Line of sight confirmation system, line of sight confirmation program, line of sight confirmation method and control device

Abstract

An example of a system for determining whether or not something is foreseeable is disclosed. [Solution] The line of sight confirmation system 1 includes an information storage unit that stores first information associated with "time" and "kilometers and vehicle speed," and second information associated with "a photograph taken from the front of the vehicle" and "the time the photograph was taken," a departure time detection unit that detects the departure time based on changes in the photographed image, a time difference measurement unit that measures the difference between the time of the first information and the time of the second information by matching "the time detected by the departure time detection unit" with "the departure time detected based on changes in vehicle speed," a shooting position identification unit that associates the photographed image with the shooting position where the photograph was taken using the time of the first information, the time of the second information and the difference, and a control device that includes a line of sight determination unit that determines whether or not the traffic signal to be confirmed is visible using the photographed image whose shooting position has been identified.

Description

Technical Field

[0001] The present disclosure relates to a visibility confirmation system that determines whether a target traffic signal to be checked is visible from a predetermined position on a railway track.

Background Art

[0002] For example, in the invention described in Patent Document 1, when there is a captured image in which a traffic signal cannot be detected normally, the imaging position corresponding to the relevant captured image is determined as a position with low visibility of the traffic signal.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problem to be Solved by the Invention

[0004] It is difficult for a system for checking visibility (visibility) of railway traffic signals to adopt a configuration that specifies an imaging position using position information obtained by GNSS satellite positioning, for the following reasons.

[0005] That is, the position of a railway traffic signal is specified by kilometer post. The kilometer post is a distance measured along the track from a reference position (for example, Tokyo Station). In other words, stating that the position of a traffic signal is at a certain kilometer post means that the traffic signal exists at a position ○◯ km away from the reference position along the track.

[0006] Since the position information obtained by GNSS satellite positioning is not information specified by kilometer post, the method of specifying an imaging position based on position information obtained by GNSS satellite positioning is not appropriate for application to a visibility confirmation system for railway traffic signals. The present disclosure discloses an example of a visibility confirmation system in view of this point. [Means for solving the problem]

[0007] A line-of-sight confirmation system that determines whether a railway signal to be checked is visible from a predetermined position on the tracks should preferably have the following configuration requirements, for example: In other words, the constituent elements are: an information storage unit (21) that stores first information relating "time" and "kilometers and vehicle speed," and second information relating "a photograph taken from the front of the vehicle" and "the time the photograph was taken"; a departure time detection unit (31) that detects the time the vehicle departed based on changes in the photographed image; a time difference measurement unit (32) that corrects the discrepancy between the time of the first information and the time of the second information; a photographic position identification unit (33) that associates the photographed image with the kilometers indicating the location where the photographed image was taken (hereinafter referred to as the photographic position) using the time corrected by the time difference measurement unit (32) (hereinafter referred to as the corrected time); and a visibility determination unit (4) that determines whether the traffic signal to be checked is visible or not using the photographed image whose photographic position has been identified by the photographic position identification unit (33).

[0008] Furthermore, it is desirable for the time difference measurement unit (32) to correct the time difference of the second information relative to the time of the first information by recognizing the discrepancy between the "time detected by the departure time detection unit (31)" and the "departure time detected based on the change in vehicle speed".

[0009] This allows the visibility confirmation system to associate the "kilometer distance included in the first information" and the "captured image included in the second information" via the correction time. Therefore, the location of the photograph and the location of the signal can be accurately determined by the kilometer distance, making it possible to appropriately determine the visibility of the railway signal being checked.

[0010] Incidentally, the symbols in each of the parentheses above are just examples showing the correspondence with the specific configurations etc. described in the embodiments described later, and this disclosure is not limited to the specific configurations etc. indicated by the symbols in the parentheses above. [Brief explanation of the drawing]

[0011] [Figure 1] This is a block diagram showing the line-of-sight confirmation system according to the first embodiment. [Figure 2] This is an explanatory diagram showing the first piece of information. [Figure 3] This is an explanatory diagram showing the second piece of information. [Figure 4] This is an explanatory diagram of the error correction process in the forecast judgment unit. [Figure 5] This is an explanatory diagram of the judgment made by the forecast judgment unit. [Modes for carrying out the invention]

[0012] The following "Embodiments of the Invention" are examples of embodiments that fall within the technical scope of this disclosure. In other words, the features defining the invention as described in the claims are not limited to the specific configurations and structures shown in the embodiments below.

[0013] Unless otherwise specified, such as "one," each component described with a reference numeral is provided at least once. The look-aware system described herein comprises at least one of the components described with a reference numeral.

[0014] (First Embodiment) <1. Overview of the Foresight System> The line of sight system is a system for determining whether a railway signal to be checked is visible from a predetermined position on the track on the side of the vehicle's direction of travel (hereinafter referred to as the judgment start position).

[0015] The line-of-sight system 1 according to this embodiment comprises at least a storage unit 2, a preprocessing unit 3, and a line-of-sight determination unit 4, as shown in Figure 1. The preprocessing unit 3 and the line-of-sight determination unit 4 in this embodiment correspond to the control device 5. This control device 5 is implemented by software being executed on a computer.

[0016] A computer is an electronic computer comprising a CPU, ROM, RAM, and the like. The software is pre-stored in a non-volatile storage unit such as a ROM or an SSD. The software is executed by the CPU.

[0017] The storage unit 2 is a rewritable non-volatile storage device. The storage unit 2 stores raw data to be processed by the preprocessing unit 3. The preprocessing unit 3 processes the raw data into data suitable for determination by the visibility determination unit 4. The visibility determination unit 4 uses the image data processed by the preprocessing unit 3 to determine whether visibility is available.

[0018] <2.Details of Visibility System> <2.1 Storage Unit> The storage unit 2 is configured to include an information storage unit 21, a signal position storage unit 22, and the like.

[0019] <Information Storage Unit> The information storage unit 21 is a storage area in which first information and second information are stored. The first information and the second information correspond to the aforementioned raw data. The first information is information acquired via an ATS device (Automatic Train Stop device) mounted on a railway vehicle (hereinafter referred to as a vehicle).

[0020] Specifically, the first information is information in which "time" is associated with "kilometer post position and vehicle speed" (see Figure 2). The acquired first information is temporarily stored in the vehicle. Thereafter, when a predetermined timing arrives, the first information is transferred to the information storage unit 21.

[0021] The second information is information in which "photographed images on the forward side of the vehicle" acquired respectively by a telephoto camera (not shown) and a wide-angle camera (not shown) mounted on the vehicle are associated with "photographing time of the photographed images" (see Figure 3).

[0022] Hereafter, "captured images" refers to all images of the front of a vehicle taken with a telephoto camera and all images of the front of a vehicle taken with a wide-angle camera, unless otherwise specified. Similarly, "camera" refers to both telephoto and wide-angle cameras.

[0023] The "time of capture of the captured image" is the time based on the clock built into the camera. Therefore, in this embodiment, the absolute time of the "time of capture" and the absolute time of the "time" included in the first information do not necessarily coincide.

[0024] Furthermore, since the clocks of the telephoto camera and the wide-angle camera are synchronized, the time displayed on the telephoto camera and the time displayed on the wide-angle camera are the same. The second piece of information is also transferred to the information storage unit 21 at the same time as the first piece of information.

[0025] <Traffic light position memory section> The signal location storage unit 22 is a storage area that stores the correspondence between a signal, the kilometer marker indicating the installation location of the signal, and the distance at which continuous line of sight should be confirmed for the signal (hereinafter referred to as installation information).

[0026] In other words, the signal location storage unit 22 stores a set ("row" or "record") of identification information (e.g., a number) indicating a specific signal, the mileage, and the distance at which continuous visibility should be confirmed for that signal.

[0027] <2.2 Pre-processing section> As shown in Figure 1, the preprocessing unit 3 includes an acquisition unit 30, a departure time detection unit 31, a time difference measurement unit 32, a shooting position identification unit 33, a kilometer mileage detection unit 34, a position correction unit 35, and a grouping unit 36, among other execution units.

[0028] The acquisition unit 30 acquires the first and second information stored in the storage unit 2 and provides the necessary information to the departure time detection unit 31 to the position correction unit 35, etc. Each execution unit is implemented by the execution of software on the CPU, as described above.

[0029] <Departure Time Detection Unit> The departure time detection unit 31 detects the time the vehicle departs based on changes in the captured image. Specifically, as shown in Figure 3, the departure time detection unit 31 estimates the time when the entire captured image can be considered to have changed from a state where there was no change in the captured image as the "time the vehicle departed".

[0030] <Time difference measurement unit> The time difference measurement unit 32 measures the difference between the time of the first information and the time of the second information. Specifically, the time difference measurement unit 32 recognizes the difference between the time of the first information and the time of the second information by measuring the difference between the "departure time detected by the departure time detection unit 31" and the "departure time detected based on the change in vehicle speed".

[0031] In other words, the first piece of information includes "time" and "vehicle speed". The time difference measurement unit 32 considers the time when the vehicle speed changes from 0 as the "vehicle departure time (hereinafter referred to as the reference departure time)".

[0032] The time difference measurement unit 32 then measures the difference between the "departure time detected by the departure time detection unit 31" and the "reference departure time". <Filming location identification section> The shooting location identification unit 33 uses the time to associate the captured image with the kilometer mileage (hereinafter referred to as the shooting location) that indicates the location where the image was taken. That is, the first information includes "time" and "kilometer mileage". The second information includes "captured image" and "time of capture".

[0033] The time difference measurement unit 32 determines the difference between the "time" included in the first information and the "time" included in the second information. Therefore, when the shooting location identification unit 33 identifies the kilometer mileage indicating the shooting location, it can take this difference into account, thereby enabling a more accurate correspondence between the captured image and the shooting location.

[0034] <Kilometer detection unit> The kilometer distance detection unit 34 uses the size of the signal that appears in the captured image to detect the kilometer distance (length along the railway line = distance) between the location where the image was taken and the signal.

[0035] "The size of a traffic light that appears in a captured image (hereinafter referred to as "display size") refers, for example, to the number of pixels that display the portion of the traffic light from the right edge to the left edge (hereinafter referred to as the horizontal width), or the number of pixels that display the portion of the traffic light from the top edge to the bottom edge (hereinafter referred to as the vertical width)."

[0036] Specifically, for example, if the horizontal width of a traffic light appearing in a captured image is displayed using n pixels, or if the vertical width is displayed using n pixels, then "n" corresponds to the display size.

[0037] "Detecting the distance in kilometers between the location where the image was taken and the traffic light using the display size" can be performed, for example, as follows: That is, the display size is a function of the distance (in kilometers) between the traffic light and the location where the image was taken (hereinafter referred to as the distance function).

[0038] In other words, when L is the distance (kilometers) between the traffic light and the position where the image was taken, and S is the display size, then L = f(S). Since the actual length of the horizontal or vertical width of each traffic light is the same, the above distance function is common to all traffic lights. The kilometer detection unit 34 stores the above distance function.

[0039] In other words, the kilometer mile detection unit 34 identifies the target traffic light by using "the shooting location detected by the shooting location identification unit 33," "the captured image included in the second information," and "the installation information stored in the traffic light location storage unit 22."

[0040] Next, the kilometer detection unit 34 uses the display size of the identified traffic light (hereinafter referred to as the identified traffic light) and a distance function to detect the distance (kilometers) between the identified traffic light and the location where the captured image was taken.

[0041] Hereinafter, this distance (kilometers) will be referred to as the corrected reference distance. For the sake of ease of understanding, in the following explanation, the kilometer distance (the distance along the railway line) between a specific signal and the location where the photograph was taken is assumed to be equal to the straight-line distance between the specific signal and the location where the photograph was taken.

[0042] In this disclosure, the method for identifying (detecting) traffic lights in captured images is, in principle, not specified. For example, the computer according to this embodiment identifies traffic lights in captured images using a known algorithm, utilizing at least one of the aspect ratio of the figure surrounding the outline of the traffic light (see Figure 5) and the traffic light's unique shape features such as color and pattern within the figure.

[0043] <Position correction section> The position correction unit 35 corrects the position from which the captured image was taken using the "correction reference distance" and the "installation location of the specific traffic signal identified by the installation information." In other words, the position correction unit 35 corrects the shooting position detected by the shooting position identification unit 33 using the correction reference distance.

[0044] In other words, the "installation information stored in the traffic signal position memory unit 22" is the most accurate information indicating the location (kilometer distance) of the traffic signal. Therefore, by identifying the location where the photograph was taken based on the location (kilometer distance) of the traffic signal indicated in the installation information, the exact location where the photograph was taken can be determined.

[0045] The corrected reference distance is information that identifies the position in which a photograph was taken, based on a specific traffic signal. The position correction unit 35 then corrects the photographed position by comparing "the position obtained by adding or subtracting the corrected reference distance from the position (kilometers) of the traffic signal indicated in the installation information, according to whether the vehicle is traveling uphill or downhill" with "the photographed position detected by the photographed position identification unit 33".

[0046] When correcting the shooting position, the kilometer detection unit 34 calculates an accurate correction reference distance using a photographed image in which the kilometer distance from the traffic light is close enough to meet a certain standard (for example, a certain kilometer value within the range of 20m to 50m) and the error in the correction reference distance is small. The position correction unit 35 then adds or subtracts this correction reference distance from the position (kilometer distance) of the traffic light indicated in the installation information, measures the difference between the kilometer distance of the shooting position based on this correction reference distance and the kilometer distance associated with the photographed image, and uses this difference to correct the kilometer distance associated with each group of photographed images extracted and grouped for each traffic light, as will be described later.

[0047] The reason why the position correction unit 35 "corrects the shooting position detected by the shooting position identification unit 33" is as follows: The kilometer mile included in the first information is a value obtained by the ATS device using some method, and may contain errors.

[0048] For example, when using a method that integrates the vehicle's wheel speed, or when using a method that integrates the speed value obtained from a Doppler laser device or GNSS receiver, the difference between the kilometer mileage included in the first information and the actual (true) kilometer mileage increases as the vehicle travels further. Therefore, it is necessary to appropriately adjust (correct) the shooting position according to the required accuracy of the shooting position.

[0049] Therefore, in this embodiment, the ATS device that acquires the first information corrects the kilometer marker to the correct value using the known information from the ground beacons that are laid intermittently along the railway line when the train passes over them.

[0050] Furthermore, the acquisition unit 30 accurately considers the kilometer value after correction due to the passage of the ground beacon as the corrected value, and interpolates it to ensure that the change in kilometer trajectory is continuous, so that the change in kilometer trajectory is not discontinuous during the correction process. <Grouping section> The grouping unit 36 ​​extracts and groups images that should belong to the same group from a large number of continuously captured images, based on the requirements of the traffic signal to be confirmed (hereinafter referred to as the target traffic signal) and the distance at which continuous visibility should be confirmed among the installation information related to the target traffic signal, in other words, the starting position for the determination.

[0051] In other words, "visible" means that it is possible to see continuously from the starting position of the determination to the location of the target signal or the location immediately adjacent to the target signal (hereinafter referred to as the signal location). Therefore, even if the target signal can be seen from the starting position, if the target signal cannot be seen at any point along the route from the starting position to the signal, it will be determined that "visibility is poor."

[0052] The grouping unit 36 ​​then groups together images taken within the range from the judgment start position to the signal position from among the numerous images taken in succession. In other words, the numerous images belonging to a particular group represent a collection of images taken of a specific target signal.

[0053] <2.3 Forecast Judgment Section> The visibility determination unit 4 determines whether the target signal is visible to each of the above groups. In this case, the visibility determination unit 4, as a general rule, determines that "visibility is not possible overall" if the target signal cannot be seen from the starting position to the signal position.

[0054] Furthermore, the computer according to this embodiment, i.e., the line-of-sight determination unit 4, as described above, recognizes the overall shape of the target traffic light for each individual captured image belonging to the above group by utilizing at least one of the aspect ratio of the figure surrounding the outline of the traffic light (see Figure 5) and the traffic light's unique shape features such as colors and patterns within the figure.

[0055] In this case, if the degree of matching of the traffic signal's unique shape, such as the color or pattern within the figure, is above a predetermined level, the visibility determination unit 4 recognizes the overall shape of the target traffic signal and determines that it is "visible." If the degree of matching is below a predetermined level, the visibility determination unit 4 determines that it cannot recognize the overall shape of the target traffic signal and determines that it is "not visible."

[0056] Furthermore, in this process, the line of sight determination unit 4 uses the display size of the traffic light in the captured image, the distance from the shooting location of the captured image to the traffic light detected using a distance function, the distance associated with the captured image, and the distance information from the installation information of the target traffic light to determine whether the traffic light appearing in the captured image matches the target traffic light.

[0057] Furthermore, if the line of sight determination unit 4 does not match, even if the overall shape of the traffic light can be recognized, it determines that it is not the target traffic light because the mileage does not match, and therefore determines that "line of sight is not possible." This situation can occur, for example, when multiple traffic lights appear in the same captured image.

[0058] However, when detecting the target traffic light from the captured image, if the overall shape of the traffic light is recognized but it is determined to be "not visible" due to the above-mentioned discrepancy in the kilometer marker, the visibility determination unit 4 will use a captured image taken before the time the current image was taken to determine whether or not visibility is possible.

[0059] Specifically, for example, if there are n images belonging to a particular group, and the first to (j-1) images are all visible, but the j-th image becomes unseen due to the aforementioned mismatch in kilometer distance, the visibility determination unit 4 will, for example, if the difference between the (j-1) image and the j-th image is smaller than a predetermined difference, reverse the determination that the j-th image was unseen due to the mismatch in kilometer distance.

[0060] More specifically, for example, if the target traffic light could not be detected in a certain image, and the reason for this is a mismatch in the kilometer markers as described above, and the target traffic light is visible in all of the images immediately preceding the image in which it could not be detected (for example, the four preceding images), and the difference in the area of ​​the traffic light between the preceding image and the image in question is smaller than a predetermined standard, then the visibility determination unit 4 will determine that the traffic light is visible in that image as well (see Figure 4).

[0061] Furthermore, if the target traffic signal could not be detected and the reason was not the above-mentioned mismatch in mileage, the visibility determination unit 4 determines that visibility is poor for the captured image in which the target traffic signal could not be detected. Incidentally, the manner in which the above-mentioned discrepancies in kilometer markers occur can vary. For example, in this embodiment, the computer is pre-trained with the outlines of traffic lights. The computer then identifies the traffic lights based on the outlines it has learned, and measures the display size of the horizontal or vertical width of those outlines using the number of pixels.

[0062] However, the display size decreases as the distance (in kilometers) between the location where the image was taken and the traffic light increases. Furthermore, the display size, or the number of pixels, is a natural number without decimal places.

[0063] Therefore, if the kilometer detection unit 34 overestimates the number of pixels by just one, the kilometer distance from the shooting position of the captured image to the traffic light, calculated using the above distance function, will deviate significantly from the true value, and the line of sight determination unit 4 may determine that "the traffic light has been detected, but the error in the kilometer distance is greater than the standard, so it is not the target traffic light."

[0064] The above-mentioned preprocessing and line-of-sight determination are performed for both the images captured by the telephoto camera and the images captured by the wide-angle camera. The line-of-sight determination unit 4 then determines that line of sight is possible if the results obtained using the images captured by at least one of the cameras indicate that line of sight is possible.

[0065] In other words, for example, even if the judgment result using the image captured by the telephoto camera is "impossible," if the image captured by the wide-angle camera is "possible," then in this embodiment, it is determined that visibility is possible.

[0066] <3. Features of the line of sight confirmation system according to this embodiment> In the line-of-sight confirmation system 1 according to this embodiment, the "kilometer distance included in the first information" and the "captured image included in the second information" are associated via the correction time. Therefore, the shooting location and the location of the signal can be accurately identified by the kilometer distance, making it possible to appropriately determine the line of sight to the railway signal to be confirmed.

[0067] The line of sight confirmation system 1 includes a position correction unit 35 that corrects the position from which the captured image was taken using the "corrected reference distance" and the "installation location of the specific traffic signal identified by the installation information." This makes it possible to ensure the required accuracy of the shooting position even if there is a difference between the kilometer distance included in the first information and the actual (true) kilometer distance.

[0068] The line of sight confirmation system 1 includes a grouping unit 36 ​​that extracts and groups images that should belong to the same group from a large number of images taken in succession, based on the target traffic signal and the starting position for determination. This makes it possible to efficiently determine whether the line of sight is clear.

[0069] In the visibility confirmation system 1, if the target traffic light cannot be detected from the captured image due to a mismatch in mileage, the visibility determination unit 4 uses an image taken before the time the uncaptured image was taken to determine whether visibility is possible. This prevents inappropriate judgments from being made.

[0070] In the line-of-sight confirmation system 1, the determination of whether a path is visible is made using images captured by at least one of the cameras, either a telephoto camera or a wide-angle camera. This effectively prevents inappropriate judgments from being made.

[0071] The visibility confirmation system 1 recognizes the overall shape of the traffic light and determines whether it is visible or not by utilizing at least one of the aspect ratio of the figure surrounding the traffic light's outline and the traffic light's unique shape features, including the colors and patterns within the figure. This prevents the visibility confirmation system 1 from incorrectly determining that visibility is possible when, for example, a human would determine that visibility is impossible. (Other embodiments) In the embodiments described above, the determination of line of sight was performed using images captured with at least one of the cameras, a telephoto camera and a wide-angle camera. However, this disclosure is not limited thereto. That is, the disclosure may be configured such that, for example, the determination of line of sight is performed using images captured with a telephoto camera or a wide-angle camera.

[0072] In the above-described embodiment, a position correction unit 33 was provided. However, this disclosure is not limited thereto. That is, the disclosure may include, for example, a configuration in which the position correction unit 33 is abolished. Alternatively, it may include a configuration in which batch correction of the kilometer distance is not performed using captured images where the kilometer distance to the traffic signal is close enough to meet a certain standard.

[0073] In the above-described embodiment, the overall shape of the target traffic light is recognized by utilizing at least one of the aspect ratio of the figure surrounding the outline of the traffic light and the traffic light's unique shape features such as colors and patterns within the figure. However, this disclosure is not limited thereto.

[0074] In other words, the system may be configured to perform a two-stage determination process: first, it uses only the aspect ratio value of the figure surrounding the outline of the traffic light to exclude anything that falls outside a predetermined range, such as an object that is not a traffic light that is mistakenly detected; and then, for those whose aspect ratio value falls within a certain range, it uses the unique shape features of the traffic light, such as the color or pattern within the figure, to recognize it as a traffic light.

[0075] Furthermore, this disclosure is not limited to the embodiments described above, but is sufficient to be consistent with the intent of the disclosures described in the embodiments described above. Therefore, it may be a configuration in which at least two of the embodiments described above are combined, or a configuration in which any of the illustrated components or components described with reference numerals in the embodiments described above are omitted. [Explanation of symbols]

[0076] 1… Line of sight confirmation system 2… Storage section 3… Pre-processing section 4… Forecast Judgment Department 5… Control device 21... Information storage section 22... Traffic light position memory section 30… Acquisition Department 31… Departure time detection unit 32… Time difference measurement unit 33… Location identification section 34... Kilometer detection unit 35…Position correction section 36… Grouping section

Claims

1. In a line-of-sight confirmation system that determines whether a railway signal to be checked is visible from a predetermined position on the track on the side of the vehicle's direction of travel (hereinafter referred to as the "determination start position"), An information storage unit stores first information associating "time" with "kilometers and vehicle speed," and second information associating "a photograph taken from the front of the vehicle" with "the time the photograph was taken." A departure time detection unit detects the time the vehicle departed based on changes in the captured image, A time difference measuring unit that corrects the discrepancy between the time of the first information and the time of the second information, A shooting location identification unit that uses the time corrected by the aforementioned time difference measurement unit to associate the captured image with the kilometer mile indicating the location where the captured image was taken (hereinafter referred to as the shooting location), The system includes a line-of-sight determination unit that uses the captured image, whose shooting position has been determined by the aforementioned shooting position determination unit, to determine whether or not the traffic signal to be inspected is visible from the front lines. The aforementioned time difference measurement unit is a forecast confirmation system that corrects the discrepancy between the time of the first information and the time of the second information by matching "the time detected by the departure time detection unit" with "the departure time detected based on the change in vehicle speed".

2. A signal location storage unit stores the correspondence between a signal, the kilometer marker indicating the installation location of the signal, and the distance at which continuous line of sight should be confirmed for the signal. A kilometer distance detection unit that uses the size of the traffic light displayed in the captured image to detect the kilometer distance between the location where the image was taken and the traffic light, A position correction unit corrects the position in which the captured image was taken by using "the kilometer detected by the kilometer detection unit" and "the location of the traffic light that appeared in the captured image, which was detected using the correspondence stored in the traffic light location storage unit." The line of sight confirmation system according to claim 1, comprising:

3. The line of sight confirmation system according to claim 2, wherein the correction by the position correction unit can be performed when the distance to the traffic signal targeted by the distance detection unit falls below a predetermined distance.

4. The line-of-sight confirmation system according to claim 3, further comprising a grouping unit that extracts and groups images that should belong to the same group from a large number of continuously captured images, based on the requirements of the traffic signal to be confirmed and the starting position for determination.

5. The line of sight confirmation system according to claim 4, wherein, when the line of sight determination unit detects a traffic light from a captured image, if the traffic light could not be detected due to a mismatch in mileage, it is possible to determine whether or not it is possible to see the traffic light by using a captured image taken before the time the captured image was taken.

6. The line-of-sight confirmation system according to any one of claims 1 to 5, wherein the captured image is an image taken using at least one of a telephoto camera and a wide-angle camera.

7. The visibility determination unit uses at least one of the aspect ratio of the figure surrounding the outline of the traffic signal and the unique shape features of the traffic signal, including the colors and patterns within the figure, to recognize the overall shape of the traffic signal and determine whether or not it is visible, as described in any one of claims 1 to 5.

8. A visibility confirmation program that determines whether a railway signal to be checked is visible from a predetermined position on the track on the side in front of the vehicle's direction of travel, In the control device, A procedure for obtaining first information, which associates "time" with "kilometers and vehicle speed," and second information, which associates "a photograph taken from the front of the vehicle" with "the time the photograph was taken," A departure time detection procedure that detects the time a vehicle departs based on changes in the captured image, A time correction procedure for correcting the time difference between the time of the first information and the time of the second information, A procedure to identify the shooting location by associating the captured image with the kilometer mile indicating the location where the image was taken (hereinafter referred to as the shooting location) using the time corrected by the aforementioned time correction procedure, A procedure for determining whether a traffic light to be inspected is visible from a distance using a photographic image with an identified shooting location, The aforementioned time correction procedure is a procedure for correcting the discrepancy between the time of the first information and the time of the second information by matching the "time detected by the departure time detection procedure" with the "departure time detected based on the change in vehicle speed," A program to check the outlook and execute the program.

9. A method for determining whether a railway signal to be checked is visible from a predetermined position on the track on the side of the vehicle's direction of travel relative to the signal, wherein the control device is The steps include obtaining first information that associates "time" with "kilometers and vehicle speed," and second information that associates "a photograph taken from the front of the vehicle" with "the time the photograph was taken," A departure time detection step that detects the time the vehicle departed based on changes in the captured image, A time correction step to correct the time difference between the time of the first information and the time of the second information, Using the time corrected in the aforementioned time correction step, the steps are to associate the captured image with the kilometer mile indicating the location where the image was taken (hereinafter referred to as the shooting location) and to identify the shooting location, The system includes a visibility determination step that uses a captured image with an identified shooting location to determine whether or not the traffic light to be inspected is visible from the outside, The aforementioned time correction step is a forecast confirmation method that corrects the discrepancy between the time of the first information and the time of the second information by matching the "time detected by the departure time detection step" with the "departure time detected based on the change in vehicle speed".

10. A control device for determining whether a railway signal to be checked is visible from a predetermined position on the track on the side in front of the vehicle's direction of travel, An acquisition unit acquires first information that associates "time" with "kilometers and vehicle speed," and second information that associates "a photograph taken from the front of the vehicle" with "the time the photograph was taken." A departure time detection unit detects the time the vehicle departed based on changes in the captured image, A time difference measuring unit that corrects the discrepancy between the time of the first information and the time of the second information, A shooting location identification unit that uses the time corrected by the aforementioned time difference measurement unit to associate the captured image with the kilometer mile indicating the location where the captured image was taken (hereinafter referred to as the shooting location), The system includes a line-of-sight determination unit that uses the captured image, whose shooting position has been determined by the aforementioned shooting position determination unit, to determine whether or not the traffic signal to be inspected is visible from the outside. The aforementioned time difference measurement unit is a control device that corrects the discrepancy between the time of the first information and the time of the second information by matching the "time detected by the departure time detection unit" with the "departure time detected based on the change in vehicle speed".

Images