JP7340413B2 - Train position determination device and train position determination method - Google Patents

Description

The present invention relates to a train position determining device and the like that determines the relative position of a train with respect to a station's fixed position.

BACKGROUND ART Conventionally, a technology has been developed that uses images taken by a camera to determine the position of a train when it stops at a fixed position at a station. For example, Patent Document 1 discloses a technique for attaching a band-shaped tape to the side of a train and photographing the band-shaped tape with a camera installed at a station to determine whether the train has stopped at a fixed position.

Techniques for determining train positions using captured images basically require a calibration process that associates the coordinate system in the real world with the coordinate system in the captured image. For example, paragraph [0033] of Patent Document 1 discloses performing a calibration process using a chess board when installing a camera.

Japanese Patent Application Publication No. 2018-86872

However, in the calibration process using a chessboard, the chessboard supported by a worker is photographed at the same position as the side of the train where the strip tape is attached, and the imaged size of the grid-shaped chessboard captured in the photographed image is By comparing the size with the actual size, the size value per pixel of the image is calculated and initialized in the system. In this work, there is a danger of workers touching cars when photographing the chess board, so it must be done in time for trains or at night when commercial trains are not running. Therefore, there has been a need for a technique for calibration processing that does not involve the danger of contact wheels.

In addition, calibration processing is not only necessary every time the camera installation location is changed, but also periodically as part of maintenance because the camera installation position and orientation may shift over time. Operation can also be considered. Therefore, a calibration processing technique with excellent maintainability has been desired.

Additionally, in addition to calibration processing, there is a need for a technology that is easy to maintain and that can be performed without the risk of contact wheels, as it relates to shooting settings, including processing such as brightness adjustment of captured images. Ta.

The problem to be solved by the present invention is to solve the problem of the danger of treadmills as a technology related to photographic settings such as calibration when determining the relative position of a train with respect to the fixed position of a station using images taken with a camera. The goal is to propose a technology that is free of maintenance problems and has excellent maintainability.

The first invention for solving the above problem is:
Image processing is performed to convert images taken by a camera installed at a station to a side-view image of the train, including distortion correction, to take a side view of a train stopped at a fixed position at the station. An image processing means (for example, the image processing unit 242 in FIG. 6),
The position where a predetermined part of the train appears in the side view image when the train stopped at the fixed position is photographed (hereinafter referred to as "imaging reference position"), and the side view image processed by the image processing means. Determining means (for example, the train in FIG. 6 position determination unit 243);
Equipped with
The determining means is
Specification means for specifying an imaged portion of a vehicle body appendage of a predetermined size provided at a predetermined position of the train, which is reflected in the side view image processed by the image processing means (for example, the imaged object extraction in FIG. 6). part 244) and
Based on a conversion coefficient between the number of pixels of the imaging portion specified by the specifying means and the predetermined size, the relative distance of the train to the fixed position is calculated from the number of pixels misaligned between the imaging reference position and the imaging position. a relative distance calculation means (for example, the relative distance calculation unit 246 in FIG. 6),
has,
This is a train position determination device.

According to the first invention, for a captured image taken by a camera installed at a station in order to photograph a train stopped at a fixed position at a station from the side, a side view of the train including distortion correction is provided. Perform image processing to convert to image. Then, when a train stopped at a fixed position is photographed, the position where a predetermined part of the train appears in a side view image (imaging reference position), and the side view image of the predetermined part shown in the side view image. The relative position of the train with respect to the fixed position is determined by comparing the position in the middle (imaging position).

Then, in determining the relative position of the train with respect to a fixed position, the imaged part of the vehicle body appendage of a predetermined size installed at a predetermined position of the train in the side view image is specified, and the number of pixels of the imaged part is determined. The relative distance of the train with respect to the fixed position is calculated from the number of misaligned pixels between the imaging reference position and the imaging position, based on the conversion coefficient between and the predetermined size of the vehicle body appendage.

Therefore, when using images taken with a camera to determine the relative position of a train with respect to a station's fixed position, it is possible to use a technology that does not involve the danger of contact cars and is easy to maintain. technology can be realized.

The second invention is, in the first invention,
The vehicle body accessory is a window or a door of the train,
This is a train position determination device.

According to the second invention, the vehicle body accessory can be a window or a door of a train.

A third invention is, in the first or second invention,
a storage unit (for example, the storage unit 250 in FIG. 6) that stores the imaging reference position;
Based on the difference between the first number of pixels associated with the imaging portion identified by the identifying means for the first train and the second number of pixels associated with the imaging portion identified by the identifying means for the second train, Maintenance necessity determination means (for example, maintenance necessity determination means in FIG. 6 No judgment unit 247);
This is a train position determination device further comprising:

According to the third invention, the necessity of maintenance is determined based on the difference in the number of pixels related to the imaged portions identified for two different trains. Therefore, instead of conventional operations that require periodic maintenance, maintenance can be performed based on the necessity of maintenance determined by the device itself, reducing the time and human costs associated with maintenance. I can do it.

The fourth invention is any one of the first to third inventions,
A predetermined object whose part or all is a single color (hereinafter, a single color part is referred to as a "single color part") is installed at the station, or is installed accompanying the camera,
The camera performs photography based on a predetermined camera setting parameter (for example, camera setting parameter 21 in FIG. 6) that changes the brightness of the photographed image,
The relative installation position and installation direction of the predetermined object and the camera are determined so that the single color portion is captured in an image taken by the camera,
Camera setting automatic adjustment means (for example, in FIG. 6 automatic adjustment section 241),
This is a train position determination device further comprising:

According to the fourth invention, the parameter values of the camera setting parameters are automatically adjusted based on the brightness value of a single-color portion (single-color imaged portion) of a predetermined object in a photographed image. Therefore, it is possible to eliminate the need for maintenance related to automatically adjusted camera setting parameters.

A fifth invention is, in the fourth invention,
The camera setting automatic adjustment means calculates a dispersion index value indicating dispersion in brightness values for the single-color imaged portion, and when the dispersion index value satisfies a predetermined dispersion threshold condition indicating that dispersion is small. automatically adjusts parameter values related to gain or exposure time by determining that the dynamic range is exceeded.
This is a train position determination device.

According to the fifth invention, a variation index value indicating variation in brightness values is calculated for a single-color imaged portion, and when this variation index value satisfies a predetermined variation threshold condition indicating that the variation is small, It is determined that the dynamic range is exceeded, and parameter values related to gain or exposure time are automatically adjusted.

The principle of the fifth invention is briefly explained as follows. Cameras are equipped with lenses to adjust the focus and amount of light, and a phenomenon called vignetting or vignetting occurs in principle. Therefore, this phenomenon of peripheral light falloff may appear in a single-color imaged portion. Utilizing this phenomenon of peripheral light falloff, it is determined whether dynamic range overflow has occurred based on variations in brightness values in single-color imaged areas, and parameter values related to gain or exposure time are automatically adjusted.

A sixth invention is, in the fifth invention,
The camera setting automatic adjustment means automatically adjusts a parameter value related to exposure time when a predetermined flicker noise generation condition based on the photographed image is satisfied.
This is a train position determination device.

According to the sixth invention, when a predetermined flicker noise generation condition based on a photographed image is satisfied, the parameter value related to the exposure time is automatically adjusted. It becomes possible to automatically determine the occurrence of flicker noise and automatically adjust the exposure time to an appropriate value.

The seventh invention is
Image processing is performed to convert images taken by a camera installed at a station to a side-view image of the train, including distortion correction, to take a side view of a train stopped at a fixed position at the station. an image processing step (for example, step S5 in FIG. 11);
When a train stopped at the fixed position is photographed, the position where a predetermined part of the train appears in the side view image (hereinafter referred to as "imaging reference position"), and the position where the predetermined part of the train appears in the image processed side view image are determined. a determination step (for example, step S7 in FIG. 11) of determining the relative position of the train with respect to the fixed position by comparing the position of the predetermined part in the side view image (hereinafter referred to as "imaging position"); ,
including;
The determination step includes:
specifying an imaged portion of a vehicle body appendage of a predetermined size provided at a predetermined position of the train, which is shown in the image-processed side view image (for example, step S71 in FIG. 11);
Calculating the relative distance of the train to the fixed position from the number of pixels misaligned between the imaging reference position and the imaging position based on a conversion coefficient between the number of pixels of the identified imaging portion and the predetermined size; (For example, steps S73 to S75 in FIG. 11),
including,
This is a train position determination method.

According to the seventh invention, it is possible to realize a train position determination method that exhibits the same effects as the first invention.

The figure which shows the example of the whole structure of a train position determination device. A schematic diagram viewed from above of a station where a train position determination device is installed. A schematic diagram of the lead car of a train stopped at a station, viewed from the front in the direction of travel. FIG. 3 is a diagram for explaining train stop position determination. FIG. 3 is a diagram for explaining train position determination processing. FIG. 2 is a block diagram showing an example of a functional configuration of a processing device. Enlarged view of the camera installation part. A schematic diagram of an image taken by a camera. FIG. 3 is a diagram for explaining dynamic range over determination. FIG. 3 is a diagram for explaining dynamic range over determination. 5 is a flowchart showing the flow of processing by the processing device. Flowchart showing the flow of automatic adjustment processing. Flowchart showing the flow of automatic adjustment processing.

Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings. Note that the present invention is not limited to the embodiments described below, and the forms to which the present invention can be applied are not limited to the following embodiments. In addition, in the drawings, the same parts are denoted by the same reference numerals.

[overall structure]
FIG. 1 is a diagram showing an example of the overall configuration of a train position determination device 2 in this embodiment. This train position determination device 2 is for determining the position of the train 1 running on the rail R, and in FIG. Train 1 is also shown in the same state. Further, FIG. 2 is a schematic diagram of a station where the train position determining device 2 is installed, viewed from above, and FIG. 3 is a schematic diagram of the leading vehicle 11 of the train 1 stopped at the station, viewed from the front side in the direction of travel. It is a diagram.

The train position determination device 2 of this embodiment is provided for each target platform 4 (more specifically, a track) at the installation station, and includes a camera 20 and a process for determining the position of the train 1 based on the photographed image of the camera 20. A device 200 is provided. The platform 4 is also provided with a platform fence 3 that is placed on the edge of the track side to separate the top of the platform 4 from the track side. The section 300 is configured to control opening and closing of the platform door 31.

The camera 20 is installed on the target platform 4 and photographs the stopped train 1 from the side. In this embodiment, as shown in FIG. 2, the camera 20 is installed above the front end of the platform 4 in the direction of travel of the train 1 with its photographing range (photographing direction) facing the track side, and when the camera 20 is stopped. The leading car 11 of a train 1 is photographed from the side. Although the manner in which the camera 20 is installed is not particularly limited, for example, as shown in FIG. 3, the camera 20 is attached to a support 41 provided on the ceiling of the platform 4 so that its installation position and orientation can be adjusted. That is, the camera 20 photographs the front (tip side) portion of the leading vehicle 11 (more specifically, the side surface of the front portion on the platform 4 side) within the photographing range, and outputs the photographed image to the processing device 200. .

Then, the processing device 200 performs image processing on the image taken by the camera 20 to determine the stop position of the train 1 that has entered the platform 4 and stopped (stop position determination).

[About stopping position determination]
FIG. 4 is a diagram for explaining the determination of the stop position of the train 1 in this embodiment. As shown in FIG. 4, the processing device 200 performs image processing of the photographed image from the camera 20 and train position determination processing based on the photographed image (side view image) after the image processing in order to determine the stop position. .

(1) Image Processing In image processing, first, distortion correction is performed to correct distortion caused by lens aberration of the camera 20. Then, the photographed image after distortion correction is subjected to perspective projection transformation to obtain a side view image 5 in which the train 1 is viewed from the side.

This side view image 5 shows the front part of the leading car 11 of the stopped train 1 and a part of the platform 4. Specifically, the window 13 and vehicle door 15 provided on the side surface on the platform 4 side in the front part of the leading vehicle 11 are shown, and the platform door of the platform fence 3 installed on the forward side of the platform 4 in the direction of travel is shown. 31 and Tobukuro 33 are visible. Further, in the leading car 11 of the train 1 of this embodiment, a reference marker 17 is provided at a predetermined position on the front portion of the side surface facing the platform 4 . The reference marker 17 is provided, for example, by applying paint of a predetermined color or pasting a sticker. Therefore, this reference marker 17 also appears in the side view image 5.

(2) Train position determination process Next, in the train position determination process, an imaging object extraction process, a misaligned pixel count calculation process, and a relative distance calculation process are performed. FIG. 5 is a diagram for explaining the train position determination process, and the upper row (1) is a schematic diagram showing the result of the imaged object extraction process for the side view image 5 of FIG. 4. Further, the lower part (2) of FIG. 5 is a schematic diagram showing a reference image obtained by converting the imaging reference position used in the process of calculating the number of misaligned pixels.

Here, the imaging reference position is predetermined as a position where a predetermined part of the train 1 appears in a side view image when the train 1 stopped at a fixed position is photographed. In this embodiment, the reference marker 17 provided on the train 1 is set as a predetermined part of the train 1, and the reference marker 17 is extracted in advance from a side view image when the train 1 is stopped at a fixed position. An imaging reference position (hereinafter also referred to as "marker imaging reference position"), which is the position of the reference marker 17 in the side view image, is set as imaging reference position data 273 (see FIG. 6).

Then, by comparing the position (imaging position; hereinafter also referred to as "marker imaging position") of the reference marker 17 shown in the side view image with the marker imaging reference position, the train 1 relative to the fixed position is Determine the relative position of.

(2-1) Imaged object extraction process In the imaged object extraction process, the imaged part of the vehicle body appendage (hereinafter also referred to as the "appendage imaged part") that appears in the side view image is extracted, and the imaged part of the vehicle body appendage that appears in the side view image is extracted. The imaged portion of the reference marker 17 located therein is extracted, and the marker imaged position is obtained. The car body accessories are of a predetermined size and are provided at predetermined positions on the train 1, and can be, for example, windows 13 and vehicle doors 15. Through this process, the imaged portions of the window 13 and vehicle door 15 that are reflected in the side view image are specified, and the imaged portion of the reference marker 17 is specified. In the example of FIG. 5(1), as shown by solid lines in the figure, the imaged portions of the window 13 and vehicle door 15 in the side view image 5 are extracted, and the position of the reference marker 17 in the side view image 5 is extracted. A marker imaging position is obtained.

(2-2) Misaligned pixel number calculation process In the misaligned pixel number calculation process, the number of misaligned pixels between the marker imaging position acquired in the imaging object extraction process and the marker imaging reference position is calculated. In the example of FIG. 5, the marker imaging position acquired as the position of the reference marker 17 in the side view image 5 and the predetermined marker imaging reference position illustrated as the position of the reference marker 17A in the reference image The number of pixels Δd between the two is determined as the number of misaligned pixels.

(2-3) Relative distance calculation process In the relative distance calculation process, first, a conversion factor K is calculated from the number of pixels of the appendage imaged part extracted in the imaged object extraction process and the actual size of the vehicle body appendage. do. Here, the sizes of the window 13 and the vehicle door 15, which are accessories to the vehicle body, are known as specifications. Therefore, the number of pixels α1 of the window 13 (here, for example, the number of pixels in the width direction), which is the appendage imaged portion in the side view image 5, and the number of pixels of the vehicle door 15 (similarly, the number of pixels in the width direction, for example) α2. , the distance corresponding to the number of misaligned pixels Δd of the reference marker 17 can be determined from the relationship with the actual size (for example, width; hereinafter referred to as "regular size") of the window 13 or vehicle door 15. This distance corresponds to the relative distance to the fixed position of the stopped train 1 (the amount of deviation between the leading end position of the leading vehicle 11 and the fixed position).

Therefore, in this embodiment, the specified sizes of the window 13 and the vehicle door 15, which are vehicle body accessories, are set as the specified size data 275 (see FIG. 6). Then, according to the following equation (1), a conversion coefficient K is obtained by dividing the specified size of the vehicle body appendage by the number of pixels α of the appendage imaged portion.
Conversion coefficient K = specified size / number of pixels α ... (1)

More specifically, when the vehicle body accessory is the window 13, the conversion coefficient K is obtained by dividing the specified size of the window 13 by the number of pixels α1 of the imaged portion of the window 13 in the side view image 5. On the other hand, when the vehicle body accessory is the vehicle door 15, the conversion coefficient K is obtained by dividing the specified size of the vehicle door 15 by the number of pixels α2 of the imaged portion of the vehicle door 15 in the side view image 5. When a plurality of imaged parts of the same type of appendage are extracted like the vehicle door 15 in FIG. 5, the conversion coefficient K may be calculated using the average value of the number of pixels. Alternatively, a range including a plurality of vehicle body appendages in the side view image 5 may be used. For example, as shown in FIG. 5, the conversion coefficient K may be determined using the number of pixels α3 of the entire imaged portion of the window 13 and the two vehicle doors 15. In that case, the width of the vehicle body portion corresponding to the range may be set as the specified size. The length of the object for which the conversion coefficient K is calculated (the length of the appendage imaged portion) is determined to be as long as possible in the side view image 5, so that the calculation error of the conversion coefficient K can be suppressed to a small value.

Once the conversion coefficient K has been determined, the relative distance ΔD is calculated by multiplying the number of misaligned pixels Δd calculated in the relative distance calculation process by the conversion coefficient K according to the following equation (2).
Relative distance ΔD = conversion coefficient K x number of positional deviation pixels Δd (2)

Thereafter, it is determined whether the train 1 has stopped at a fixed position based on the determined relative distance ΔD. In this embodiment, as shown in FIG. 2, for example, the relative distance from the home position is within a predetermined value A11 for each of the over position (overtravel position) and the short position (unreached position) with respect to the home position. The range is predetermined as the fixed position stop range A1. Then, if the relative distance ΔD calculated by equation (2) is less than or equal to the distance A11, it is determined that the train 1 has stopped at a fixed position.

[Functional configuration]
FIG. 6 is a block diagram showing an example of the functional configuration of the processing device 200. As shown in FIG. 6, the processing device 200 includes an operation section 210, a display section 220, a communication section 230, a control section 240, and a storage section 250. can be configured as a computer.

As described above, the camera 20 is installed on the platform 4 to photograph the train 1 stopped at the station from the side, and is set to a predetermined camera setting parameter 21 that changes the brightness of the photographed image when changed. Shoot based on the following. In this embodiment, camera setting parameters 21 include analog gain, digital gain, and exposure time.

The operation section 210 includes push switches, dials, and the like, and the display section 220 includes an LED, a small liquid crystal display, and the like. The operating section 210 and the display section 220 are mainly used by workers during maintenance.

The communication unit 230 is a wired or wireless communication device realized by, for example, a wireless communication module, a router, a modem, a TA, a wired communication cable jack, a control circuit, etc. etc.). Photographed images input frame by frame from the camera 20 via the communication section 230 are stored in the storage section 250 as entrance-side photographed image data 255.

The control unit 240 is realized by electronic components such as a processor such as a CPU (Central Processing Unit) or a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), or an FPGA (Field-Programmable Gate Array), and is connected to each part of the device. Data input/output control is performed between. Then, it performs various arithmetic processing based on predetermined programs, data, data input via the communication unit 230, etc., and controls the operation of the processing device 200. This control section 240 includes an automatic adjustment section 241, an image processing section 242, a train position determination section 243, and a maintenance necessity determination section 247. These functional units may be processing blocks realized as software by executing a program, or may be circuit blocks realized by hardware circuits such as ASIC or FPGA. This embodiment will be described as a processing block realized as software when the control unit 240 executes the automatic adjustment program 251 and the stop position determination program 253.

Prior to determining the stop position, the automatic adjustment unit 241 performs automatic adjustment processing using the captured image from the camera 20 to automatically adjust parameter values of predetermined camera setting parameters. In this embodiment, the values of analog gain, digital gain, and exposure time of the camera setting parameters 21 are automatically adjusted. Then, by outputting the photographed image when the automatic adjustment processing is normally completed to the image processing unit 242, it is used as a target for stop position determination. According to this, it becomes possible to perform the stop position determination using a captured image of appropriate image quality that has been subjected to brightness adjustment, etc., and it is possible to improve the accuracy of the stop position determination.

Here, automatic adjustment processing will be explained. FIG. 7 is an enlarged view of the installation part of the camera 20 shown in FIG. 3. Further, FIG. 8 is a schematic diagram of an image 6 taken by the camera 20. As shown in FIG. 7, the camera 20 of this embodiment includes an adjustment board 23 attached so that part or all of the surface facing the camera 20 overlaps with the photographing range for automatic adjustment processing. For example, this can be achieved by using a plate whose entire surface is a single color as the adjustment board (predetermined object) 23 and attaching it to the camera 20. Although the color is not particularly limited as long as it is a single color, it may be white, for example. In this embodiment, as shown by the broken line in FIG. 8, the surface (single color portion) of the adjustment board 23 is captured as a band-shaped area (single color imaged portion) 61 over the entire width at the bottom end of the photographed image 6. Thus, the relative installation position and installation orientation of the adjustment board 23 and camera 20 are determined.

Note that the entire surface of the adjustment board 23 may not be of a single color, but a portion of the surface may be of a single color, or may be of a type that transmits light. It is sufficient if it is attached to 20. Furthermore, the single-color imaging portion does not need to be long enough to span the entire width like the single-color imaging portion 61 in FIG. It is sufficient if it is configured as follows. For example, if it is possible to capture the brightness gradient by capturing a single color image part with a length of 1/3 or more of the left and right angle of view, then the size of the adjustment board 23 and the adjustment board 23 should be adjusted so that that length can be captured. It is only necessary that the relative position and angle between the board 23 and the camera 20 be set.

Furthermore, it is not limited to the bottom end of the photographed image, but may also be configured such that a vertically band-shaped single-color imaging portion is captured at the left end or right end. The installation position and installation direction relative to the camera 20 may be determined in advance. In that case as well, it is sufficient that the camera 20 is configured so that a single color imaged portion can be captured with a length that can capture at least the brightness gradient in the brightness distribution of the camera 20.

Furthermore, the adjustment board 23 is not limited to the configuration in which the adjustment board 23 is attached so that the single-color imaged portion is captured on the camera 20, but may be installed separately from the camera 20 at the station. For example, it is also possible to install an adjustment board so that it is suspended under the roof of the platform 4 so that it is visible in the photographing range of the camera 20. In that case, the relative installation position of the adjustment board and the camera 20 should be determined so that the single color part of the adjustment board can be photographed with a length that can capture the brightness gradient in the brightness distribution of the camera 20. Direction is determined.

Alternatively, even without using an adjustment board, if there is a single color part on the side of the platform fence 3 or the side of the platform 4 of the train 1 that is within the shooting range, you can simply select the area in the shot image where that part is captured. It can also be used as a one-color imaging area. According to this, it becomes possible to perform automatic adjustment processing without attaching the adjustment board 23 to the camera 20.

In the automatic adjustment process, dynamic range over determination, flicker occurrence determination, and center brightness determination are performed using the brightness value of the single-color imaged portion 61 in the photographed image, and the camera setting parameters 21 are adjusted according to the determination results. Automatically adjusts each value of analog gain, digital gain, and exposure time. In this embodiment, the center line of the single-color imaging portion 61 is defined as the inspection line L1, and each determination is made based on the luminance value on this inspection line L1.

(1) Dynamic range over determination FIGS. 9 and 10 are diagrams for explaining dynamic range over determination. FIG. 9 shows each position on the inspection line L1 centered on the center C shown in FIG. It is a graph in which the luminance value distribution on the inspection line L1 under a standard illumination environment is plotted as an axis. In addition, FIG. 10 shows the luminance value distribution under the standard illuminance environment (middle row), the luminance value distribution on the inspection line L1 under the high-illuminance environment during the day (upper row), and the low-illuminance environment at night. It is a diagram showing the luminance value distribution (lower stage) on the lower inspection line L1 along a common horizontal axis.

As described above, the camera 20 suffers from the phenomenon of peripheral light falloff. In the automatic adjustment process, a dynamic range over-determination is performed using the characteristics of the luminance value distribution due to this peripheral light falloff. Specifically, as shown in the luminance value distribution on the inspection line L1 in the middle of FIGS. A phenomenon may occur in which the luminance value of the peripheral portion (the end on the inspection line L1) is lower than that of the surrounding area.

On the other hand, under the high illuminance environment shown in the upper part of FIG. The luminance value distribution above has a flat distribution except for the edges (so-called blown-out highlight state). Similarly, under the low illuminance environment shown in the lower part of FIG. 10, the ideal value of the portion indicated by the dashed line on both sides is below the lower limit of the luminance value, and the luminance value distribution on the inspection line L1 is flat in the portion on both sides. distribution (so-called black-filled state). In this way, when there is a blown-out highlight state, the luminance gradient has a flat distribution in the central portion (near center C), and when there is a blown-out shadow state, the distribution is flat on both sides. Therefore, by monitoring the difference in brightness distribution, it is possible to determine whether the image is overexposed or underexposed.

In addition, saturation of the image sensor installed in the camera occurs regardless of the upper and lower limit ranges of the brightness value, as shown by the SS line in Figure 10, even though the brightness value is output within the normal range. A distribution with a flat luminance gradient occurs in the central portion (near center C). That is, by incorporating this method of monitoring the brightness gradient, it is also possible to monitor the signal saturation inside the image sensor mounted on the camera.

Therefore, the variations SU and SS of each luminance value on the inspection line L1 under a high illuminance environment in the upper row are larger than the variations T of each luminance value on the inspection line L1 under a standard illuminance environment shown in the middle row, etc. becomes smaller. Similarly, the variation SL of each luminance value on the inspection line L1 under the low-light environment in the lower row is smaller than the variation T under the standard illuminance environment in the middle row (T>SU, SL, SS). .

Therefore, in the dynamic range over determination, a variation index value indicating the variation in luminance values on the inspection line L1 is calculated, and if the variation index value satisfies a predetermined variation under/under threshold condition indicating that the variation is small, the dynamic range is Determined as over. In this embodiment, the standard deviation of the luminance values on the inspection line L1 is used as the variation index value. The variation threshold condition is set as "the variation index value (standard deviation) is less than or equal to a predetermined value" or the like.

If the obtained variation index value satisfies the variation threshold condition, it is determined that the dynamic range is exceeded. In other words, it is determined that there is a blown-out highlight state, a blown-out shadow state, or a state of signal saturation inside the image sensor.

After that, the analog gain is adjusted depending on whether the problem is overexposure, underexposure, or signal saturation inside the image sensor, and in some cases, the exposure time is adjusted. Whether the image is in a blown-out highlight state or a blown-out shadow state can be determined based on the level of the luminance value on the inspection line L1.

(2) Flicker occurrence determination In flicker occurrence determination, occurrence of flicker noise is determined. For example, the occurrence of flicker is determined using the variation index value determined in the dynamic range over determination. Specifically, since captured images can be acquired at predetermined time intervals such as every frame, the difference between the dispersion index value of the currently captured image and the dispersion index value of the previous captured image indicates that there is a large amount of dispersion. When the predetermined excessive variation threshold condition shown in FIG. The excessive variation threshold condition is set as "the variation index value (standard deviation) is greater than or equal to a predetermined value". If it is determined that flicker noise has occurred, the exposure time is adjusted.

(3) Center brightness determination In the center brightness determination, first, the center brightness of the single-color imaging portion 61 is determined. For example, the brightness value at the center (center C) of the inspection line L1 is set as the center brightness. The maximum value or average value of the brightness values on the inspection line L1 can also be set as the center brightness. Then, the digital gain is adjusted depending on whether the calculated center brightness is within a predetermined tolerance range. For example, the upper and lower limits of the allowable range are set as the allowable range data 271 (see Figure 6), and the digital gain is adjusted when the center luminance exceeds the upper limit or is less than the lower limit of the allowable range. .

Returning to FIG. 6, the image processing unit 242 is a functional unit that performs image processing, and performs distortion correction on the captured image input from the automatic adjustment unit 241, and performs perspective projection transformation on the captured image after distortion correction. , generate a side view image. The side view image can also be said to be a bird's-eye view conversion image viewed from the side.

The train position determination unit 243 is a functional unit that performs train position determination processing, and includes an imaged object extraction unit 244, a misaligned pixel count calculation unit 245, and a relative distance calculation unit 246.

The imaged object extraction unit 244 is a functional unit that performs imaged object extraction processing, and uses various image recognition techniques to extract the imaged part of an appendage appearing in the side view image, and also extracts the imaged part of the reference marker. Obtain the marker imaging position.

The positional deviation pixel number calculation unit 245 is a functional unit that performs positional deviation pixel number calculation processing, reads a preset marker imaging reference position from the imaging reference position data 273, and calculates the positional deviation pixel number Δd with respect to the marker imaging position. calculate.

The relative distance calculation unit 246 is a functional unit that performs relative distance calculation processing, and calculates a conversion coefficient K from the number of pixels α of the appendage imaging area and the specified size of the vehicle body appendage, and calculates the conversion coefficient K and the position. Based on the number of misaligned pixels Δd, the relative distance ΔD of the train 1 stopped at the station from the fixed position is calculated. It is also determined whether the position is over the normal position or the position is short.

The maintenance necessity determination unit 247 is a functional unit that performs maintenance necessity determination processing when it is determined that the train 1 has stopped at a fixed position, and determines the necessity of maintenance using the maintenance history data 277. The maintenance here refers to maintenance related to correction of the imaging reference position data 273 or inspection of the installation position and installation orientation of the camera 20.

To this end, in this embodiment, the number of pixels of the appendage imaged portion and the values of the conversion coefficient K when it is determined that the train 1 has stopped at a fixed position in the stop position determination are accumulated and saved as maintenance history data 277. I'll keep it. That is, each time it is determined that the train 1 has stopped at a fixed position, the maintenance history data 277 includes the number of pixels of the appendage imaged portion extracted at that time and the conversion coefficient K calculated at that time. are added in association with the date and time.

In the maintenance necessity determination process, the maintenance necessity determination unit 247 first sets the number of pixels of the past appendage imaged portion stored in the maintenance history data 277 as the first pixel number, and determines the current stop position. The number of pixels of the imaged part of the appendage is set as the second number of pixels, and the necessity of maintenance is determined based on the difference therebetween. The number of pixels in the attached object imaging area when the train 1 stops at a fixed position should not change significantly unless there is a deviation in the installation position or orientation of the camera 20, which is why it has changed significantly. In this case, the accuracy of stopping position determination may be reduced.

Specifically, for example, if the difference between the first number of pixels and the second number of pixels is less than or equal to a predetermined threshold, it is determined that maintenance is required. The first number of pixels may be the number of pixels of the appendage imaged portion related to the previous stop position determination, or may be the average number of pixels of the appendage imaged portion related to the stop position determination for a predetermined number of times in the past. , it may be the number of pixels of the appendage imaging portion related to the stop position determination immediately after the previous maintenance.

Note that a configuration may be adopted in which the necessity of maintenance is determined using a conversion coefficient K instead of the number of pixels of the appendage imaged portion. In that case, the past conversion coefficient K stored in the maintenance history data 277 is set as the first conversion coefficient K, the conversion coefficient K related to the current stop position determination is set as the second conversion coefficient K, and the difference between them is set as the first conversion coefficient K. The necessity of maintenance is determined based on the following.

Subsequently, when the maintenance necessity determining unit 247 determines that maintenance is necessary, it performs notification control to that effect. For example, this can be achieved by displaying a message to the station staff or the like to the effect that maintenance is required on a display monitor installed at a suitable location within the station premises. Specifically, a message prompting the user to adjust the installation position and orientation of the camera 20 and to reset the imaging reference position data 273 is displayed.

Then, the maintenance necessity determining unit 247 adds the number of pixels and the conversion coefficient K of the appendage imaged portion related to the current stop position determination to the maintenance history data 277.

The storage unit 250 is realized by a storage medium such as an IC memory, a hard disk, or an optical disk. In this storage unit 250, programs for operating the processing device 200 and realizing various functions of the processing device 200, data used during the execution of the program, etc. are stored in advance, or are stored each time processing is performed. Memorized temporarily. For example, the storage unit 250 stores an automatic adjustment program 251, a stop position determination program 253, approach side photographed image data 255, parameter setting data 260, permissible range data 271, imaging reference position data 273, and a specified size. Data 275 and maintenance history data 277 are stored.

The automatic adjustment program 251 is a program for causing the control section 240 to function as the automatic adjustment section 241. Further, the stop position determination program 253 is a program for causing the control unit 240 to function as the image processing unit 242, the train position determination unit 243, and the maintenance necessity determination unit 247.

The parameter setting data 260 includes at least the initial values 261 of analog gain, digital gain, and exposure time, which are the objects of adjustment by the automatic adjustment unit 241, among the camera setting parameters 21, and the current values 263 of these analog gain, digital gain, and exposure time. Store.

[Processing flow]
FIG. 11 is a flowchart showing the flow of processing performed by the processing device 200. The processing described here can be realized in the processing device 200 by the control unit 240 reading out the automatic adjustment program 251 and the stop position determination program 253 from the storage unit 250 and executing them.

First, the automatic adjustment unit 241 starts automatic adjustment processing (step S1). FIG. 12 is a flowchart showing the flow of automatic adjustment processing. Further, FIG. 13 is a diagram showing the flow of processing that follows when the dynamic range over determination is negative in step S107 of FIG. 12.

As shown in FIG. 12, in the automatic adjustment process, the automatic adjustment unit 241 first reads an initial value 261 from the parameter setting data 260, and initializes the analog gain, digital gain, and exposure time that are to be adjusted (step S101). ). Through this process, the values of analog gain, digital gain, and exposure time are automatically set as their initial values.

Then, if the photographed image is acquired from the camera 20 (step S103: YES), the automatic adjustment unit 241 acquires the luminance value of each pixel on the inspection line L1 (see FIG. 8) (step S105).

Subsequently, the automatic adjustment unit 241 performs a dynamic range over determination, and if the determination is negative (step S107: NO), the process moves to step S131 in FIG. 13. On the other hand, if it is determined that the dynamic range has exceeded (affirmative determination) (step S107: YES), it is determined whether it is a blown-out highlight state or a blown-up shadow state, and the process branches (step S109).

That is, when the automatic adjustment unit 241 determines that the overexposure state is present, the automatic adjustment unit 241 refers to the current value 263, and if the analog gain is not the minimum gain (step S111: NO), decreases the analog gain by one step and changes the camera settings. The parameters 21 are updated (step S113). At that time, the analog gain of the current value 263 is also rewritten with the updated value. After that, the process returns to step S103.

Further, the automatic adjustment unit 241 refers to the current value 263 when the analog gain is the minimum (step S111: YES), and shortens the exposure time by one step when the exposure time is not the minimum (step S115: NO). The camera setting parameters 21 are updated (step S117). At that time, the current value 263 of exposure time is also rewritten with the updated value. After that, the process returns to step S103.

Then, if the exposure time is the minimum (step S115: YES), the automatic adjustment unit 241 performs control to notify an error and notify that maintenance is required, since both the analog gain and the exposure time cannot be adjusted. (Step S119). After that, the process returns to step S103.

On the other hand, if the automatic adjustment unit 241 determines in step S109 that the blackout condition is present, the automatic adjustment unit 241 refers to the current value 263, and increases the analog gain by one step if the analog gain is not the maximum gain (step S121: NO). The camera setting parameters 21 are updated (step S123). At that time, the analog gain of the current value 263 is also rewritten with the updated value. After that, the process returns to step S103.

Further, if the analog gain is the maximum (step S121: YES), and if the exposure time is not the maximum (step S125: NO), the automatic adjustment unit 241 increases the exposure time by one step and sets the camera setting parameter 21. Update (step S127). At that time, the current value 263 of exposure time is also rewritten with the updated value. After that, the process returns to step S103.

Then, when the exposure time is the maximum (step S125: YES), the automatic adjustment unit 241 controls to notify an error and notify that maintenance is required, since both the analog gain and the exposure time cannot be adjusted. (Step S129). After that, the process returns to step S103.

Next, in step S131 in FIG. 13, the automatic adjustment unit 241 performs center brightness determination. When the center brightness exceeds the upper limit of the allowable range set in the allowable range data 271, the process moves to step S133; when the center brightness is less than the lower limit of the allowable range, the process moves to step S139; and when it is within the allowable range, the process moves to step S139. The process moves to step S145.

First, in step S133, the automatic adjustment unit 241 refers to the current value 263, and if the digital gain is not the minimum (step S133: NO), decreases the digital gain by one step and updates the camera setting parameter 21 (step S135). ). At that time, the digital gain of the current value 263 is also rewritten with the updated value. After that, the process returns to step S103 in FIG. 12.

On the other hand, when the digital gain is the minimum (step S133: YES), the automatic adjustment unit 241 performs control to notify an error and notify that maintenance is required (step S133: YES). Step S137). After that, the process returns to step S103 in FIG. 12.

Next, in step S139, the automatic adjustment unit 241 refers to the current value 263, and if the digital gain is not the maximum (step S139: NO), increases the digital gain by one step and updates the camera setting parameter 21 (step S141). At that time, the digital gain of the current value 263 is also rewritten with the updated value. After that, the process returns to step S103 in FIG. 12.

On the other hand, if the digital gain is the maximum (step S139: YES), the automatic adjustment unit 241 performs control to notify an error and notify that maintenance is required (step S139: YES). Step S143). After that, the process returns to step S103 in FIG. 12.

Next, in step S145, the automatic adjustment unit 241 performs flicker occurrence determination. If it is determined that flicker noise has occurred (step S145: YES), the current value 263 is referred to, and if the exposure time is not the maximum (step S147: NO), the exposure time is increased by one step and the camera setting parameter is 21 (step S149). At that time, the current value 263 of exposure time is also rewritten with the updated value. After that, the process returns to step S103 in FIG. 12.

On the other hand, if the exposure time is the maximum (step S147: YES), the automatic adjustment unit 241 performs control to notify an error and notify that maintenance is required (step S147: YES). Step S151). After that, the process returns to step S103 in FIG. 12.

Then, if it is determined in step S145 that flicker noise has not occurred (step S145: NO), the automatic adjustment unit 241 normally ends the automatic adjustment process for the captured image acquired in step S103 of FIG. , outputs the photographed image to the image processing section 242 (step S153). Thereafter, the process returns to step S103 in FIG. 12 and the above-described process is repeated.

Return to FIG. 11. Once the automatic adjustment process is started, the automatic adjustment section 241 enters a standby state until the photographed image is outputted to the image processing section 242. Then, when the photographed image is output, that is, when the automatic adjustment section 241 outputs the photographed image to the image processing section 242 in step S151 of FIG. 13 (step S3: YES), the image processing section 242 Image processing is performed to generate a side view image (step S5).

Subsequently, the train position determination unit 243 performs train position determination processing (step S7). That is, first, the imaged object extraction unit 244 performs imaged object extraction processing, extracts the appendage imaged portion from the side view image, and extracts the imaged portion of the reference marker 17 to obtain the marker imaged position (step S71). . Next, the positional deviation pixel number calculation unit 245 performs positional deviation pixel number calculation processing and calculates the number of positional deviation pixels between the marker imaging position and the marker imaging reference position (step S73). Next, the relative distance calculation unit 246 performs relative distance calculation processing to calculate the conversion coefficient K, and uses this to calculate the relative distance ΔD (step S75).

Subsequently, the train position determination unit 243 monitors the relative distance ΔD calculated in step S75, and determines whether the train 1 has stopped (step S11). Specifically, the relative distance ΔD calculated by the relative distance calculation process a predetermined number of times in the recent past is held, and when the relative distance ΔD does not change, it is determined that the train 1 has stopped. For example, if the amount of change from the previous value remains below a predetermined threshold a predetermined number of times, it is determined that the train 1 has stopped.

If it is determined that the train 1 has stopped (step S13: YES), the train position determining unit 243 determines whether the train 1 has stopped at a fixed position using the relative distance ΔD at that time (step S13: YES). S15). That is, when the relative distance ΔD is less than or equal to the predetermined value A11 shown in FIG. 2 (within the fixed position stopping range A1), it is determined that the train 1 has stopped at the fixed position. On the other hand, if it is outside the home position stop range A1, a short circuit (the home position has not been reached) or an overrun (overrun beyond the home position) is determined. Alternatively, the distance between the home position and the stop position may be calculated from the image. In that case, the relationship between the installation position of the camera 20 and the fixed position, the relationship between the number of pixels and the distance of one pixel between the fixed position and the stop position in the captured image of the camera 20, etc. can be used. .

If it is determined that the train 1 has stopped at a fixed position (step S17: YES), a stop notification is output to the drive control unit 300 (step S19).

On the other hand, if the train 1 has not stopped at the regular position (step S17: NO), a notification control to that effect is performed (step S21), and then the process returns to step S3 to repeat the above-described process. The notification control in step S21, for example, controls the output of a warning sound and the display output of a message to notify the driver that the stop position needs to be corrected. Then, if it is determined that the train 1 has stopped again (step S13: YES), and then it is determined that the train 1 has stopped at a fixed position (step S17: YES), the process moves to step S19.

After outputting the stop notification in step S19, the maintenance necessity determination unit 247 performs maintenance necessity determination processing (step S23). For example, the maintenance necessity determination unit 247 reads the number of pixels of the past appendage imaged portion from the maintenance history data 277, sets it as the first pixel number, and in step S7, the number of pixels of the appendage imaged portion extracted last. is the second number of pixels, and the necessity of maintenance is determined based on the difference between the two. When the maintenance necessity determining unit 247 determines that maintenance is necessary (step S25: YES), it performs notification control to that effect (step S27), and ends this process.

As explained above, according to the present embodiment, distortion correction is performed on the captured image of the camera 20 that captures the stopped train 1 from the side, and the captured image after the distortion correction is subjected to perspective projection transformation to display the train 1. A side view image viewed from the side can be generated.

On the other hand, a reference marker 17 is provided on the side surface of the platform 4 of the train 1 that is within the shooting range of the camera 20 when the train 1 stops at a station, and a side view image of the train 1 when it stops at a fixed position is provided. The position of the reference marker 17 in the side view image (marker imaging reference position) and the position of the reference marker 17 in the generated side view image (marker imaging position) are used to determine the position of the train 1 relative to the fixed position. Relative distances can be determined.

Specifically, based on the conversion coefficient K between the number of pixels α of the appendage imaged part shown in the side view image and its default size, the number of pixels Δd of the positional deviation between the marker imaging reference position and the marker imaging position is calculated. , the relative distance ΔD of the train 1 with respect to the fixed position can be determined.

Therefore, as a technology related to settings for shooting such as calibration when determining the relative position of the train 1 with respect to the fixed position of the station using images taken with the camera 20, there is no danger of contact cars and it is easy to maintain. Excellent technology can be realized.

Furthermore, according to the present embodiment, the number of pixels and the conversion coefficient K of the captured object portion in the side view image when it is determined that the train 1 has stopped at a fixed position can be stored as a history. Then, when it is determined that train 1 has newly stopped at the fixed position, it is determined based on the difference between the number of pixels and conversion coefficient K of the current appendage imaged part and the number of pixels and conversion coefficient K of the past appendage imaged part. It is possible to determine whether maintenance is necessary.

Specifically, if the difference between the two is large, it is determined that maintenance is necessary. In that case, a message is displayed prompting the user to adjust the installation position and orientation of the camera 20 and to reset the imaging reference position data 273 (that is, the marker imaging reference position), indicating that maintenance is necessary. can be notified. Therefore, it is possible to automatically detect a shift in the installation position or orientation of the camera 20 due to the passage of time, etc., so maintenance can be performed appropriately when the need arises. Become.

Furthermore, even if it is necessary to change the installation position of the camera 20 apart from maintenance, the installation position and orientation of the camera 20 can be adjusted in the same way as during maintenance, and the imaging reference position data 273 can be set. By repairing it, changes in the installation of the camera 20 can be easily accommodated.

Further, according to the present embodiment, the adjustment board 23 having a single color portion can be installed, for example, in association with the camera 20, and the single color portion can be photographed with the single color portion within the photographing range. Then, based on the brightness value of the single-color imaged portion in the photographed image, parameter values related to each camera setting parameter of analog gain, digital gain, and exposure time can be automatically adjusted. Therefore, maintenance related to these camera setting parameters can be made unnecessary.

For example, the adjustment board 23 is attached to the camera 20 so that a single color portion appears in a band shape over the entire lower end of the photographed image. Then, if the luminance dispersion index value obtained for the single-color imaged portion satisfies the dispersion under-dispersion threshold condition, it is determined that the dynamic range has been exceeded, and in that case, the analog gain is automatically adjusted or the exposure time is automatically adjusted. be able to.

Furthermore, if the variation index value obtained for the single-color imaging portion satisfies the excessive variation threshold condition, it is determined that flicker noise has occurred, and in that case, the exposure time can be automatically adjusted. Furthermore, it is possible to determine whether or not the center brightness of the single-color imaging portion is outside the allowable range, and to automatically adjust the digital gain if it is outside the allowable range.

Therefore, it is possible to realize a technology related to settings for photographing including processing such as brightness adjustment of a photographed image, which is free from the danger of touch wheels and has excellent maintainability.

In the above embodiment, the conversion coefficient K is calculated using car body accessories such as the window 13 and the vehicle door 15 provided on the train 1. If the size is within the range and the size is known, it can be used to calculate the conversion coefficient K. For example, the platform door 31 or door pocket 33 (see FIG. 4, etc.) of the platform fence 3 may be used. In that case, the specified size of the platform door 31 and door pocket 33 may be set in advance as specified size data. Then, the imaged portions of the platform door 31 and the door pocket 33 are extracted from the side view image, and the conversion coefficient K is calculated in the same manner as in the above embodiment.

Further, in the above embodiment, the reference marker 17 provided on the train 1 is used as a predetermined part of the train 1, and is used to determine whether the train 1 is stopped at a fixed position. For example, an antenna, a passenger door, etc.) can also be used as a predetermined part. In that case, a specific object to be used as a predetermined region is extracted in advance from a side view image when the train 1 is stopped at a fixed position, and the position in the side view image is set as an imaging reference position. Then, the relative position of the train 1 with respect to the fixed position is determined by comparing the position (imaging position) of the specific object in the side view image with the imaging reference position.

1 train, 11 lead vehicle, 13 window, 15 vehicle door, 17 reference marker, 2 train position determination device, 20 camera, 21 camera setting parameter, 23 adjustment board, 200 processing device, 240 control unit, 241 automatic adjustment unit, 242 image processing section, 243 train position determination section, 244 imaged object extraction section, 245 positional deviation pixel number calculation section, 246 relative distance calculation section, 247 maintenance necessity determination section, 250 storage section, 251 automatic adjustment program, 253 stop position Judgment program, 255 Approach side photographed image data, 260 Parameter setting data, 261 Initial value, 263 Current value, 271 Tolerance range data, 273 Imaging reference position data, 275 Regulation size data, 277 Maintenance history data, 3 Platform fence, 31 platform door, 33 door pocket, 300 drive control section, 4 platform, 5 side view image, 6 photographed image

Claims (7)

Image processing is performed to convert images taken by a camera installed at a station to a side-view image of the train, including distortion correction, to take a side view of a train stopped at a fixed position at the station. image processing means;
The position where a predetermined part of the train appears in the side view image when the train stopped at the fixed position is photographed (hereinafter referred to as "imaging reference position"), and the side view image processed by the image processing means. determining means for determining the relative position of the train with respect to the fixed position by comparing the position of the predetermined part shown in the image in the side view image (hereinafter referred to as "imaging position");
repeatedly performing image processing by the image processing means and determination of the relative position by the determination means,
In the repeated execution , the determination means:
identifying an imaged portion of a vehicle body appendage of a predetermined size provided at a predetermined position of the train, which is shown in the side view image;
calculating a conversion coefficient for converting the number of pixels of the identified imaging portion and the predetermined size;
Calculating the relative distance of the train with respect to the fixed position from the number of positional deviation pixels between the imaging reference position and the imaging position based on the conversion coefficient ;
and determining whether the train has stopped at the fixed position using the relative distance at stop, which is the relative distance when the relative distance stops changing;
Train position determination device. The vehicle body accessory is a window or a door of the train,
The train position determining device according to claim 1. storage means for storing the imaging reference position;
a first number of pixels for the identified imaging portion when the relative distance for the first train stops changing; and a first number of pixels for the identified imaging portion when the relative distance for the second train stops changing; Based on the difference from the second number of pixels, it is determined whether or not maintenance is necessary for correcting the data of the imaging reference position stored in the storage means or inspecting the installation position and orientation of the camera. means for determining whether maintenance is necessary;
The train position determination device according to claim 1 or 2, further comprising: Image processing is performed to convert images taken by a camera installed at a station to a side-view image of the train, including distortion correction, to take a side view of a train stopped at a fixed position at the station. image processing means;
The position where a predetermined part of the train appears in the side view image when the train stopped at the fixed position is photographed (hereinafter referred to as "imaging reference position"), and the side view image processed by the image processing means. determining means for determining the relative position of the train with respect to the fixed position by comparing the position of the predetermined part shown in the image in the side view image (hereinafter referred to as "imaging position");
Equipped with
The determining means is
identification means for identifying an imaged portion of a vehicle body appendage of a predetermined size provided at a predetermined position of the train, which is shown in the side view image processed by the image processing means;
Based on a conversion coefficient between the number of pixels of the imaging portion specified by the specifying means and the predetermined size, the relative distance of the train to the fixed position is calculated from the number of pixels misaligned between the imaging reference position and the imaging position. relative distance calculation means for
has
A predetermined object whose part or all is a single color (hereinafter, a single color part is referred to as a "single color part") is installed at the station, or is installed accompanying the camera,
The camera performs photography based on predetermined camera setting parameters;
The relative installation position and installation direction of the predetermined object and the camera are determined so that the single color portion is captured in an image taken by the camera,
A variation index value indicating the variation in luminance value is calculated for the single color portion in the photographed image, and if the variation index value satisfies a predetermined variation threshold condition indicating that the variation is small, dynamic detection is performed. camera setting automatic adjustment means that determines that the range is over and automatically adjusts the parameter value of the camera setting parameter related to gain or exposure time;
A train position determination device further comprising : The camera setting automatic adjustment means automatically adjusts a parameter value related to exposure time when a predetermined flicker noise generation condition based on the photographed image is satisfied.
The train position determination device according to claim 4 . Image processing is performed to convert images taken by a camera installed at a station to a side-view image of the train, including distortion correction, to take a side view of a train stopped at a fixed position at the station. an image processing step;
When a train stopped at the fixed position is photographed, the position where a predetermined part of the train appears in the side view image (hereinafter referred to as "imaging reference position"), and the position where the predetermined part of the train appears in the image processed side view image are determined. a determination step of determining the relative position of the train with respect to the predetermined position by comparing the position of the predetermined part in the side view image (hereinafter referred to as "imaging position");
The image processing step and the determination step are repeatedly executed,
The determination step includes, in the repeated execution,
identifying an imaged portion of a vehicle body appendage of a predetermined size provided at a predetermined position of the train, which is shown in the side view image;
calculating a conversion coefficient for converting the number of pixels of the identified imaging portion and the predetermined size;
Calculating the relative distance of the train with respect to the fixed position from the number of positional deviation pixels between the imaging reference position and the imaging position based on the conversion coefficient ;
and determining whether the train has stopped at the fixed position using a relative distance at stop, which is a relative distance when the relative distance stops changing.
Train position determination method. Image processing is performed to convert images taken by a camera installed at a station to a side-view image of the train, including distortion correction, to take a side view of a train stopped at a fixed position at the station. an image processing step;
When a train stopped at the fixed position is photographed, the position where a predetermined part of the train appears in the side view image (hereinafter referred to as "imaging reference position"), and the position where the predetermined part of the train appears in the image processed side view image are determined. a determination step of determining the relative position of the train with respect to the predetermined position by comparing the position of the predetermined part in the side view image (hereinafter referred to as "imaging position");
including;
The determination step includes:
identifying an imaged portion of a vehicle body appendage of a predetermined size provided at a predetermined position of the train, which is shown in the image-processed side view image;
Calculating the relative distance of the train to the fixed position from the number of pixels misaligned between the imaging reference position and the imaging position based on a conversion coefficient between the number of pixels of the identified imaging portion and the predetermined size; ,
including ;
A predetermined object whose part or all is a single color (hereinafter, a single color part is referred to as a "single color part") is installed at the station, or is installed accompanying the camera,
The camera performs photography based on predetermined camera setting parameters;
The relative installation position and installation direction of the predetermined object and the camera are determined so that the single color portion is captured in an image taken by the camera,
A variation index value indicating the variation in luminance value is calculated for the single color portion in the photographed image, and if the variation index value satisfies a predetermined variation threshold condition indicating that the variation is small, dynamic detection is performed. determining that the range is over and automatically adjusting parameter values of the camera setting parameters related to gain or exposure time;
A train position determination method further comprising :

Images