JP4858316B2 - Trolley wire wear measuring device by image processing - Google Patents

Description

  The present invention relates to a trolley wire wear measuring apparatus using image processing. In particular, the present invention relates to an apparatus for measuring the width of a worn portion of a trolley wire.

The trolley line that supplies power to the electric railway vehicle comes into contact with the current collector every time the vehicle passes. For this reason, the trolley wire gradually wears while the electric railway vehicle is operated, and if it is not replaced, it eventually breaks and causes an accident.
Therefore, the trolley wire has a wear limit, and the trolley wire is replaced based on the wear limit to ensure the safety of the electric railway vehicle.

There are two methods for measuring the wear of the trolley wire: a method of directly measuring the thickness of the trolley wire, and a method of calculating the width of the wear portion of the trolley wire and converting the thickness of the trolley wire from the wear width. .
As a method of directly measuring the thickness of the trolley wire, first, there is a method of measuring the thickness of the trolley wire using a ruler such as a caliper. This is a method in which the thickness of the trolley wire portion that the operator wants to measure is manually measured using a ruler such as a vernier caliper, and the thickness of the trolley wire to be measured can be reliably obtained. On the other hand, the measurement is laborious and cannot be automated, so it is difficult to measure a long distance section.

Another method for directly measuring the thickness of the trolley wire is to use an optical sensor. This is done by pressing the rotating roller against the trolley wire, and measuring the amount of light received at the trolley wire portion sandwiched between the devices with a laser irradiation device and a light receiving device attached so that the trolley wire is sandwiched between the roller stands. The thickness of the wire is converted. Although this method can continuously measure the thickness of the trolley wire, it is necessary to operate at low speed because it involves contact with the trolley wire. In addition, it is impossible to use the unit where there is a structure that collides with the sensor such as air section and anchor, and it is necessary to move the device away from the measurement position so that it does not collide with the existing structure. .
As a method of measuring the width of the trolley wire wear portion, there is a method of measuring the trolley wire wear portion by irradiating a sodium lamp or laser light (for example, see Patent Document 1). This uses the fact that the lower part of the trolley wire has a round gourd shape, and the width of the cracked part becomes wider as the trolley line is flattened due to wear, and the thickness of the trolley line at that point is determined from the wear width. Convert.

  To measure the trolley wire wear part, the position of the light receiving part with the line sensor is precisely adjusted so that the reflected light from the trolley wire wear part is received by regular reflection when irradiated from a light source such as a sodium lamp or laser light. The trolley wire wear part is made into a whiteout state by imaging strong light due to regular reflection, and the width of the trolley wire wear part is obtained from the width of the whiteout part receiving the strong light. This method is non-contact and can be operated at high speed. However, it is easily affected by noise such as clamps sandwiching the trolley wire and structures reflected in the background, and there is no way to check the wrong measurement result due to some noise, which causes problems as trolley wire wear. In the end, it is confirmed by using a method of directly measuring the thickness of the trolley wire. In addition, it is necessary to precisely adjust the positions of the light source and the light receiving device so as to receive specularly reflected light.

In response to this, a technique has been developed in which the wear width is measured by image processing from a line sensor image, and the thickness of the trolley wire at that location is converted from the measured value of the wear width ("Trolley wire wear measurement by image processing"). Device (Japanese Patent Application No. 2006-273523) ", hereinafter referred to as a prior proposal 1). According to the previous proposal 1, it is possible to measure a long distance in a short time without contact.
As shown in FIG. 13, the basic idea of the image processing in the prior proposal 1 is that a line sensor 10 is used as an image input means, and a normal lighting fixture 2 is used for illumination instead of a sodium lamp or laser light. It is what you use.

That is, the line sensor 10 is installed on the roof of the inspection vehicle 3 so as to look up vertically, the scanning line of the line sensor 10 is perpendicular to the traveling direction of the inspection vehicle 3, and the scanning line crosses the trolley line 4. Like that. On the roof of the inspection vehicle 3, a pantograph 5 that comes into contact with the trolley wire 4 is provided, and the vicinity of the trolley wire 4 that comes in contact with the pantograph 5 is illuminated by the lighting device 2 and the state of the trolley wire 4 is lined up. Photographed by the sensor 10.
An image signal of the trolley wire 4 photographed by the line sensor 10 is input to a measurement personal computer (hereinafter referred to as PC) 6 installed in the vehicle. The measurement PC 6 arranges the luminance signals of the scanning lines obtained from the line sensor 1 in time series, creates a line sensor image (planar image), and stores it in the recording device 7 as an input image.

As the measurement PC 6, for example, as shown in FIG. 17, a line sensor image creation unit 20, memories 30 and 40, a discriminant analysis binarization processing unit 50, a noise removal processing unit 60, and a trolley wire wear unit edge detection unit 70. The trolley wire wear portion width calculation unit 80 is configured.
The measuring PC 6 performs image processing on the input line sensor image according to the flowchart shown in FIG. 16 to obtain the width of the worn portion of the trolley wire according to the following procedure.

[Acquisition of line sensor image]
First, the image signals acquired by the line sensor 10 are arranged in time series by the line sensor image creating unit 20 and stored in the memory 30 as a line sensor image (step S1).

[Enhancement of trolley wire wear by discriminant analysis binarization]
Next, the line sensor image is sent to the discriminant analysis binarization processing unit 50 together with the parameters via the memory 40, and binarization processing is performed by the discriminant analysis binarization processing unit 50 (step S2).
Here, the worn portion of the trolley wire is a portion where the trolley wire is scraped by a pantograph, and therefore, the trolley wire has a stronger gloss than the portion that is not worn. Therefore, the worn portion of the trolley wire is also the background portion on the line sensor image. Is taken as a band-like portion having a different luminance value compared to

The discriminant analysis binarization processing unit 50 sets a threshold value so as to separate a trolley wire wear portion (pantograph contact surface) and other background portions (existing structures, etc.), and the threshold value is used for the line sensor image. A binarization process is performed to construct a binarized line sensor image in which a worn portion of the trolley line is emphasized.
As shown in FIG. 14, the binarized line sensor image has a trolley wire wear portion in a white area and a background portion in a black area.
Further, the discriminant analysis binarization processing unit 50 uses a discriminant analysis binarization method in order to cope with the difference in the displacement of the trolley line and the intensity of the reflected light from the trolley line.

The “discriminant analysis binarization method” is a method for determining a threshold according to an image, and there is a block of pixels (hereinafter referred to as a class) that gathers in a certain range of luminance values in a histogram in each image. A relatively good threshold value is determined for any image so that the variance ratio of intra-class variance and inter-class variance regarding the background and pattern area is maximized when binarized.
For this reason, depending on the environment at the time of imaging when the threshold value is determined as a fixed value, problems such as emphasis and extraction of parts other than the worn parts of the trolley wire, or the worn parts themselves are not extracted are solved.

[Noise removal from binarized line sensor image]
When a binarized line sensor image is constructed from the line sensor image by binarization processing, fine dot-like noise may be included depending on the scratches on the trolley wire worn portion or the background portion.
Therefore, when the binarized line sensor image is sent to the noise removal processing unit 60 via the memory 40, the noise removal processing unit 60 performs expansion and contraction processing of the binarization processing to remove these noises (steps). S3).

[Edge detection of trolley wire wear part]
Subsequently, when the binarized line sensor image from which noise is removed is sent to the trolley wire wear part edge detection unit 70 via the memory 40 together with the parameters, the trolley line wear part edge detection unit 70 displays white on the binarization line sensor image. Edges on both sides of the trolley wire wear portion represented by the area are detected (step S4).

When these edge points are searched from the left for a certain line, the point that changes from the black area of the background to the white area of the worn area is the edge point on the left side of the worn area, and from the white area of the worn area to the black area of the background. The changing point can be detected as an edge point on the right side of the wear portion.
By performing this process for each upper to lower line of the image, as shown in FIG. 15, the edge of the trolley line wear portion relating to one binarized line sensor image is detected.

[Calculation of trolley wire wear part width]
Further, the edge data on both sides of the trolley wire wear portion detected from the binarized line sensor image is sent to the trolley wire wear portion width calculation unit 80 together with the parameters and the trolley wire height data via the memory 40, and the trolley wire wear portion. The width calculation unit 80 calculates the distance between the edge points on both sides on one scanning line of the line sensor 10 as the width on the image of the trolley line wear portion (step S5).

Specifically, from the height from the line sensor 10 to the trolley line 4 at this time, the lens focal length, the sensor width, and the number of sensor pixels, the image resolution which is the degree of the actual dimension [mm] with respect to one pixel [pix] [ mm / pix] is calculated, and the width of the trolley wire wear portion is multiplied by the image resolution to calculate the width of the trolley wire wear portion.
As for the height from the line sensor 10 to the trolley wire 4 used for the calculation, another means (for example, “trolley wire wear measuring device by image processing (Japanese Patent Application No. 2005-068793)”, hereinafter referred to as the prior proposal 2). It is determined using data obtained by the above.

The edge data thus obtained, the trolley wire wear part, the line sensor image used for the calculation, data indicating the corresponding line number, and the like are recorded.
As described above, the trolley wire wear measuring device based on the image processing of the previous proposal 1 is a non-contact method, and thus can be operated at high speed and can measure a long distance section in a short time. .

  In addition, because the sensor is installed at a position away from existing structures such as points, air sections, and anchors due to the structure of the device, compared to the method of directly measuring the thickness of the trolley wire using a rotating roller and an optical sensor. There is no need to consider collisions with existing structures, and line sensor images can be taken basically in all sections, and trolley lines in the measurement section and image data of existing structures in the vicinity are acquired. Can do.

In addition, there is no need to use special illumination, there is no difficulty in handling that takes into account the effects on the human body, such as using laser light, and there is no need to perform precise alignment between the light source and the light receiving device. There is not.
Furthermore, it is possible to automatically calculate a relatively good threshold value in any imaging environment, and the phenomenon that the trolley line due to the low brightness that occurs when the threshold value is determined by a constant is not improved is improved. In addition, since the line sensor image remains as data, it is possible to confirm the problem of the portion having a problem with the trolley wire wear by looking at the image of the portion.

JP-A-10-194015

  The trolley wire wear can be measured by directly measuring the thickness of the trolley wire using a ruler such as a caliper, directly measuring the thickness of the trolley wire using an optical sensor, and the width of the trolley wire wear group using a sodium lamp. There are a method for measuring by irradiating laser beam and converting the thickness of the trolley wire from the worn portion, and a trolley wire wear measuring method by the image processing of the previous proposal 1, which has the following problems.

(1) In the case of the method of directly measuring the thickness of the trolley wire using a ruler such as a caliper, since the measurement is performed manually by an operator, the measurement is troublesome and cannot be automated. It is difficult to measure between.

(2) In the case of the method of directly measuring the thickness of the trolley wire using an optical sensor, it is necessary to operate at a low speed because it involves contact between the rotating roller and the trolley wire.
In addition, because the structure is such that the trolley wire is sandwiched between sensors, it cannot be used where there are structures that collide with sensors such as points, air sections, and anchors. It is necessary to move the device away from the measurement position so that it does not collide.

(3) When the width of the trolley wire wear part is measured by irradiating a sodium lamp or laser light and the thickness of the trolley wire is converted from the wear width, first, special illumination light such as a sodium lamp or laser light is prepared. In particular, when laser light is used, it is necessary to consider the influence on the human body, so care must be taken in handling. Next, since it is easily affected by noise such as clamps sandwiching the trolley wire and structures reflected in the background, there is no way to confirm the wrong measurement result due to some noise, so trolley wire wear There is no means for confirming the cause of a problem portion, and eventually it takes time to confirm using a method of directly measuring the thickness of the trolley wire. In addition, it is necessary to precisely adjust the positions of the light source and the light receiving device so as to receive specularly reflected light.

(4) The trolley wire is not continuous with a single trolley wire over the entire area, but a plurality of trolley wires are connected and used for convenience of manufacturing and construction. At this connection point (hereinafter referred to as “section section”), as shown in FIG. 18, since the trolley wire is usually overlapped by one or two overhead pole sections, at the beginning and end of the overlapping section, As shown in FIG. 19, there is a portion that does not cross the image vertically as a trolley line in the binarized line sensor image in the measurement section. Therefore, as shown on the left side of FIG. 19, if only the white area (transverse white area) that crosses vertically in the binarized line sensor image is recognized as a trolley line wear part, the binarized line as shown on the right side of FIG. 19. A white area (non-crossing white area) that does not cross in the sensor image cannot be recognized as a trolley line. In other words, if the trolley line disappears in the middle of the binarized line sensor image, the image is not traversed up and down, so that it may not be recognized as a trolley line and may be processed as background noise.

A trolley wire wear measuring device by image processing according to claim 1 of the present invention that solves the above problems, line sensor creating means for creating a line sensor image by arranging the luminance signals of the scanning lines obtained from the line sensor in time series, Binarizing means for emphasizing a trolley wire wear portion by binarizing the line sensor image into a binarized line sensor image; noise removing means for removing noise from the binarized line sensor image; an edge detecting means for detect an edge of the trolley wire worn portion from the binarized line sensor image that has been removed of noise, trolley wire wear to calculate the wear width of the trolley wire from the distance between edge points of the trolley wire worn portion In a trolley wire wear measuring apparatus using image processing having a width calculating means, the line sensor image is divided into M lines and numbered consecutively. An image that is N lines (M ≧ N) of the M lines in the (n−1) th line sensor image and is continuous in time series with the nth line sensor image (hereinafter referred to as “line sensor image”). ) And N lines of the M lines in the (n + 1) th line sensor image, which are continuous in time series with the nth line sensor image (hereinafter referred to as a preliminary line image). And a preliminary line image connecting means for creating a line sensor image comprising (M + 2N) lines, and the noise removing means, the edge detecting means and the trolley wire wear width calculating means are provided by the preliminary line image connecting means. Each process is performed on the created line sensor image including (M + 2N) lines.

  A trolley wire wear measuring apparatus using image processing according to claim 2 of the present invention for solving the above-mentioned problem is the trolley wire wear measuring apparatus according to claim 1, wherein binarization is performed among the line sensor images including the (M + 2N) lines. When a white area crossing in the time series direction (hereinafter referred to as a transverse white area) exists in the preliminary line image after processing, a partial area in the time series direction is obtained after the binarization processing in the n-th line sensor image. A pattern analysis means for determining that a white area that crosses (hereinafter, non-crossing white area) is a trolley wire wear portion is added.

  The trolley wire wear measuring apparatus according to claim 3 of the present invention for solving the above-mentioned problems is the trolley wire wear measuring apparatus according to claim 1, wherein the nth line sensor image comprising the (M + 2N) lines. When a white line crossing in the time-series direction exists in the line sensor image after binarization processing, or binarization is performed on the n-th line sensor image among the line sensor images composed of the (M + 2N) lines. There is a non-crossing white area that partially crosses in the time-series direction after processing, and the non-crossing white area is continuous with the cross-white area that crosses in the time-series direction in the preliminary line image after binarization processing. If any of the cases is true, pattern analysis means for determining that the non-crossing white area is a trolley wire wear portion is added.

As described in paragraph 0022, in the image after binarization processing using the image of the line sensor position where the image analysis is performed, a continuous white area that traverses up and down is extracted as a trolley line. When the image after the valuation processing did not cross the top and bottom, it was processed as noise.
On the other hand, according to the present invention, using the previous analysis area and the next analysis area that are continuous in time series, the continuous white area that traverses the set section of these areas up and down and the white area that continues in the current analysis area are also included. By recognizing as a trolley line, it is possible to recognize and extract a trolley line even if the image after binarization processing in the current analysis area such as a section section does not cross vertically.

The best mode for carrying out the present invention will be described below with reference to FIG.
FIG. 1A is a conceptual diagram of image processing relating to trolley wire wear measurement according to the present invention, and FIG. 1B is a conceptual diagram schematically showing the inside of a binarization storage area.
That is, the line sensor images of all the measurement sections are stored in the hard disk drive (hereinafter referred to as HDD) 100, read out from the HDD 100, and developed in the binarization storage area 200.

In the binarization storage area 200, the line sensor image is obtained by arranging one line of image data (scanning line luminance signal) in time series t ₁ , t ₂ , t ₃ , t ₄ ,.
After the binarization process, when the line sensor image is divided into M lines and given a serial number n, the first line sensor image is arranged with M pieces of image data of one line in time series t _{1 to} t _M. The second line sensor image is an image data of one line arranged in _M time series t _{M to} t _{M + M. As} a general formula, the n th line sensor image is One line of image data is arranged in _M time series t _{M ·} ( _{n−1) +1 to} t _{M · n} .

As described above, the line sensor image divided for each M line is obtained by arranging one line of image data in M time series, and is used in the image processing described in the background section. It is the same as the image.
The line sensor image with the serial number n is, for example, temporarily stored in the HDD 300 and then read from the HDD 300, and the (n−1) th line sensor is read from the nth line sensor image. N lines (M ≧ N) of the M lines in the image, and a spare line that is continuous in time series with the nth line sensor image and N lines of the M lines in the (n + 1) th line sensor image Then, the n-th line sensor image is connected to a preliminary line image that is continuous in time series to create a line sensor image composed of (M + 2N) lines.

The trolley wire wear width measuring device according to the image processing according to the present invention uses the line sensor image formed of the (M + 2N) lines thus created as a determination target in the subsequent image processing.
Therefore, in the present invention, compared with the binarized line sensor image used in the conventional image processing, the area of the image is expanded by the amount of the preliminary line image, that is, the image of 2N lines. It will be a thing.

A trolley wire wear measuring apparatus using image processing according to the first embodiment of the present invention is shown in FIGS. FIG. 7 shows a flowchart of the process, and FIG. 8 shows an example of the apparatus configuration.
Also in this embodiment, as described in the background art section, as shown in FIG. 13, the line sensor 10 is used as an image input means, and illumination is not a sodium lamp or laser light, but a normal lighting fixture 2. They are common in that they are used.

Further, in this embodiment, as shown in FIG. 8, in the trolley wire wear measuring apparatus using the image processing described in the background art section, a preliminary line connection process is performed instead of the line sensor image creation unit 20 and the memory 30. A line sensor image creation unit 21 and a memory 31 are provided, and a pattern analysis processing unit 90 that performs pattern analysis is further added.
That is, as shown in FIG. 8, the luminance signals of the scanning lines acquired by the line sensor are arranged in time series by the line sensor image creating unit 21 and stored in the memory 31 (step S1). Utilizing this, the preliminary line image is connected (step S6).

Here, if the number M of lines that have been subjected to the image analysis described in the Background Art column is, for example, 1000 lines, the analysis area has conventionally been every 1000 lines, but in this embodiment, every 1000 lines. The line sensor image is created by connecting the preliminary line portions overlapping the previous and next analysis areas above and below the image.
For example, as shown in FIGS. 2 and 3, a 1000-line image B, which is an analysis area where conventional image analysis has been performed, and a 100-line image, which is a part of an image in the previous analysis area adjacent to the analysis line, are reserved lines. Assuming that an image A is an image of 100 lines that is a part of an image in the next analysis area, which is a preliminary line image C, the preliminary line images A and C are connected to an image B of 1000 lines.

For example, when the image pickup speed of the inspection vehicle is 40 km / h and when the image is picked up at the same frequency, the movement distance per line differs. Therefore, an appropriate number of lines is set according to the imaging environment of the inspection vehicle.
In this embodiment, the spare line N is set to 100 lines as described above, and therefore the following image processing is performed using a line sensor image of 1000 + 100 + 100 = 1200 lines.

That is, as described in the background art section, the line sensor image in which the preliminary line images are connected in this way is subjected to discriminant analysis binarization processing (step S2), noise removal processing (step S3), trolley wire wear portion. After the edge detection process (step S4), it is sent to the pattern analysis processing unit 90 to perform the pattern analysis process (step S7).
In the pattern analysis processing, as shown in the right side portion and the left side portion of FIG. 4, as described above in paragraph 0022, the crossing white area crossing the 1000-line image B is determined as a trolley line.

  Further, in the pattern analysis process, as shown in the left part of FIG. 5, a transverse white area that crosses the preliminary line image A and a non-crossing white area that crosses the white area up to the middle of the image B of 1000 lines are continuous. In this case, the non-crossing white area is determined as a trolley line, and as shown in the right part of FIG. 6, the white area crossing from the middle of the image B of 1000 lines and the crossing white area crossing the preliminary line image C Is characterized in that a non-crossing white area is determined as a trolley line.

That is, in the image analysis described above in paragraph 0022, it is determined whether or not the trolley line is only in the analysis area of the 1000-line image B. On the other hand, the pattern analysis in this embodiment is performed in addition to the 1000-line image B. The 100-line preliminary line image A and image C are also used to determine whether or not the trolley line is used.
Therefore, in this embodiment, the trolley line that is processed as noise in the section shown in FIG. 18 is surely extracted as compared with the case where only the 1000-line image B image is determined as the trolley line. It can be done.

  The configuration and processing other than the above, such as the subsequent trolley wire wear portion width detection processing (step S5), are the same as the configuration and processing shown in FIGS. 16 and 17 and have the same effect.

FIGS. 11 and 12 show a trolley wire wear measuring apparatus using image processing according to a second embodiment of the present invention. FIG. 11 shows a flowchart of the process, and FIG. 12 shows an example of the apparatus configuration.
Also in this embodiment, as described in the background art section, as shown in FIG. 13, the line sensor 10 is used as an image input means, and illumination is not a sodium lamp or laser light, but a normal lighting fixture 2. Is to be used.

  That is, the line sensor 10 is installed on the roof of the inspection vehicle 3 so as to look up vertically, the scanning line of the line sensor 10 is perpendicular to the traveling direction of the inspection vehicle 3, and the scanning line crosses the trolley line 4. Like that. On the roof of the inspection vehicle 3, a pantograph 5 that comes into contact with the trolley wire 4 is provided, and the vicinity of the trolley wire 4 that comes in contact with the pantograph 5 is illuminated by the lighting device 2 and the state of the trolley wire 4 is lined up. Photographed by the sensor 10.

  An image signal of the trolley wire 4 photographed by the line sensor 10 is input to a measurement personal computer (hereinafter referred to as PC) 6 installed in the vehicle. The measurement PC 6 arranges the luminance signals of the scanning lines obtained from the line sensor 1 in time series, creates a line sensor image (planar image), and stores it in the recording device 7 as an input image.

As the measurement PC 6, for example, as shown in FIG. 12, a line sensor image creation unit 22, memories 32 and 40, a binarization processing unit 51, a noise removal processing unit 60, a trolley wire wear part edge detection unit 70, a trolley The line wear portion width calculation unit 80 is configured.
According to the flowchart shown in FIG. 11, the measurement PC 6 performs image processing on the input line sensor image to obtain the width of the worn portion of the trolley wire according to the following procedure.

[Acquisition of line sensor image]
First, the image signals acquired by the line sensor 10 are arranged in time series by the line sensor image creating unit 22 and stored in the memory 32 as a line sensor image (step T1).

[Connection of line sensor images]
The line sensor image stored in the memory 32 includes a predetermined number of lines arranged in time series. As shown in FIG. 9, the upper and lower ends of the image are processed by a noise filter or the like in the subsequent image processing process. There are parts that cannot be processed with respect to several lines, and data loss occurs.
In order to prevent the phenomenon that data of several lines is lost, it is supposed to be continuous, but the image that is the target of processing is connected vertically to the image to be processed. A concatenation process for concatenating several lines of images is performed (step T2).

That is, the line sensor image creation unit 22 performs the (n−1) th line sensor image with respect to the nth line sensor image with respect to the line sensor image stored in the memory 32, as shown in FIG. A part (preliminary line image) and a part (preliminary line image) of the (n + 1) th line sensor image are connected to form an image processing area.
Therefore, in the following image processing, there are portions that cannot be processed with respect to the top and bottom lines of the image, and even if data loss occurs in the preliminary line image, data loss for the n-th line sensor image has not occurred. Can be prevented.

[Enhancement of trolley wire wear by binarization]
Next, the line sensor image is sent to the binarization processing unit 51 through the memory 40 together with the parameters, and binarization processing is performed by the binarization processing unit 50 (step T3).
Here, the worn portion of the trolley wire is a portion where the trolley wire is scraped by a pantograph, and therefore, the trolley wire has a stronger gloss than the portion that is not worn. Therefore, the worn portion of the trolley wire is also the background portion on the line sensor image. Is taken as a band-like portion having a different luminance value compared to

The binarization processing unit 50 sets a threshold value so as to separate a trolley wire wear part (pantograph contact surface) and other background parts (existing structures, etc.), and uses the threshold value for 2 to the line sensor image. A binarization line sensor image is constructed by performing a binarization process and emphasizing a worn portion of a trolley wire.
As shown in FIG. 14, the binarized line sensor image has a trolley wire wear portion in a white area and a background portion in a black area.

[Noise removal from binarized line sensor image]
When a binarized line sensor image is constructed from the line sensor image by binarization processing, fine dot-like noise may be included depending on the scratches on the trolley wire worn portion or the background portion.
Therefore, when the binarized line sensor image is sent to the noise removal processing unit 60 via the memory 40, the noise removal processing unit 60 performs expansion and contraction processing of the binarization processing to remove these noises (steps). T4).

[Edge detection of trolley wire wear part]
Subsequently, when the binarized line sensor image from which noise is removed is sent to the trolley wire wear part edge detection unit 70 via the memory 40 together with the parameters, the trolley line wear part edge detection unit 70 displays white on the binarization line sensor image. Edges on both sides of the trolley wire wear portion represented by the area are detected (step T5).

When these edge points are searched from the left for a certain line, the point that changes from the black area of the background to the white area of the worn area is the edge point on the left side of the worn area, and from the white area of the worn area to the black area of the background. The changing point can be detected as an edge point on the right side of the wear portion.
By performing this process for each upper to lower line of the image, as shown in FIG. 15, the edge of the trolley line wear portion relating to one binarized line sensor image is detected.

[Calculation of trolley wire wear part width]
Further, the edge data on both sides of the trolley wire wear portion detected from the binarized line sensor image is sent to the trolley wire wear portion width calculation unit 80 together with the parameters and the trolley wire height data via the memory 40, and the trolley wire wear portion. The width calculation unit 80 calculates the distance between the edge points on both sides on one scanning line of the line sensor 10 as the width on the image of the trolley line wear portion (step T6).

Specifically, from the height from the line sensor 10 to the trolley line 4 at this time, the lens focal length, the sensor width, and the number of sensor pixels, the image resolution which is the degree of the actual dimension [mm] with respect to one pixel [pix] [ mm / pix] is calculated, and the width of the trolley wire wear portion is multiplied by the image resolution to calculate the width of the trolley wire wear portion.
The height from the line sensor 10 used for the calculation to the trolley line 4 is obtained using data obtained by another means such as the previous proposal 2 or the like.

  The edge data thus obtained, the trolley wire wear part, the line sensor image used for the calculation, data indicating the corresponding line number, and the like are recorded.

As described above, according to the trolley wire wear measuring apparatus using image processing according to the present embodiment, the following effects can be obtained.
(1) Since it is a non-contact method, it can be operated at high speed, and a long distance section can be measured in a short period.
(2) Due to the structure of the device, sensors are installed at positions away from existing structures such as points, air sections, anchors, etc. Compared to the method of directly measuring the thickness of the trolley wire using a rotating roller and an optical sensor. Therefore, it is not necessary to consider the collision with the existing structure, and the measurement can be continuously performed even in the place where the existing structure exists.
(3) Data omission at each image edge is eliminated, and continuous trolley wire wear width data can be obtained.
(4) Unlike simple interpolation of trolley wire wear width data, data is created from actual images by linking image data that moves back and forth in time, so the data interpolation unit can also obtain more accurate data. .
(5) Basically, it is possible to capture line sensor images in all sections, and it is possible to acquire trolley lines in the measurement section and image data of existing structures in the vicinity.
(6) There is no need to use special lighting.
(7) Compared to the method using laser light, there is no need to consider the influence on the human body, and handling is simple.
(8) Since it is not necessary to receive the reflected light of the trolley wire wear part by regular reflection, there is no trouble of performing precise alignment between the light source and the light receiving device.
(9) Since the line sensor image in the measurement section remains, it is possible to confirm the problem by looking at the image of the portion where there is a problem with trolley wire wear.

  The present invention can be widely used industrially as a device for measuring the width of a worn portion of a trolley wire.

FIG. 1A is a conceptual diagram of image processing relating to trolley wire wear measurement according to the present invention, and FIG. 1B is a conceptual diagram schematically showing the inside of a binarization storage area. It is explanatory drawing which shows the example of the binarization line sensor image which a white area crosses image A, B, C. It is explanatory drawing which shows the connection of an image analysis area. It is explanatory drawing which shows the example of the binarization line sensor image with which the crossing white area which crosses the image A or C and the crossing white area which crosses the image B are continuous. It is explanatory drawing which shows the example of the binarization line sensor image in which the crossing white area which crosses the image A, and the non-crossing white area which partially crosses the image B continue. It is explanatory drawing which shows the example of the binarization line sensor image with which the white area which crosses the image C, and the white area which crosses the image B continue. It is a flowchart of the trolley wire wear part measuring apparatus which concerns on Example 1 of this invention. It is a block diagram of the trolley wire wear part measuring apparatus which concerns on Example 2 of this invention. It is explanatory drawing of the line sensor image which has a data missing part. A part of the (n−1) th line sensor image (preliminary line image) and a part of the (n + 1) th line sensor image (preliminary line image) are connected to the nth line sensor image. It is explanatory drawing which shows an image processing area. It is a flowchart of the trolley wire wear part measuring apparatus which concerns on Example 2 of this invention. It is a block diagram of the trolley wire wear part measuring apparatus which concerns on Example 2 of this invention. It is the schematic which shows the example of arrangement | positioning of the trolley wire wear measurement line sensor. It is explanatory drawing which shows the example of the binarization line sensor image of a trolley wire wear part. It is explanatory drawing which shows the example of edge detection of a trolley wire wear part. It is a flowchart of the trolley wire wear measurement which concerns on background art. It is a block diagram of the trolley wire wear measuring apparatus which concerns on background art. It is explanatory drawing of a trolley line connection location (section area). It is explanatory drawing which shows the example of the binarization line sensor image of a trolley line connection location terminal.

Explanation of symbols

2 Lighting equipment 3 Inspection vehicle 4 Trolley wire 5 Pantograph 6 PC for measurement
7 recording device 10 line sensor 20, 21, 22 line sensor image creation unit 30, 31, 32, 40 memory 50 discriminant analysis binarization processing unit 51 binarization processing unit 60 noise processing unit 70 trolley wire wear unit edge detection unit 80 Trolley wire wear part width calculation part 100,300 HDD (hard disk drive)
200 Storage area for binarization A, B, C Image

Claims (3)

Line sensor creation means for creating a line sensor image by arranging the luminance signals of the scanning lines obtained from the line sensor in time series, and binarizing the line sensor image to obtain a binarized line sensor image. and binarizing means emphasizing portion to detect a noise removal means for removing noise of the binary line sensor image, the edges of the trolley wire worn portion from the binarized line sensor image that has been removed of noise In the trolley wire wear measuring device by image processing, comprising edge detection means and trolley wire wear width calculation means for calculating the wear width of the trolley wire from the distance between edge points of the trolley wire wear portion, the line sensor image is converted into M lines. M lines in the (n−1) -th line sensor image with respect to the n-th line sensor image that is numbered consecutively every time. Of these, N lines (M ≧ N) and an image that is continuous in time series with the nth line sensor image (hereinafter referred to as a preliminary line image) and of the M lines in the (n + 1) th line sensor image Preliminary line image coupling means for creating a line sensor image consisting of (M + 2N) lines by connecting N-line and n-th line sensor images continuous in time series (hereinafter referred to as preliminary line images). The noise removing means, the edge detecting means, and the trolley line wear width calculating means perform each process on a line sensor image composed of (M + 2N) lines created by the preliminary line image connecting means. A trolley wire wear measuring device using image processing. When a white area (hereinafter referred to as a transverse white area) crossing in the time series direction exists in the preliminary line image after the binarization process among the line sensor images composed of the (M + 2N) lines, the nth line A pattern analysis means for determining that a white area partially traversing in the time-series direction after the binarization processing in the sensor image (hereinafter, non-crossing white area) is a trolley line wear part is added. Item 1. A trolley wire wear measuring device according to item 1. Of the line sensor images composed of the (M + 2N) lines, the n-th line sensor image has a transverse white area that crosses in the time-series direction after binarization processing, or a line composed of the (M + 2N) lines Among the sensor images, the n-th line sensor image has a non-crossing white area that partially crosses in the time-series direction after binarization processing, and the non-crossing white area is the binarized image after the binarization process. A pattern analysis means for determining that the non-crossing white area is a trolley wire wear part when it corresponds to any of the cases where it is continuous with the crossing white area traversing in the time-series direction in the preliminary line image is characterized. The trolley wire wear measuring apparatus according to claim 1.

Images