 41g into the turnout, so that each turnout is provided. First, the image data for the number of images is acquired by the infrared camera 2 (step T1), and then the image data acquired by the infrared camera is input to the image processing unit 41b (step T1) without installing a dedicated sensor. Step T2), the first step or the second step is selected in the selection step (step T3), and when the first step is selected, the line detection step (step T4), the turnout detection step (step T5), The step of outputting the presence / absence of the turnout (step T6) is carried out in order, and when the second step is selected, the line detection step (step T4), the step of determining whether or not to enter the turnout (step T7), By sequentially performing the step (step T8) of outputting whether or not to enter the turnout, it is possible to detect the line 5 and the turnout 50, and further to determine whether or not to enter the turnout. As a result, the processing unit 41h processes the monitor 42 and the speaker 43 (step T9), and the cost can be reduced.

In particular, the entry possibility determination unit 41g to the turnout recognizes the type of branching in addition to opening and closing the turnout, so that it is possible to determine whether or not to enter the turnout. This provides driving support to the vehicle driver.
If a telephoto lens is attached to the visible light camera 2a, the applicable distance can be set farther than that of the laser, so that the speed limit to the vehicle can be relaxed.

The turnout detection device and the turnout detection method of the present invention are widely available in industry.

1 Vehicle 2 Infrared camera 2a Visible light camera 3 Infrared light illumination device 3a Visible light illumination device 4 Measurement PC
5 Line 41 Image processing unit 41a Equipment data unit 41b Image input unit 41c Line detection unit 41d Turnout detection unit 41e Turnout open / close recognition unit 41f Storage unit 41g Result processing unit 42 Monitor 43 Speaker 50 Turnout 50a Tongue rail 50b Lead rail 50c Guard rail 51 Point part 52 Lead part 53 Crossing part A Area at any V coordinate (constant distance from the vehicle) of the image B Turnout detection distance C Track center coordinate D Straight line E Enterability judgment position F Feature evaluation point G Turnout detection area H Line position

Claims (9)

A turnout detection device including a photographing means installed on the roof of a vehicle and photographing a railroad track in front of the vehicle and an image processing means for image processing an image photographed by the photographing means.
The photographing means is an infrared camera corresponding to the wavelength of infrared light.
The image processing means
An image input unit for inputting an image taken by the infrared camera,
A line detection unit that detects the line from the image input by the image input unit, and
A turnout detection unit that detects the turnout from the image using the position on the image of the line detected by the line detection unit, and
A turnout detector, characterized in that it comprises. The turnout detection device according to claim 1, wherein an infrared light illuminating means for illuminating a railroad track in front of the vehicle with infrared light is further installed on the roof of the vehicle. The image processing means further
Turnout open / close recognition that recognizes the opening / closing of the turnout from the image by using the position on the image of the line detected by the line detection unit and the position on the image of the turnout detected by the turnout detection unit. Department and
The turnout detection device according to claim 1 or 2, wherein the turnout detection device is provided. The image processing means further
The type of the turnout and the opening / closing of the turnout are recognized from the image by using the position on the image of the line detected by the line detection unit and the position on the image of the turnout detected by the turnout detection unit. However, the approach direction of the turnout is known, and the approachability determination unit that determines whether or not to enter the turnout based on the type of the turnout and the opening and closing of the turnout,
The turnout detection device according to claim 1 or 2, wherein the turnout detection device is provided. A turnout detection method that photographs the railroad tracks in front of the vehicle with an infrared camera, which is an imaging means corresponding to the wavelength of infrared light, installed on the roof of the vehicle, and processes the image captured by the infrared camera. And
An image input process for inputting an image taken by the infrared camera, and
A line detection step of detecting the line from an image input in the image input step,
A turnout detection step of detecting the turnout from the image using the position on the image of the line detected in the line detection step,
A turnout detection method characterized by comprising. The turnout detection method according to claim 5, wherein the line in front of the vehicle is illuminated with infrared light by an infrared light illuminating means installed on the roof of the vehicle. Turnout open / close recognition that recognizes the opening / closing of the turnout from the image by using the position on the image of the line detected by the line detection step and the position on the image of the turnout detected by the turnout detection step. Process and
The turnout detection method according to claim 5 or 6, wherein the turnout detection method is provided. Using the position on the image of the line detected by the line detection step and the position on the image of the turnout detected by the turnout detection step, the type of the turnout and the opening and closing of the turnout can be opened and closed from the image. The approach determination step of recognizing, making the approach direction of the turnout known, and determining whether or not to enter the turnout based on the type of the turnout and the opening and closing of the turnout,
The turnout detection method according to claim 5 or 6, wherein the turnout detection method is provided. Using the position on the image of the line detected by the line detection step and the position on the image of the turnout detected by the turnout detection step, the type of the turnout and the opening and closing of the turnout can be opened and closed from the image. The approach determination step of recognizing, making the approach direction of the turnout known, and determining whether or not to enter the turnout based on the type of the turnout and the opening and closing of the turnout,
A selection step in which a first step including the line detection step and the turnout detection step or a second step including the line detection step and the entry possibility determination step is selected.
The turnout detection method according to claim 5 or 6, wherein the turnout detection method is provided.

Images