JP5776378B2 - Crossover measuring device by image processing - Google Patents

Description

  The present invention relates to a three-dimensional position measurement of a crossover line by image processing, and more particularly to a crossover measurement apparatus by image processing that performs robust measurement against background brightness fluctuations in the daytime.

  Although there is a trolley line as an electric railway facility, it is necessary to cross this trolley line at the place where the tracks cross. Of the trolley lines that intersect, the line that intersects differently from the main line is called a “crossover”. Conventionally, the three-dimensional position of the crossover line is measured so that the crossover line does not hinder the running of the railway vehicle.

  As a device for measuring the three-dimensional position of the crossover line, a line sensor camera and a lighting device are installed on the roof of the vehicle, and the discriminant analysis is performed on the image acquired from the line sensor camera by illuminating the crossover line. Some binarization is used to extract only the crossover line as a white region and obtain the three-dimensional position of the crossover line in a non-contact manner (see, for example, Patent Document 1). In addition, this patent document 1 also discloses that measurement is performed in the daytime by removing disturbance light such as sunlight using a sodium lamp and a bandpass filter and obtaining a good image.

  In addition, as a device for measuring the trolley line, binarization processing is performed on the image acquired from the line sensor camera, and the white area larger than the thickness parameter of the trolley line is inverted to the black area as an empty part. There is also one that extracts only the trolley wire wear surface and measures the wear amount of the trolley wire (see, for example, Patent Document 2).

  There is also a pantograph peripheral obstacle detection device in which a mask that does not process the background portion is set using a preset grayscale threshold (see, for example, Patent Document 3).

JP 2009-192394 A JP 2007-271445 A JP 2006-250775 A

  However, in the case of the one described in Patent Document 1, it is necessary that only the crossover portion is a white region and the background portion is a black region when binarized. There was a possibility that the crossover line and the background could not be separated because of the white area. In addition, although a sodium lamp and a bandpass filter are separately prepared and attached so that measurement can be performed even in the daytime, there is a problem in that the preparation of the apparatus is expensive and the apparatus configuration becomes complicated.

Here, when the daytime measurement is performed using the apparatus described in the cited document 1, it is conceivable to adopt an apparatus configuration in which illumination is removed.
When a crossover is photographed in such a configuration, the background is brightly reflected by sunlight (when binarized, a white area), and the crossover appears black as a sunlight shield (when binarized, a black area). Therefore, it is considered that the crossover line and the background can be separated even in the daytime, and the three-dimensional position measurement of the crossover line in the daytime can be performed.

However, when the three-dimensional position measurement of the crossover line is performed with such an apparatus, the following problems have been considered.
(1) When there is a brightness spot (shadow due to weather change or clouds, etc., hereinafter simply referred to as “bright spot”) in the background, the thickness parameter as disclosed in Patent Document 2 is used. In this method, there is a possibility that the brightness spot is erroneously detected as a crossover. In addition, there is a possibility that the background part and the part that is not so cannot be distinguished by the method of setting a fixed grayscale threshold value as disclosed in Patent Document 3.
(2) When sunlight strikes the crossover and sunlight reflection occurs, a part of the crossover is photographed extremely brightly. When this image is binarized, the part that was captured extremely brightly is the background. As a result, a white area is formed, and the width of the black area of the crossover portion is reduced, so that the crossover line width is calculated to be smaller than the actual width. Since the three-dimensional position such as the crossover line height is calculated based on the crossover line width, correct measurement cannot be performed when sunlight reflection occurs.

  In view of the above, an object of the present invention is to provide a crossover measuring apparatus using image processing capable of separating a crossover line and a background with high accuracy.

A crossover measuring apparatus using image processing according to the first aspect of the present invention for solving the above-described problems is arranged on the roof of the vehicle so that the scanning line direction is perpendicular and horizontal to the traveling direction of the vehicle. In a crossover measuring apparatus using image processing comprising a line sensor camera for photographing and an image processing means installed inside the vehicle, the image processing means converts an image signal input from the line sensor camera in time series. Line sensor image creation means for creating line sensor images arranged side by side, discriminant analysis binarization processing means for creating a binarized image by performing discriminant analysis binarization processing on the input image, and input Noise removal processing means for generating a noise-removed image by performing noise removal processing on the input image, and a crossover section error detecting edge points on both sides of the crossover line from the input image. Detecting means; background separation processing means for separating the crossover line from the background of the input image; and a crossover line for calculating the height of the crossover line from the positions of the edge points on both sides of the crossover line And a crossover portion deviation calculating means for calculating a deviation of the crossover line from the height of the crossover line, and the background segmentation processing means includes edge points on both sides of the crossover line. Background luminance average value calculation processing means for calculating a background luminance average value by aggregating luminance values in a certain range on the background side based on the edge point for the input image based on the position; and the background luminance A threshold value is set based on the average value, and brightness distributions in a predetermined range centering on edge points on both sides of the crossover line of the input image based on the threshold value are compared to compare the reflection of sunlight. Sunlight to determine presence or absence Characterized in that comprising a morphism determination processing unit.

The crossover measuring device by image processing according to the second invention for solving the above-mentioned problem is the crossover measuring device by image processing according to the first invention, wherein the discriminant analysis binarization processing means is the line sensor. Discriminant analysis binarization processing is performed on the image to create the binarized image, and the noise removal unit performs noise removal processing on the binarized image to create the noise removal image, Crossover line edge detection processing means detects edge points on both sides of the crossover line with respect to the noise-removed image, and the background luminance average value calculation processing means calculates the background luminance average value using the line sensor image. And the said sunlight reflection determination process means determines the presence or absence of the said sunlight reflection using the said line sensor image, It is characterized by the above-mentioned.

The crossover measuring device by image processing according to the third invention for solving the above-mentioned problems is the crossover measuring device by image processing according to the first or second invention, wherein the solar reflection determination processing means When it is determined that there is light reflection, the measurement result is interpolated based on the data before and after the measurement is completed.

A crossover measuring apparatus using image processing according to a fourth aspect of the present invention for solving the above-described problems is provided on the roof of the vehicle so that the scanning line direction is set perpendicularly and horizontally to the traveling direction of the vehicle. In a crossover measuring apparatus using image processing comprising a line sensor camera for photographing and an image processing means installed inside the vehicle, the image processing means converts an image signal input from the line sensor camera in time series. Line sensor image creation means for creating line sensor images arranged side by side, discriminant analysis binarization processing means for creating a binarized image by performing discriminant analysis binarization processing on the input image, and input Noise removal processing means for generating a noise-removed image by performing noise removal processing on the input image, and a crossover section error detecting edge points on both sides of the crossover line from the input image. Detecting means; background separation processing means for separating the crossover line from the background of the input image; and a crossover line for calculating the height of the crossover line from the positions of the edge points on both sides of the crossover line And a crossover portion deviation calculating means for calculating a deviation of the crossover line from the height of the crossover line, and the background segmentation processing means sets the input image to an arbitrary section. Dividing and performing standard deviation background removal processing for each of the sections to determine whether or not the object is reflected in the section for each of the sections, and the object is reflected in the section. When the object is not shown in an adjacent section, the area from the adjacent section to the object is regarded as a background, and the object is not shown in the section, and the adjacent section If the object is reflected, And the standard deviation background removal processing means for creating a background removed image that interval the between region and the compartment from the object to between the compartment performing the standard deviation background removal processing regarded as background, both sides of edge point of the connecting wire Background luminance average value calculation processing means for calculating a background luminance average value by aggregating luminance values in a certain range on the background side based on the edge point for the input image based on the position; and the background luminance A threshold value is set based on the average value, and brightness distributions in a predetermined range centering on edge points on both sides of the crossover line of the input image based on the threshold value are compared to compare the reflection of sunlight. A sunlight reflection determination processing means for determining presence / absence, and when the crossover line height calculation means determines that there is no sunlight reflection in the sunlight reflection determination processing means, edges on both sides of the crossover line The height of the crossover is calculated from the position of the point.
A crossover measuring device by image processing according to a fifth aspect of the present invention for solving the above problem is the crossover measuring device by image processing according to the fourth aspect of the invention, wherein the standard deviation background removal processing means is the standard deviation. When there is an area that is not regarded as a background in the section subjected to the background removal process, the section is subdivided and the standard deviation background removal process is performed again for each of the subdivided sections. For each section, it is determined whether or not the object is shown in the section, and if the target is shown in the section and the target is not shown in the adjacent section, the adjacent section The area from the object to the object is regarded as the background, and when the object is not reflected in the section and the object is reflected in the adjacent section, the object from the adjacent section is Ward Between regions and the compartment to perform the standard deviation background removal processing regarded as background, characterized in that to create the background removed image.
A crossover measuring device by image processing according to a sixth aspect of the present invention for solving the above problem is the crossover measuring device by image processing according to the fourth aspect of the invention, wherein the size of the arbitrary section is the object. It is characterized in that the section in which is present is not continuous.

A crossover measuring device using image processing according to a seventh aspect of the present invention for solving the above problem is the crossover measuring device using image processing according to any one of the fourth to sixth aspects of the invention, wherein the sunlight reflection is performed. When it is determined by the determination processing means that there is sunlight reflection, the measurement result is interpolated based on the data before and after the measurement.

A crossover measuring apparatus using image processing according to an eighth aspect of the present invention for solving the above problem is the crossover measuring apparatus using image processing according to any one of the fourth to seventh inventions, wherein the standard deviation background A removal processing unit performs a standard deviation background removal process on the line sensor image to create the background removal image by removing the background in the line sensor image, and the discriminant analysis binarization processing unit performs the background removal image. The binarized image is subjected to discriminant analysis binarization processing to create the binarized image, and the noise removal unit performs noise removal processing on the binarized image to create the noise-removed image, and the transition Line edge detection processing means detects edge points on both sides of the crossover line with respect to the noise-removed image, and the background luminance average value calculation processing means uses the line sensor image to calculate the background luminance average. Calculates the sunlight reflecting determination processing means and judging the presence or absence of the sunlight reflecting using the line sensor image.

According to the crossover measuring apparatus using image processing according to the first aspect of the present invention, a line sensor camera that photographs the crossover line by setting the scanning line direction on the roof of the vehicle so that the scanning line direction is perpendicular and horizontal to the traveling direction of the vehicle; In the crossover measuring device by image processing provided with image processing means installed inside the vehicle, the image processing means creates a line sensor image in which image signals input from the line sensor camera are arranged in time series Line sensor image creating means for performing discriminant analysis binarization processing on the input image to create a binarized image, and noise removal processing on the input image Noise removal processing means for creating a noise-removed image by performing cross-over edge detection means for detecting edge points on both sides of the crossover line for the input image, and for the input image Background separation processing means for separating the connecting line from the background, a connecting line height calculating means for calculating the height of the connecting line from the positions of the edge points on both sides of the connecting line, and the connecting line from the height of the connecting line A crossover portion deviation calculating means for calculating the deviation of the background , and the background segmentation processing means is based on the positions of the edge points on both sides of the crossover line , and the background is based on the edge points for the input image. Background luminance average value calculation processing means for calculating a background luminance average value by aggregating luminance values in a certain range on the side, and setting a threshold value based on the background luminance average value and an image input based on the threshold value Because it consists of sunlight reflection judgment processing means that compares the brightness distribution of a predetermined range centering on the edge points on both sides of the crossover line and judges the presence or absence of sunlight reflection, the influence of sunlight reflection on the line sensor image If there is It can also perform crossover wire measurement accurately. Further, it is not necessary to use a device such as a sodium lamp or a band-pass filter, and the daytime measurement of the crossover can be performed with an inexpensive and simple device.

According to the crossover measuring apparatus using image processing according to the second invention, the discriminant analysis binarization processing means performs discriminant analysis binarization processing on the line sensor image to create a binarized image, and removes noise. The means performs noise removal processing on the binarized image to create a noise removal image, and the crossover edge detection processing means detects edge points on both sides of the crossover line from the noise removal image, and the background luminance average Since the value calculation processing means calculates the background luminance average value using the line sensor image and the sunlight reflection determination processing means determines the presence or absence of sunlight reflection using the line sensor image, the sunlight reflection is reflected on the line sensor image. Even when there is an influence of, crossover measurement can be performed accurately. Further, it is not necessary to use a device such as a sodium lamp or a band-pass filter, and the daytime measurement of the crossover can be performed with an inexpensive and simple device.

According to the crossover measuring device by the image processing according to the third invention, when it is determined that there is sunlight reflection in the sunlight reflection determination processing means, the measurement result is interpolated based on the data before and after the measurement, The influence on crossover measurement by sunlight reflection can be eliminated.

According to the crossover measuring apparatus using image processing according to the fourth aspect of the present invention, a line sensor camera that photographs the crossover line by setting the scanning line direction on the roof of the vehicle perpendicular to the traveling direction of the vehicle and horizontally. In the crossover measuring device by image processing provided with image processing means installed inside the vehicle, the image processing means creates a line sensor image in which image signals input from the line sensor camera are arranged in time series Line sensor image creating means for performing discriminant analysis binarization processing on the input image to create a binarized image, and noise removal processing on the input image Noise removal processing means for creating a noise-removed image by performing cross-over edge detection means for detecting edge points on both sides of the crossover line for the input image, and for the input image Background separation processing means for separating the connecting line from the background, a connecting line height calculating means for calculating the height of the connecting line from the positions of the edge points on both sides of the connecting line, and the connecting line from the height of the connecting line A crossover portion deviation calculating means for calculating the deviation of the image, and the background segmentation processing means divides the input image into arbitrary sections, and performs standard deviation background removal processing for each section, for each section. If there is an object in the section, and the object is in the section, and the object is not in the adjacent section, the area from the adjacent section to the object If the object is not shown in the section and the object is shown in the adjacent section, the area from the object in the adjacent section to the section and the section are set as the background. Performs standard deviation background removal processing for background removal. And the standard deviation background removal processing means for creating an image, based on the position of both sides of the edge points of the crossover line, respectively on the input image by aggregating a luminance value of a range of background-side relative to the edge point Background luminance average value calculation processing means for obtaining a background luminance average value, and a threshold value is set based on the background luminance average value, and the edge points on both sides of the crossover line of the image input based on the threshold value are set as the center. And a sunlight reflection determination processing means for determining the presence or absence of sunlight reflection by comparing the luminance distribution of a predetermined range, and the crossover line height calculation means determines that there is no sunlight reflection in the sunlight reflection determination processing means Since the height of the crossover line is calculated from the positions of the edge points on both sides of the crossover line, the crossover line can be identified even if there is a brightness spot in the background in the input image. , Misdetection of trolley wire wear surface The output can be further reduced. Further, it is not necessary to use a device such as a sodium lamp or a band-pass filter, and the daytime measurement of the crossover can be performed with an inexpensive and simple device.
According to the crossover measuring apparatus using image processing according to the fifth aspect of the present invention, when there is an area that is not regarded as the background in the section where the standard deviation background removal processing means has performed the standard deviation background removal processing, the section The standard deviation background removal process is performed for each subdivided section again, and it is determined whether or not the object is reflected in the section for each subdivided section. If the target object is not shown in the adjacent section, the area from the adjacent section to the target object is regarded as the background, and if the target object is not shown in the section, If the object is reflected, the background from the object in the adjacent section to the section and the standard deviation background removal process that regards the section as the background is performed to create a background-removed image. Carving Ukoto can.
According to the crossover measuring apparatus using image processing according to the sixth aspect of the present invention, the size of an arbitrary section is such that the section where the object exists is not continuous. Can be carved.

According to the crossover measurement by the image processing according to the seventh invention, when the sunlight reflection determination processing means determines that there is sunlight reflection, the measurement result is interpolated based on the data before and after the measurement is completed. It is possible to further reduce the erroneous detection of the wire wear surface.

According to the crossover measuring apparatus using image processing according to the eighth aspect of the invention, the background removal image obtained by removing the background in the line sensor image by the standard deviation background removal processing means performing the standard deviation background removal processing on the line sensor image. The discriminant analysis binarization processing unit performs discriminant analysis binarization processing on the background removal image to create a binarized image, and the noise removal unit performs noise removal processing on the binarized image. The crossover edge detection processing means detects edge points on both sides of the crossover line with respect to the noise removal image, and the background luminance average value calculation processing means uses the line sensor image to detect the background luminance. Since the average value is calculated and the sunlight reflection determination processing means determines the presence or absence of sunlight reflection using the line sensor image, erroneous detection of the trolley wire wear surface can be further reduced.

It is the schematic which shows the example of installation of the crossover measuring apparatus by the image process which concerns on Example 1 of this invention. It is a block diagram which shows schematic structure of the computer for measurement of the crossover measuring apparatus by the image process which concerns on Example 1 of this invention. It is explanatory drawing which shows an example of the standard deviation background removal process which concerns on Example 1 of this invention. It is explanatory drawing which shows the other example of the standard deviation background removal process which concerns on Example 1 of this invention. It is a flowchart which shows the flow of the trolley wire wear measurement process which concerns on Example 1 of this invention. It is a block diagram which shows schematic structure of the computer for measurement of the crossover measuring apparatus by the image process which concerns on Example 2 of this invention. It is a flowchart which shows the flow of the trolley wire wear measurement process which concerns on Example 2 of this invention. It is explanatory drawing which shows an example of the line sensor image which concerns on Example 2 of this invention. It is explanatory drawing which shows an example of the binarized image which concerns on Example 2 of this invention. It is explanatory drawing which shows an example of the luminance distribution of the edge point vicinity which concerns on Example 2 of this invention. It is a block diagram which shows schematic structure of the computer for measurement of the crossover measuring apparatus by the image process which concerns on Example 3 of this invention. It is a flowchart which shows the flow of the trolley wire wear measurement process which concerns on Example 3 of this invention.

  The details of the crossover measuring device by image processing according to the present invention will be described below with reference to the drawings.

(Standard deviation background removal)
A first embodiment of a crossover measuring apparatus using image processing according to the present invention will be described with reference to FIGS. The present embodiment does not perform binarization processing on an image acquired from a line sensor camera as it is, but uses a standard deviation as a preprocessing to determine a region where the brightness is substantially uniform, thereby improving the brightness on the background. Then, the influence of the spots is removed, and then the binarization process is performed to separate the background and the jumper line.

As shown in FIG. 1, in this embodiment, a line sensor camera 2 is installed on the roof of a vehicle 1, and a measuring computer 3 and a recording device 4 as image processing means are arranged in the vehicle 1. ing.
The direction of the line sensor camera 2 is set so that the vertical direction is photographed, and the scanning line direction is the direction of sleepers of rails (not shown), in other words, the direction orthogonal to the traveling direction of the vehicle 1. . Thereby, the scanning line of the line sensor camera 2 crosses the crossover line 7 which intersects the main line 6.

  As shown in FIG. 2, the measurement computer 3 includes a line sensor image creation unit 3a as a line sensor image creation unit, a standard deviation background removal processing unit 3b as a background segmentation processing unit, and a discriminant analysis binarization processing unit. Discriminant analysis binarization processing section 3c, noise removal processing section 3d as noise removal processing means, crossover edge detection section 3e as crossover edge detection means, and crossover section as crossover height calculation means A height calculation unit 3f and a crossover part deviation calculation unit 3g as a crossover part deviation calculation unit are provided.

  The line sensor image creation unit 3a creates a line sensor image (planar image) by arranging the luminance signals of the scanning lines input from the line sensor 2 in time series. The created line sensor image is input to the standard deviation background removal processing unit 3b via the memories M1 and M2.

  The standard deviation background removal processing unit 3b performs standard deviation background removal processing on the input line sensor image. An image created by the standard deviation background removal processing unit 3b (hereinafter referred to as background removal image) is input to the discriminant analysis binarization processing unit 3c via the memory M2.

  Hereinafter, the process in the standard deviation background removal processing unit 3b will be described in detail with reference to FIG. In the crossover measuring apparatus using image processing according to the present embodiment, the standard deviation background removal processing unit 3b, as shown in FIG. 3A, displays the line sensor image 11 in an arbitrary section (first section 12a in the present embodiment). , Second section 12b, third section 12c), and standard deviation background removal processing is performed for each section.

  First, the standard deviation background removal process is performed on the first section 12a. In the example shown in FIG. 3A, since the crossover 7 as the object is not shown in the first section 12a, the standard deviation value becomes low and it is determined that the entire first section 12a is the background. Thus, the first section 12a determined to be the background is painted white as a painted section so as not to affect the subsequent processing (FIG. 3B).

  Subsequently, a standard deviation background removal process is performed on the second section 12b. In the example shown in FIG. 3A, since the crossover line 7 is shown in the second section 12b, the standard deviation value becomes high, and some structure (here, the crossover line 7. In the following, the reflection is shown). It is judged that there is a thing). When it is determined that the reflection is reflected in this way, it is determined whether the adjacent section and the section where the standard deviation background removal processing is completed (here, the first section 12a) is a filled section. In the case of the painted section, a process of painting the area up to the reflected image as a background portion is performed in white (FIG. 3C).

  Next, a standard deviation background removal process is performed on the third section 12c. In the example shown in FIG. 3A, since the crossover 7 is not shown in the third section 12c, the standard deviation value is lowered and the entire third section 12c is determined as the background, and is the same as the first section 12a. , It is painted white as a filled section. Furthermore, the section determined to be the background in the sections other than the first section 12a is an adjacent section, and whether or not the section in which the standard deviation removal processing is completed (here, the second section 12b) is a painted section. If it is not a painted section, the area from the reflection in the adjacent section to the third section 12c is regarded as a background part and is painted white (FIG. 3 (d)).

  Here, as shown in FIG. 4A, for example, two trolley lines 7-1 and 7-2 are captured, and there are reflections in adjacent sections, or two in one section. When there are two or more reflection objects in the line sensor image 11 due to the presence of two reflection objects, as a result of performing the above-described processing, two trolleys as shown in FIG. There may be a case where the space between the lines 7-1 and 7-2 is not filled. In such a case, the area that has not been filled is subdivided into sections as shown in FIG. 4C, and the same processing is performed again, and the processing for filling the background between the two reflections is performed. As shown in FIG. 4D, it is possible to separate the reflected object and the background.

In addition, when the distance between the reflections when two or more reflections are captured in the line sensor image 11 can be predicted in advance, one segment or If the size of a section is set so that two or more reflections do not exist in adjacent sections, the reflection and the background can be separated by a single process, which is preferable. Specifically, when the division size of the section is B and the minimum distance between the reflections is L, the section may be divided so as to satisfy the following expression (1).
B <L / 2 (1)

  Through the above processing, a background-removed image 13 in which the brightness spots on the background are removed is obtained. The background removed image 13 is input to the discriminant analysis binarization processing unit 3c via the memory M2.

  The discriminant analysis binarization processing unit 3c performs binarization processing on the input background-removed image by the discriminant analysis binarization processing method to create a binarized image. Here, the discriminant analysis binarization processing method is a method of automatically setting a threshold according to the state of an image when performing binarization processing. The created binarized image is input to the noise removal processing unit 3d via the memory M2.

  The noise removal processing unit 3d creates a noise-removed image from the input binarized image by removing the flaws at the crossover portion included in the image, fine dot-like noise, and the like. The noise-removed image is input to the crossover edge detection unit 3e via the memory M2.

  For example, when searching for a certain line from the left with respect to the input noise-removed image, the crossover edge detection unit 3e detects a point that changes from white to black as the left edge point, and then from black to white. A process of detecting the changing point as the right edge point is performed. This process is performed for each line from the top to the bottom of the image. Information on the detected edge point (hereinafter referred to as edge data) is input to the crossover line height calculator 3f via the memory M2.

  The crossover line height calculation unit 3f is based on the input edge data, the distance between the left and right edge points input as parameters, the previously known crossover line size data, the lens focal length, the sensor width, and the number of sensor pixels. Calculate the height of the crossover. The calculated height of the crossover line is input to the crossover line deviation calculation unit 3g via the memory M2 as crossover line height data.

  The crossover part deviation calculation unit 3g receives the input crossover line height data, the center of gravity position of the left and right edge points input as parameters, the installation position of the line sensor camera 2, the lens focal length, the sensor width, and the number of sensor pixels. Calculate the displacement of the crossover from The calculated crossover deviation amount is recorded as crossover deviation amount data in the recording device 4 via the memory M2.

  In addition to the above-described crossover deflection amount data, the recording device 4 records line sensor images, background-removed images, binarized images, noise-removed images, edge data, crossover-line height data, and the like via memories M1 and M2. Is done.

  The flow of image processing in the measurement computer 3 of the crossover measuring apparatus using image processing according to the present embodiment will be described with reference to FIG.

  As shown in FIG. 5, in this embodiment, the measurement computer 3 first creates a line sensor image by arranging the image signals input from the line sensor camera 2 in time series in the line sensor image creation unit 3a. Next, the standard deviation background removal processing unit 3b performs the above-described standard deviation background removal processing on the line sensor image to create a background removed image (step P2).

  Thereafter, the discriminant analysis binarization processing unit 3c performs discriminant analysis binarization processing on the background removed image to create a binarized image (step P3), and then the noise removal processing unit 3d performs the binarized image. Is subjected to noise removal processing to create a noise-removed image (step P4).

  Subsequent to step P4, the crossover line edge detection unit 3e performs a process of extracting a boundary between white indicating the background of the noise-removed image and black indicating the crossover 7 as left and right edge points (step P5). Thereby, edge data is obtained.

Subsequent to Step P5, the jumper line height calculation processing unit 3f calculates the jumper line height as jumper line height data from the distance between the left and right edge points and the preset parameter described above (Step S5). P6) Subsequently, the crossover deflection amount is calculated from the crossover height data and the previously set parameters in the crossover deflection calculation processing unit 3g (step P7).
By performing the processing described above, the three-dimensional position of the crossover line can be measured.

  In the line sensor image, the sky as a background becomes a space with no sharpness when clouds and blue sky are reflected, whereas the trolley line 7 to be measured is clearly focused and bright and dark. In this embodiment, this is used to divide the image into arbitrary sections (in this embodiment, the first section 12a, the second section 12b, and the third section 12c) and take the standard deviation in the section to obtain the luminance. Check the degree of variation. At this time, if the luminance variation in the section is large, it is determined that the crossover 7 is reflected, and if the luminance variation is small, the background is determined. And about the area judged to be a background, the process which fills with white so that it may not affect subsequent processing is performed, and the erroneous detection of the crossover 7 is reduced by clarifying that this area is a background.

  Furthermore, in a section where no painting is performed, when the painting process is performed on the adjacent section, the painting process is also performed on the area from the neighboring section to the reflection in the section. By doing this, it is possible to fill the blank area (pixels that are background but not filled) between the painted section and the captured section of the section that is judged to have a reflected object, and more erroneous detection. Can be reduced. In the present embodiment, it is assumed that pixels having a luminance difference equal to or greater than a certain value compared to the average luminance value of the painted section are detected as the reflected object locations.

According to the above-described crossover measuring apparatus using image processing according to the present embodiment, the crossover line 7 and the background can be separated even when brightness spots are present on the background. Can be reduced.
In addition, even if it is determined that there is a reflection object, if the adjacent area is a painted area, the reflection object is also painted, so that the crossover line and the background can be more precisely separated.
Further, by photographing the crossover line 7 without illuminating it and displaying it on the screen in black, the crossover line 7 can be constructed with an inexpensive and simple apparatus configuration without using a device such as a sodium lamp and a bandpass filter. Daytime measurements can be made.

  In this embodiment, the standard deviation background removal processing unit 4f shows an example in which the area regarded as the background is painted black. However, the area regarded as the background is not limited to being painted black. For example, it is regarded as the background. It goes without saying that various changes can be made without departing from the spirit of the present invention, such as storing the region in the memory space as a mask pixel and making the mask pixel a pixel that will not be processed thereafter.

(Sunlight reflection judgment processing)
A second embodiment of the crossover measuring apparatus using image processing according to the present invention will be described with reference to FIGS. In the present embodiment, an erroneous measurement of the crossover due to sunlight reflection is prevented by “sunlight reflection determination processing” in which an image reflected by sunlight is determined as an error.

  In this embodiment, instead of the standard deviation background removal processing section 3b of the first embodiment shown in FIGS. 1 and 2, the background luminance average value calculation processing section 3h and the sunlight reflection determination as background segmentation processing means shown in FIG. A processing unit 3i is added. The other configuration is substantially the same as the configuration shown in FIGS. 1 and 2 and described above, and hereinafter, similar members will be denoted by the same reference numerals and redundant description will be omitted, and different points will be mainly described.

  As shown in FIG. 6, in this embodiment, the measurement computer 3 includes a line sensor image creation unit 3a, a discriminant analysis binarization processing unit 3c, a noise removal processing unit 3d, a crossover edge detection unit 3e, and a background luminance average value. The calculation processing unit 3h, the sunlight reflection determination processing unit 3i, the crossover line height calculation unit 3f, and the crossover line deviation calculation unit 3g are configured.

  The background luminance average value calculation processing unit 3h obtains the background luminance average value A from the input edge data and the background luminance totaling range W (see FIG. 9) input as a parameter. Here, the background luminance totaling range W is a fixed range set in advance in the outer direction of the crossover line of the edge point obtained by the crossover line edge detection unit 3e, and the background luminance average value A is totalized within the background luminance totaling range W. It is an average value of luminance values. The calculated background luminance average value A is input to the sunlight reflection determination processing unit 3i via the memory M2.

  The sunlight reflection determination processing unit 3i uses the fact that the sunlight reflection portion has a luminance value larger than that of the background, and the sum of the input background luminance average value A and a preset setting width α input as a parameter. Is used as a threshold value, and the luminance distributions near the edge points are compared. The determination result is input to the crossover line height calculation processing unit 3f via the memory M2.

The crossover line height calculation processing unit 3f knows in advance the distance between the input edge data and the left and right edge points that are input as parameters when the input determination result is that there is a sunlight reflection spot. The height of the crossover is calculated from the crossover size data, the lens focal length, the sensor width, and the number of sensor pixels. The calculated height of the crossover line is input to the crossover line deviation calculation unit 3g via the memory M2 as crossover line height data. On the other hand, if the input determination result is that there is no sunlight reflecting portion, the height of the crossover portion is not calculated, and is recorded in the recording device 4 via the memory M2 as missing data.
In addition, the crossover line which became the missing data is interpolated from normal crossover line width data of a portion determined to have no sunlight reflection after the measurement.

  The recording device 5 includes a line sensor image, a binarized image, a noise-removed image, a background luminance average value A, a setting width α, a determination result in the sunlight reflection determination processing unit, edge data, a crossover line height data, a crossover line. Deviation data and the like are recorded.

  The flow of image processing in the measurement computer 4 of the crossover measuring apparatus using the image processing according to this embodiment will be described with reference to FIGS.

  As shown in FIG. 7, in the measurement computer 3 according to the present embodiment, first, the line sensor image generator 3a arranges the image signals input from the line sensor camera 2 in chronological order. (Step P11). Subsequently, the discriminant analysis binarization processing unit 3c performs a discriminant analysis binarization process on the line sensor image to create a binarized image (step P12). Subsequently, the noise removal processing unit 3d performs noise removal processing on the binarized image to create a noise-removed image (step P13).

Subsequent to step P13, the crossover line edge detection unit 3d performs processing for extracting the boundary between white indicating the background and black indicating the crossover 7 as the left and right edge points 9L and 9R from the noise-removed image ( Step P14). Thereby, edge data is obtained.
Subsequent to step P14, the background luminance average value calculation processing unit 3h obtains a background luminance average value based on the edge data (step P15).

Here, details of the process of calculating the background luminance average value will be described with reference to FIGS.
For example, as shown in FIG. 8A, when there is no sunlight reflection spot in the line sensor image 11, the background luminance average value calculation processing unit 3h, as shown in FIG. Based on the edge data detected in 3e, the luminance values in the background luminance totaling range W, which is a fixed range in the outer direction of the crossover line 7, from the positions corresponding to the edge points 9L and 9R on the line sensor image 11 are totalized. The average value is obtained.
On the other hand, for example, as illustrated in FIG. 8B, when the sunlight reflection spot 8 is captured in the line sensor image 11, the background luminance average value calculation processing unit 3 h detects the crossover line edge detection unit 3 e. Based on the edge data, as shown in FIG. 9B, the luminance value within the background luminance totaling range W that is a fixed range in the outer direction of the crossover line 7 from the position corresponding to the edge points 9L and 9R on the line sensor image 11. To calculate the average value.
Thus, the background luminance average value A is obtained.

  Following step P15, the value obtained by adding the preset setting width α to the background luminance average value A in the sunlight reflection determination processing unit 3i is set as the threshold value S, while the line sensor image 11 has an edge for each line. The luminance distributions in the predetermined ranges on the left and right of the points 9L and 9R are compared (step P16). At this time, when the luminance distribution does not exceed the threshold value S as shown in FIG. On the other hand, if there is a portion exceeding the threshold value S as shown by a dashed line in FIG.

If it is determined in step P16 that there is no sunlight reflection, the crossover line height is calculated from the distance between the left and right edge points 9L and 9R and the preset parameter described above in the crossover line height calculation processing unit 3f. Is calculated as crossover line height data (step P17), and the crossover line deviation calculation processing unit 3g calculates the crossover line deviation amount from the crossover line height data and the above-described preset parameters (step S17). P18).
On the other hand, if it is determined in step P16 that there is sunlight reflection, the crossover line height calculation unit 3f and the crossover line knitting calculation unit 3g are not processed, but are recorded in the recording device 4 as missing data, and the measurement ends. The crossover line three-dimensional position is interpolated from the crossover line height data at the location where it was determined later that there was no sunlight reflection (step P19).

  With the above processing, the three-dimensional position of the crossover can be measured while preventing erroneous measurement due to sunlight reflection.

  According to the above-described crossover measuring apparatus using image processing according to the present embodiment, even when the line sensor image 11 is affected by sunlight reflection, the crossover 7 is measured with high accuracy. Can do. Further, by photographing the crossover line 7 without using illumination, the crossover line 7 is imaged in black on the line sensor image 11, so that an inexpensive and simple apparatus configuration without using a device such as a sodium lamp and a bandpass filter is used. The daytime measurement of the crossover line 7 can be performed.

(Crossover measurement)
The details of the third embodiment of the crossover measuring apparatus using image processing according to the present invention will be described with reference to FIGS. In this embodiment, the background luminance average value calculation processing section and the sunlight reflection determination processing section described in the second embodiment are added to the first embodiment described above. The other configuration is substantially the same as the configuration shown in FIGS. 1 and 2 and described above, and hereinafter, similar members will be denoted by the same reference numerals and redundant description will be omitted, and different points will be mainly described.

  As shown in FIG. 11, in the crossover measuring apparatus using image processing according to the present embodiment, the measurement computer includes a line sensor image creation unit 3 a, a standard deviation background removal processing unit 3 b, a discriminant analysis binarization processing unit 3 c, A noise removal processing unit 3d, a crossover line edge detection unit 3e, a background luminance average value calculation processing unit 3h, a sunlight reflection determination processing unit 3i, a crossing line height calculation unit 3f, and a crossing line part deviation calculation unit 3g It is prepared for.

  The recording device 4 includes a line sensor image, a background removal image, a binarized image, a noise removal image, edge data, a background luminance average value, a determination result in the sunlight reflection determination processing unit, a crossover line height data, a crossover line bias. Quantity data and the like are recorded.

  The flow of image processing by the measurement computer 4 of the crossover measuring apparatus using image processing according to the present embodiment will be described with reference to FIG.

  As shown in FIG. 12, in this embodiment, first, the line sensor image creation unit 3a performs a process of creating a line sensor image by arranging the image signals input from the line sensor camera 2 in time series. (Step P21) Subsequently, the standard deviation background removal processing unit 3b performs the above-described standard deviation background removal processing on the line sensor image to create a background removed image (Step P22).

  Thereafter, the discriminant analysis binarization processing unit 3c performs discriminant analysis binarization processing on the background removed image to create a binarized image (step P23), and then the noise removal processing unit 3d performs the binarized image. Is subjected to noise removal processing to create a noise-removed image (step P24).

  Subsequent to step P24, the crossover line edge detection unit 3d performs a process of extracting a boundary between white indicating the background and black indicating the crossover 7 as the left and right edge points from the noise-removed image (step P25). . Thereby, edge data is obtained.

  Subsequent to step P25, the background luminance average value calculation processing unit 3h obtains the background luminance average value A based on the edge data (step P26).

  Subsequent to Step P26, a value obtained by adding the preset setting width α to the background luminance average value A in the sunlight reflection determination processing unit 3i is set as the threshold value S, while the line sensor image is set for each line. The luminance distributions near the edge points are compared (step P27).

If it is determined in step P27 that there is no sunlight reflection, the crossover line height is calculated from the distance between the left and right edge points in the crossover line height calculation processing unit 3f and the preset parameter described above. It is calculated as height data (step P28), and the crossover deflection amount is calculated from the crossover height data and the previously set parameters in the crossover deflection calculation processing unit 3g (step P29).
On the other hand, when it is determined in step P27 that there is sunlight reflection, the crossover line height calculation unit 3f and the crossover line part calculation unit 3g are not processed, but are recorded in the recording device 4 as missing data, and the measurement ends. The crossover line three-dimensional position is interpolated from the crossover line height data at the location where it was later determined that there was no sunlight reflection (step P30).

  The three-dimensional position of the crossover line can be measured by the processing described above.

  In this embodiment, the operator selects whether or not the standard deviation background removal process and the standard deviation background removal process are necessary according to the shooting environment and shooting conditions (for example, setting the process ON / OFF in the setting file) It is preferable to be able to do this.

  Further, the crossover measuring device by image processing according to the present invention is not limited to the above-described first to third embodiments, and various modifications can be made without departing from the spirit of the present invention. Needless to say.

  The present invention is suitable for application to a crossover measuring apparatus using image processing.

DESCRIPTION OF SYMBOLS 1 Vehicle 2 Line sensor camera 3 Measurement computer 3a Line sensor image preparation part 3b Standard deviation background removal process part 3c Discriminant analysis binarization process part 3d Noise removal process part 3e Crossover part edge detection part 3f Crossover part height calculation Section 3g Crossover section deviation calculation section 3h Background luminance average value calculation processing section 3i Sunlight reflection determination processing section 4 Recording device 5 Pantograph 6 Trolley line 7 Crossover line 8 Sunlight reflection portion 11 Line sensor image 12a First section 12b First Two sections 12c Third section 13 Background removed image A Background brightness average value M1, M2 Memory S Threshold W Background brightness count range

Claims (8)

An image provided with a line sensor camera that images a crossover line on the roof of the vehicle so that its scanning line direction is perpendicular and horizontal to the traveling direction of the vehicle, and image processing means installed inside the vehicle In the crossover measuring device by processing,
The image processing means
Line sensor image creating means for creating a line sensor image in which image signals input from the line sensor camera are arranged in time series;
Discriminant analysis binarization processing means for creating a binarized image by performing discriminant analysis binarization processing on the input image;
Noise removal processing means for performing noise removal processing on the input image to create a noise removal image; and
A crossover edge detection means for detecting edge points on both sides of the crossover for the input image;
Background segmentation processing means for segmenting the connecting line and the background with respect to the input image;
A crossover line height calculating means for calculating a height of the crossover line from positions of edge points on both sides of the crossover line;
A crossover part deviation calculating means for calculating a deviation of the crossover from the height of the crossover;
With
The background separation processing means is
Based on the positions of the edge points on both sides of the crossover line, the background luminance average for calculating the background luminance average value by summing up the luminance values of a certain range on the background side based on the edge points for the input image. Value calculation processing means;
A threshold value is set based on the average value of the background luminance, and a predetermined range of luminance distributions centering on edge points on both sides of the crossover line of the input image is compared based on the threshold value. connecting wire measuring device according to images processed it characterized by comprising the sunlight reflecting determination processing means determines the presence or absence of light reflection. The discriminant analysis binarization processing means performs discriminant analysis binarization processing on the line sensor image to create the binarized image,
The noise removing unit performs noise removal processing on the binarized image to create the noise removed image,
The crossover edge detection processing means detects edge points on both sides of the crossover for the noise-removed image,
The background luminance average value calculation processing means calculates the background luminance average value using the line sensor image,
The sunlight reflecting determination processing unit connecting wire measuring apparatus using image processing according to claim 1, wherein the determining the presence or absence of the sunlight reflecting with said line sensor image. The image processing transition according to claim 1 or 2, wherein when the sunlight reflection determination processing means determines that there is sunlight reflection, the measurement result is interpolated based on previous and subsequent data after the measurement is completed. Line measuring device. An image provided with a line sensor camera that images a crossover line on the roof of the vehicle so that its scanning line direction is perpendicular and horizontal to the traveling direction of the vehicle, and image processing means installed inside the vehicle In the crossover measuring device by processing,
The image processing means
Line sensor image creating means for creating a line sensor image in which image signals input from the line sensor camera are arranged in time series;
Discriminant analysis binarization processing means for creating a binarized image by performing discriminant analysis binarization processing on the input image;
Noise removal processing means for performing noise removal processing on the input image to create a noise removal image; and
A crossover edge detection means for detecting edge points on both sides of the crossover for the input image;
Background segmentation processing means for segmenting the connecting line and the background with respect to the input image;
A crossover line height calculating means for calculating a height of the crossover line from positions of edge points on both sides of the crossover line;
A crossover part deviation calculating means for calculating a deviation of the crossover from the height of the crossover;
With
The background separation processing means is
The input image is divided into arbitrary sections, standard deviation background removal processing is performed for each of the sections to determine whether or not an object is captured in the section for each section, and the target is included in the section When an object is shown and the object is not shown in an adjacent section, the area from the adjacent section to the object is regarded as the background, and the object is not shown in the section When the object is shown in an adjacent section, a background removal image is generated by performing standard deviation background removal processing that regards the area from the object in the adjacent section to the section and the section as a background. Standard deviation background removal processing means ;
Based on the positions of the edge points on both sides of the crossover line, the background luminance average for calculating the background luminance average value by summing up the luminance values of a certain range on the background side based on the edge points for the input image. Value calculation processing means;
A threshold value is set based on the average value of the background luminance, and a predetermined range of luminance distributions centering on edge points on both sides of the crossover line of the input image is compared based on the threshold value. A sunlight reflection determination processing means for determining the presence or absence of light reflection,
The crossover line height calculation means calculates the height of the crossover line from the positions of the edge points on both sides of the crossover line when the sunlight reflection determination processing means determines that there is no sunlight reflection. connecting wire measuring device according to images processed you characterized. When there is an area that is not regarded as a background in the section in which the standard deviation background removal processing unit has performed the standard deviation background removal processing, the section is subdivided and standardized again for each of the subdivided sections. A deviation background removal process is performed to determine whether or not the object is shown in the section for each of the subdivided sections, and the object is shown in the section and the section is adjacent to the section. When the object is not shown, the area from the adjacent section to the object is regarded as the background, and the object is shown in the adjacent section when the object is not shown in the section. the image processing according to claim 4, wherein creating a background subtraction image by performing the standard deviation background removal processing regarded as background between regions and said section to between the compartment from the object of the adjacent section if it is That a crossover measuring device. 5. The crossover measuring apparatus by image processing according to claim 4 , wherein the size of the arbitrary section is such that a section where the object exists is not continuous. In any one of claims 4 to 6 when it is determined that there is sunlight reflecting characterized by interpolating the measurement results based on the data before and after the after end of measurement in the sunlight reflecting determination processing unit Trolley wire wear measurement by the described image processing. The standard deviation background removal processing unit performs a standard deviation background removal process on the line sensor image to create the background removed image in which the background in the line sensor image is removed,
The discriminant analysis binarization processing means performs discriminant analysis binarization processing on the background removed image to create the binarized image,
The noise removing unit performs noise removal processing on the binarized image to create the noise removed image,
The crossover edge detection processing means detects edge points on both sides of the crossover for the noise-removed image,
The background luminance average value calculation processing means calculates the background luminance average value using the line sensor image,
The crossover measurement by image processing according to any one of claims 4 to 7, wherein the sunlight reflection determination processing means determines the presence or absence of the sunlight reflection using the line sensor image. apparatus.