JP2014013256A - Track displacement measurement apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a track displacement measurement apparatus capable of reliably detecting track anomaly with an inexpensive configuration.SOLUTION: A track displacement measurement apparatus includes: a plurality of photographing targets A1-A4, B1-B4 located on or proximate to a railroad track 32; a digital camera 11 which captures an image that includes all of the plurality of targets; and a control unit which controls the timing for the digital camera 11 to acquire images, stores acquired images, computes coordinates of each target using each of the acquired images, and stores computation results. The control unit has a function to compute coordinates of the targets from acquired images and measure displacement of the track from temporal change of the coordinates since start of the operation obtained from each result of the coordinate computation on every acquired image and, in addition, has a self diagnosis function to diagnose presence of target anomaly and lens anomaly of the camera.

Description

  The present invention belongs to a track displacement measuring apparatus for measuring a track displacement such as a railroad track.

  When underpass construction or shield construction is underway on a railroad track, it is necessary to constantly monitor the displacement of the track, because if the track is displaced due to such construction, it will lead to accidents such as train derailment. There is. In order to measure these subsidences and uplifts in order to safely proceed with these constructions, monitoring devices such as displacement sensors and subsidometers are installed on the ground near the track for monitoring (for example, Patent Document 1). ). When the ground displacement exceeds the safety limit, measures such as emergency stop of the train are taken.

JP 11-247108 A

  However, in the method of monitoring the displacement of the track using the above-mentioned squat meter, it is not possible to directly check the displacement of the track, so if the first alarm occurs, the staff will go to the site and directly measure the track. Need to do. However, it is difficult to secure personnel at night, and problems remain with respect to train safety management and instantaneous response.

  In order to solve these problems, the Kevlar fiber laid parallel to the track is fixed at two points, and the two-dimensional track displacement between the Kevlar fiber and the track is detected by image processing at the intermediate point. A method for measuring track displacement has been proposed based on a method and a result of measuring the position of each target with a measuring device called a total station by placing a reflective target on a railway track.

  However, in the case of the former method, when imaging is performed from two directions because the Kevlar fiber is displaced two-dimensionally, the data obtained from the captured image includes a distortion component, and this distortion component is corrected. Is needed. Further, in order to measure one measuring point, it is necessary to lay Kevlar fiber, for example, in a 10-m section centered on the measuring point, which not only complicates the apparatus configuration but also requires protection of Kevlar fiber. .

  In the latter method, since it is impossible to measure all targets simultaneously, there is a problem that the influence of time or weather change cannot be said to be the same in the measurement results of all targets.

  Accordingly, an object of the present invention is to provide an orbit displacement measuring apparatus that can reliably detect an orbit abnormality with an inexpensive configuration.

  According to the present invention, an initial image is obtained by photographing a plurality of shooting targets arranged on or near a railway track and an image that is arranged at a predetermined position and includes the plurality of targets at the time of initial setting. And a camera for capturing an image including the plurality of targets at the time of actual measurement to obtain a measurement image, and comparing the initial image and the measurement image on coordinate axes on a common image, and the measurement image The target is extracted from the target, the coordinates of the specific part for each target are calculated, and the displacement of the trajectory is calculated from the time-dependent change of the coordinates of the specific part from the start of operation based on the calculation result of the coordinates of the specific part of each target for each imaging. A control unit for measuring, and at the time of initial setting, the control unit targets a reference image including a target set in the initial image. Each time, the coordinates of a specific part of each target in each reference image are calculated and stored in the storage device, and further, a search range set to center the reference image for each target. The control unit stores the search range for each target in the measurement image defined by the coordinate axes common to the initial image and including the plurality of targets in actual measurement. The measurement search range is set in the same region as the region where the reference value is set, and the reference image of the corresponding target is moved pixel by pixel within the measurement search range, so that the correlation coefficient with the reference image is the highest. A high area is obtained as an area including the corresponding target, the coordinates of a specific part of the area are calculated, and the target is photographed under the control of the control unit. Each target is equipped with a light emitting circuit that automatically emits a light source in time, each target has a plurality of LEDs arranged in a ring shape, and the light emission pattern is formed in a ring shape, and the control unit performs the actual measurement. When the correlation coefficient is low, the plurality of LEDs are individually turned off sequentially from all lighting states, and each time shooting is performed.From the obtained captured image, the cause of the low correlation coefficient is a target abnormality, A trajectory displacement measuring device is provided that performs self-diagnosis of any abnormality of a camera lens.

  The trajectory displacement measuring apparatus according to the present invention captures a plurality of targets with a camera so as to fit into one image, and performs coordinate calculation of each target based on the instantaneously obtained image. The behavior does not include a temporal element.

  Further, since image data is used without using Kevlar fibers, analog error factors do not intervene in coordinate calculation, and it is not necessary to consider protection of Kevlar fibers at all.

  According to the present invention, by using an image processing means such as a target and a camera and a personal computer configured only with inexpensive electronic components, it is possible to automatically perform orbital displacement measurement instantaneously, and in relative change There is no influence of errors caused by the environment. In addition, the data stored and processed by the personal computer is digital data, and since the stored data is a digital value, it can be easily applied to numerical operations such as statistical processing.

It is a figure which shows roughly the structure of the orbital displacement measuring apparatus by this invention. It is a figure which shows the flow of the coordinate calculation by this invention. It is a figure which shows the example of a setting of the reference | standard image and search range by this invention. It is a figure which shows the example of a search of the measurement image by this invention. It is a figure explaining the condition which measures orbital displacement by the present invention. It is a figure which shows an example of the arrangement | positioning condition of a digital camera. It is a figure explaining the relationship between the actual displacement by this invention, and an apparent displacement. It is a figure which shows an example of the target which comprised the light emitting circuit and the light source. It is a figure which shows the boundary of the brightness of the target in the structure of FIG. It is a figure which shows an example of the control circuit of the target which makes a light emission pattern variable. It is a figure which shows the target which made the light emission pattern by this invention ring shape.

  Next, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a diagram schematically showing the configuration of an orbital displacement measuring apparatus according to the present invention.

  The control unit 10 is configured by an electronic circuit having a storage device capable of mounting a program capable of displaying an image on a display and storing data and controlling an externally connected device, such as an industrial board personal computer.

  The digital camera 11 controls the shooting time, the number of shots, the shutter speed, and the like according to a control signal from the control unit 10 and simultaneously shoots a plurality of targets 12 as shooting targets. A digital image taken by the digital camera 11 is transferred to the control unit 10.

  As will be described in detail later, the target 12 in this embodiment is circular and has a light emission function, and the light emission timing is controlled by the control unit 10.

  The control unit 10 stores the transferred image in the storage device, calculates a specific part of the target 12, for example, a two-dimensional coordinate of the center of gravity, from the image, and stores the calculation result in the storage device.

  FIG. 2 is a diagram showing a flow of coordinate calculation performed by the control unit 10.

1. Initial setting The image obtained by simultaneously shooting multiple targets at the time of installation is set as the initial image, and the target image is cut out and numbered for each target from the initial image and set as the “reference image”. Is calculated and stored in the storage device (step S1).

  Further, the distances L1 to Ln (n is the number of targets 12 installed) from the digital camera 11 to each target 12 are actually measured and stored in the storage device as initial values for calculation.

  With reference to FIG. 3, a method for cutting out a reference image for each target from an initial image obtained at the time of initial setting will be described. The reference image is cut out by displaying an initial image on the display and then operating by a keyboard operation by an operator. At this time, a certain range of regions centered on the reference image (hereinafter referred to as a search range) is obtained. Is set. The reference image is usually a quadrangle, and the size differs for each target. This is because the distances L1 to Ln from the digital camera to the plurality of targets are different as will be described later. The search range is also a quadrangle, and the size is different for each reference image.

  FIG. 3 shows an example of setting a plurality of reference images (target images) in the initial image and a search range for each reference image. The operator displays an initial image taken by the digital camera on the display, and sequentially cuts out an area where each target exists as a reference image. For example, when the operator designates one of the two diagonals of the square reference image (P1 and P2 in FIG. 3) with a touch pen, a cursor, or the like, it is within the square range defined by P1 and P2. This is performed by extracting the image of the above as a reference image. The operator also sets, as a search range, diagonal coordinates of a region having an area that is at least twice as large as the reference image with the reference image as the center as a search range when performing template matching between a measurement image and a reference image, which will be described later. The area ratio of the search range to the reference image becomes larger as the required measurement range is wider. This setting is also performed by designating one of the two diagonals of the quadrangular search range (P3 and P4 in FIG. 3), as in the case of cutting out the reference image. The search range is set for the purpose of limiting the area for comparison with the reference image for each target. By arranging the targets so that the search ranges of all targets do not overlap in the initial image, each target is set. The position of is never misidentified.

  In FIG. 3, the number of targets is 6, and numbers 1 to 6 are assigned in descending order of size. Hereinafter, the largest reference image may be referred to as reference image 1, and the search range may be referred to as search range 1.

  As described above, in the case of FIG. 3, in the initial setting, with respect to the target number 1, in addition to the reference image 1, the coordinates of P3 and P4 (the coordinates of the search range 1 in the initial image), the reference image 1 The barycentric coordinates of the target and the distance from the digital camera are stored. Such data is stored for six targets. The center-of-gravity coordinates of the target can be calculated by a template matching method to be described later, but the midpoint of the line segment connecting P3 and P4 may be calculated as the center-of-gravity coordinates. This is because in the reference image, the center of the reference image and the center of gravity of the target always coincide with each other, and also coincide with the center of the search range. In this way, the center of gravity coordinates of the target can be obtained with an accuracy of 1 pixel or less by obtaining the center of gravity coordinates of the target.

2. Measurement (imaging) time determination The control unit 10 stands by until the measurement time set by the operator (step S2).

3. The photographing control unit 10 causes the plurality of targets 12 to emit light at the measurement time, and photographs them with the digital camera 11 (step S3). Of course, this shooting is performed in the same state as the shooting of the initial image.

4). Image Storage The control unit 10 stores the captured image as a captured image in a storage device in the control unit 10 (step S4).

5. Target Extraction FIG. 4 is a diagram showing a search example of a measurement image according to the present invention, and corresponds to FIG. Since the initial image of FIG. 3 and the captured image of FIG. 4 are images captured in the same state, it goes without saying that they are represented by a common coordinate axis. First, as shown on the left side of FIG. 4, the control unit 10 is the same as P3 and P4 that define the search range 1 set as the target of number 1 in the initial image of FIG. Set the search range to the coordinates. FIG. 4 shows that the target is shifted diagonally to the lower right in the set search range 1. Subsequently, as shown on the right side of FIG. 4, the control unit 10 moves the reference image 1 pixel by pixel from one diagonal to the other within the search range 1 and correlates with the reference image 1 by template matching. Get the highest region. By obtaining the center-of-gravity coordinates of the target in this region, the center-of-gravity coordinates of the target number 1 can be obtained with an accuracy of 1 pixel or less. Such processing is also performed for the search ranges 2 to 6 set in FIG.

  Template matching is used as a basic operation in the field of image processing. In this embodiment, in order to extract the position (coordinates) of the measurement image including the number 1 target from the search range 1 in the captured image obtained in step S3, the control unit 10 uses the correlation coefficient of the reference image 1 with respect to the measurement image. While obtaining R, the reference image 1 is shifted by one pixel at a time, and the coordinate having the highest correlation coefficient R is obtained. By this processing method, the coordinates of the measurement image in the captured image, that is, the coordinates indicating the target area can be obtained in units of one pixel (step S5). Here, the coordinates of the measurement image are represented by the diagonal coordinates (in FIG. 3, the coordinates of P1 and P2).

  The correlation coefficient is a value representing the similarity between the two-dimensional luminance distributions of the reference image and the measurement image, and becomes 1 when the distributions completely match. The correlation coefficient R when the vertical and horizontal sizes of the reference image are x and y is obtained by the following equation (1).

  Where f (i, j) is the luminance of the coordinate position (i, j) in the reference image, s (i, j) is the luminance and sign of the coordinate position (i, j) in the region overlapping the reference image in the measurement image Symbols with a bar above f and a symbol s are the average values of the luminances of the portions of the reference image and the measurement image that overlap the reference image.

6). Coordinate calculation The coordinates obtained in the process of step S5 are in units of one pixel. However, by calculating the center of gravity of a circular target, it is possible to calculate the center of gravity coordinates of the target in units of one pixel or less.

  First, the control unit 10 performs binarization processing on the measurement image defined by the coordinates obtained in step S5, thereby converting bright pixels to white and dark pixels to black. Thereafter, the weighted average of the X and Y coordinates of the white pixels is performed to obtain the coordinates of the center of gravity of the circular area in the photographed image (step S6).

7). Conversion to displacement The control unit 10 obtains a difference between the center of gravity coordinates obtained by the process of step S6 and the center of gravity coordinates indicated by the initial value, and multiplies the difference by a proportional coefficient corresponding to the distance L, thereby obtaining an initial value. Relative displacement from the barycentric coordinates shown is obtained (step S7). The proportionality coefficient according to the distance L is a value set according to the distances L1 to Ln in order to obtain an actual displacement amount because the displacement of the center of gravity coordinates on the image is only the displacement amount within the image. It is.

8). Data storage The relative displacement amount (actual displacement amount) from the center-of-gravity coordinates and the center-of-gravity coordinates indicated by the initial values obtained at steps S6 and S7 is stored (step S8).

  FIG. 5 is a diagram showing a situation in which orbital displacement is measured according to the present invention. A track (rail) 32 is installed on a sleeper 31 embedded in the road bed on the roadbed with a space. A plurality of targets 12 (A 1, A 2, A 3, A 4, B 1, B 2, B 3, B 4) are fixed to the sleepers 31 on both sides of the track 32 using a fixing jig.

  FIG. 6 is a diagram showing an arrangement state of the digital camera 11 according to the present invention.

  In order to accommodate all of the plurality of targets 12 (here, the targets A1 to A4 on one side) within the angle of view, the digital camera 11 is arranged at a position away from the target installation section.

By the way, each railway company uses the following four track measurement items for track management.
Gauge strain: The distance between the left and right rails is deviated, and represents the amount of increase or decrease relative to the basic gauge.
Level strain: Expresses the difference in height between the left and right rail heads.
Street strain: Represents unevenness in the length direction of the rail side surface. Generally, it is expressed as a horizontal distance between the rail and the yarn in the center portion by stretching a 10 m yarn on the rail side surface.
High and low strain: Represents unevenness in the length direction of the rail head surface. Generally, it is expressed by a vertical distance between the rail and the thread at the center portion by applying a 10 m thread on the rail head surface.

  Since each of the above items can be obtained by measuring the displacement in the direction orthogonal to the trajectory, a digital camera is placed on or near the trajectory.

  By calculating a straight line (10 m chord) based on the coordinates of three targets respectively placed 5 m before and after the target to be obtained, and obtaining the horizontal and vertical distances from the target to the straight line, the above 4 Strain and high / low strain can be obtained.

  FIG. 7 is a diagram showing the relationship between the actual displacement of the target and the apparent displacement obtained from the captured image.

  Since the digital camera cannot be placed in the trajectory, the optical axis of the digital camera must be tilted with respect to the trajectory direction with respect to the vertical direction and the horizontal direction in order to fit all targets in the same image. That is, the position of the digital camera 11 is shifted in a direction away from the track 32 and is positioned higher than the track 32.

  Considering the spatial coordinates with the installation position of the digital camera 11 as the origin, the horizontal direction orthogonal to the track 32 is the X axis, the direction parallel to the track 32 is the Y axis, and the height direction is the Z axis. It is assumed that the optical axis of the digital camera 11 forms an angle θx with respect to the Y axis on a horizontal plane, and forms an angle θz with respect to the X axis on a vertical plane. In this case, the apparent displacement ΔvX, ΔvZ of a target 12 obtained from the captured image of the digital camera 11 and the actual displacement ΔX (displacement in the X-axis direction), ΔZ (displacement in the Z-axis direction) are as described above. When the spatial coordinates of the target 12 are (X0, Y0, Z0), the following relationship is established.

ΔX = ΔvX × (1 / cosθx)
= ΔvX × √ (X0 ² + Y0 ² ) / Y0 (approximate because ΔX << X0)
ΔZ = ΔvZ × (1 / cosθz)
= ΔvZ × √ (Y0 ² + Z0 ² ) / Y0 (approximate because ΔZ << Z0)
If the installation position of the digital camera is located far away from the target, Y0 >> X0, Y0 >> Z0 holds,
ΔX = ΔvX
ΔZ = ΔvZ
Holds.

  In other words, it is possible to obtain a displacement measurement result equivalent to the actual displacement by performing the calculation without correcting the coordinates of each target.

  The displacement measurement as described above is also performed on the targets B1 to B4. The control unit 10 determines whether or not the displacement amount from the initial value is within a preset range for all targets, and if there is a target having a displacement amount exceeding the set range, notifies that fact. To do.

  FIG. 8 is a diagram showing a configuration of a target including a light emitting circuit and a light source according to the present invention.

  In the coordinate calculation by image processing, the higher the contrast between the luminance of the target and the surrounding luminance, the clearer the boundary between light and dark and the smaller the error in binarization, so that the accuracy of specifying the coordinates is increased.

  In this example, the target is made to emit light in a circular shape. For this reason, a plurality of LEDs 12-1 having low power consumption and long life are arranged in a ring shape, and the emitted light is transmitted through the diffusion plate 12-2. In this configuration, the brightness of transmitted light is made uniform.

  The light that has passed through the diffuser 12-2 is limited to a circular optical path by hollowing out the center of the presser plate 12-3, and a bordered circle of light and darkness is formed by installing a frame-shaped shading mask 12-4. For clarity, the light emitting part of the target is circular and has uniform brightness as seen from the digital camera, and the brightness contrast with the surroundings is high.

  Instead of emitting light from the target itself, there is also a method to increase the contrast between the target and its surroundings by applying a reflective coating to the target and receiving the reflected light using a camera flash. In order to take pictures, the flash needs a large amount of light. In addition, since it is necessary to shoot the flash light over the entire shooting range, the living environment of neighboring residents is disturbed.

  When the target itself emits light as in this example, it is possible to minimize light leakage to the neighboring environment by masking the direction other than the direction of the digital camera, and consider scattering of light due to reflected light. Since it is not necessary to do so, it is efficient in terms of power.

  FIG. 9 is a diagram showing a light / dark boundary of the target having the configuration of FIG.

  When the horizontal or vertical cross section of one target is viewed, there are two bright and dark boundaries, and the light emitting part is circular, so the midpoint of the line segment connecting these two boundaries is the horizontal or vertical target. It becomes the coordinates of the center of gravity.

  FIG. 10 is a diagram showing a control circuit for a plurality of LEDs for making the light emission pattern of the target variable.

  Here, power is supplied to the ten LEDs 1 to LED 10 from the DC power source 30, and a FET 41 and a resistor 42 are connected as a switching element between the DC power source 30 and each LED 1 to LED 10, and control from the control unit 10 is performed. By controlling on / off of the FET 41 with a signal, light emission only at the time of measurement is possible.

  When the correlation coefficient in step S5 is low, self-diagnosis is performed in which the FET 41 is individually controlled to turn off the LEDs 1 to 10 in order and take an image. For example, if dirt is attached to the camera lens, the focal length is far away, so it blurs and affects a wide range on the image, but if the LED is defective individually, the defective LED is photographed darkly. Therefore, it is possible to identify whether it is due to a target abnormality or a camera lens abnormality.

  Furthermore, by turning off the LEDs in order, if the target is abnormal, if the surface of the target is dirty, the brightness of the captured image will differ between when it is turned on and when it is turned off. Since there is no difference in the brightness of the captured image, it is possible to specify a detailed failure location and determine whether the target itself can be repaired or the target surface can be cleaned.

  Since it is necessary to enter the track to repair and replace the target, it is necessary to apply to the railway company in advance and perform the work in the presence of railway company personnel when there is no railway operation. By grasping in advance whether it is replacement or cleaning, the contents of entry work into the track become clear. If the camera lens is abnormal, if the camera is installed outside the management site, it can be repaired at an arbitrary time and can be quickly recovered.

  FIG. 11 is a diagram showing an example of a target having a light emission pattern according to the present invention in a ring shape.

  The target according to this example is a presser plate in which a circular plate is attached as a light shielding mask 12-6 at the center of a presser plate similar to the presser plate 12-3 instead of the presser plate 12-3 of the target described in FIG. 12-5 is used.

  By using such a holding plate 12-5 to form a ring-shaped light emission pattern, the cross section of the outer circumference circle (the outer peripheral edge of the light transmission portion) and the cross section of the inner circumference circle (the inner periphery of the light transmission portion) when calculating the coordinates of the target It is possible to use a total of four light / dark boundaries formed in the above. By calculating the center-of-gravity coordinates obtained from the two light / dark boundaries formed by the cross-section of the outer circle and the center-of-gravity coordinates determined from the two light / dark boundaries formed by the cross-section of the inner circle, Specific accuracy is improved.

  According to the orbital displacement measuring apparatus described above, it is possible to automatically perform orbital displacement measurement instantaneously by using a target composed of only inexpensive electronic components and a digital camera. Not only is there no influence of the error caused by it, but since the stored data is a digital value, it can be easily applied to numerical operations such as statistical processing.

  In addition, the self-diagnosis function of the apparatus has a great effect that it can detect an apparatus abnormality.

  Note that the imaging means in the present invention is not limited to a digital camera but may be any camera that can obtain an image signal. When the image signal is an analog signal, an analog-digital converter is provided in front of the control unit 10.

10 control unit 11 digital camera 12 target 31 sleeper 32 orbit

Claims (2)

A plurality of shooting targets arranged on or near the railway track; and
An initial image is obtained by capturing an image including the plurality of targets arranged at a predetermined position at the time of initial setting, and a measurement image is obtained by capturing an image including the plurality of targets at the time of actual measurement. Camera for
The initial image and the measurement image are compared on the coordinate axes on the common image, the target is extracted from the measurement image, the coordinates of the specific part for each target are calculated, and the coordinates of the specific part of each target for each imaging A control unit that measures the displacement of the trajectory from the time-dependent change of the coordinates of the specific part from the start of operation based on the calculation result of
In the initial setting, the control unit stores a reference image including the target set in the initial image in a storage device for each target, and calculates coordinates of a specific part of each target in each reference image. Storing in the storage device, further storing in the storage device a search range set to center the reference image for each target,
In the actual measurement, the control unit is provided in the same area as the area where the search range is set for each target in the measurement image defined by the coordinate axes common to the initial image and including the plurality of targets. After setting the measurement search range, the reference image of the corresponding target is moved pixel by pixel within the measurement search range, and the region having the highest correlation coefficient with the reference image includes the corresponding target. Obtain as a region and calculate the coordinates of a specific part of that region,
Under the control of the control unit, the target comprises a light emitting circuit that automatically emits a light source in accordance with the photographing time,
Each target has a plurality of LEDs arranged in a ring shape and its light emission pattern is in a ring shape.
When the correlation coefficient is low during the actual measurement, the control unit turns off the plurality of LEDs individually in order from the fully lit state, and takes a picture each time. An orbital displacement measuring apparatus that performs self-diagnosis to determine whether the cause of a low relationship number is a target abnormality or a camera lens abnormality.   The track displacement measuring apparatus according to claim 1, wherein the plurality of targets include a predetermined number of targets arranged at equal intervals along the railroad track, and the control unit includes: Select a target target and two targets adjacent to it, calculate the coordinates of specific parts of a total of three targets, calculate a straight line based on the coordinates of the specific parts of the two targets, An orbital displacement measuring apparatus for measuring a horizontal distance and a vertical distance from a target target to the calculated straight line.

Images