JP5175528B2 - Tunnel lining crack inspection system - Google Patents

Description

  The present invention relates to a technique for inspecting whether or not a crack has occurred in a tunnel lining with high accuracy and shortening the inspection cycle.

A conventional tunnel photography vehicle is a vehicle in which four CCD line sensor cameras are mounted on a railroad vehicle (maintenance vehicle that can travel on both roads and tracks) and covers the tunnel at about 10 km / h while traveling in the tunnel. The entire inner wall of the craft is photographed continuously in a plane. By scanning at intervals of 1 mm in the circumferential direction and 1 mm in the traveling direction, cracks are detected with a resolution of 1 mm / pixel or more. The composition of the image data photographed by each camera is automatically processed after the position of the image photographed by each camera is manually adjusted. In the synthesized image, the state of the tunnel wall surface and deformation such as cracks are written on the screen by a pen touch method by hand.
JP 2007-26255 A Japanese Patent Laid-Open No. 07-0667101 Japanese Patent Laid-Open No. 02-236787

However, the crack detection method using the tunnel photographing vehicle as described above has the following problems.
(B) Since the tunnel wall surface is photographed at a speed of about 10 km / h, the inspection period by photographing is about once every two years for photographing outside the business hours. May be delayed.

(B) Since the image data photographed by each camera is manually combined, there is a problem that a lot of labor and time are required for the work by the worker.
(C) There has been a problem in that deformations such as cracks are manually extracted from the photographed image, and enormous labor and time are required for the work by the worker. In addition, in order to manually check whether or not the deformed portion has progressed, there is a problem that it takes a lot of labor and time for the work by the worker.

In order to solve such a problem, there is a method of inspecting the presence or absence of the development of cracks generated in the tunnel lining using image processing as in Patent Documents 1 to 3 below.
For example, as exemplified in Patent Document 1, a method of extracting the progress of deformation points such as cracks from two images with different shooting times, such as a tunnel wall surface, is also conceivable. Specifically, in such an extraction method, the edge of the image is extracted based on the first actual image data of the measured surface, and the second actual image after a predetermined time has elapsed on the same measured surface. A step of extracting an edge of the image based on the data, a difference between the specific pixel in the second edge image, the pixel in the first edge image corresponding to the specific pixel and a peripheral pixel of the pixel, It comprises a difference processing step for outputting the smallest difference value as a difference value, and a step for outputting the difference obtained in this difference processing step as an advanced deformation point. Note that straight lines other than the deformed point are extracted and deleted by the Hough transform.

  However, in the crack extraction method described in Patent Document 1, the image quality is different because the timing of capturing the actual image data of the measured surface is different, and noise is present in the difference value extracted from the two actual image data. It is not preferable because there is a problem that the progress of the included transformation point is incorrect.

  Further, as exemplified in Patent Document 2, a method of accurately detecting an abnormality of a device accompanied by a visual state change using image information of the captured device is also conceivable. Specifically, in such an extraction method, the edge portion of the captured image of the monitoring target region is extracted by the edge extraction portion, and this edge portion is used as a mask image. Next, a difference image between the picked-up image and the picked-up image of the monitoring target area taken at a time from when the picked-up image is taken is obtained by a calculation unit, and the difference image is masked with the mask image by a logic calculation unit. To do. The difference image obtained as a result is binarized by a binarization threshold value by a logic operation unit to obtain a binarized image, and the abnormality determination unit determines presence / absence of abnormality from the size of the area of the binarized image. .

  However, in the crack extraction method described in Patent Document 2, the image quality may be different between the mask image and the photographed image, and the extracted difference value includes noise and is output as an abnormal point. There is a problem that the progress of this is inaccurate.

  Further, as exemplified in Patent Document 3, the difference value of the captured image obtained by capturing the monitoring target at a predetermined time interval is cumulatively added for each pixel, and the contour line that forms the background from the obtained cumulative added image In addition, the cumulative addition image after the contour line removal is projected in the moving direction of the monitoring object and the direction perpendicular to the moving direction, and the state change pixel is set using a predetermined threshold value for each direction. An extraction method is also conceivable.

  However, in the crack extraction method described in Patent Document 3, the image quality is different because the time when the captured image is captured differs, and noise is included in the difference value extracted from the two captured images. There is a problem that the state change pixel is inaccurate, which is not preferable.

  The present invention has been made in view of such problems, and the purpose of the present invention is to highly accurately inspect the presence or absence of cracks occurring in the tunnel lining and shorten the inspection cycle. The purpose is to improve the safety of tunnel lining against concrete peeling.

The tunnel lining crack inspection apparatus according to claim 1, which has been made to solve the above-described problem, is a tunnel lining crack inspection apparatus that inspects the presence or absence of the development of cracks generated on the inner wall of the tunnel lining, An imaging unit that continuously captures the entire circumference of the inner wall in a plane, a storage unit that stores a captured image captured by the captured image, a crack extracted from the captured image stored in the storage unit, and the storage unit Or at least one of the captured image stored by the storage means and the reference image used as a reference when determining the progress of cracks in the inner wall. And performing image correlation to extract crack differences, and based on the extracted differences, image processing for inspecting whether or not the inner wall crack has progressed. And means, wherein the image processing means, said extracting an image of the object that are regularly arranged on the inner wall of the tunnel lining, and deletes the extracted image from the captured image from the captured image, Estimating that the thin and long one is likely to be cracked, and extracting the crack from the photographed image by comparing the shape of the general crack to be memorized with the shape of the crack, and the image processing means It is characterized by inspecting whether or not the inner wall has cracks from the tendency of the progress of the difference based on three or more extraction results .

  According to the tunnel lining crack inspection apparatus of the present invention configured as described above, it is possible to inspect the progress of cracks generated in the tunnel lining with high accuracy and to shorten the inspection cycle. Therefore, it is possible to improve the safety against the concrete peeling of the tunnel lining. In addition, since photography can be performed during business hours, the inspection cycle is shortened in this respect, and safety is increased. Moreover, it is not necessary to prepare a dedicated vehicle for extracting cracks in the tunnel lining, and an existing vehicle can be used.

In this case, it is conceivable that, as in claim 2 , the image processing means can additionally register a general crack shape. If comprised in this way, the precision which test | inspects the presence or absence of the progress of a crack can be improved more.

Further, as in claim 3 , it is conceivable that the image processing means performs processing for emphasizing the extracted crack. By emphasizing the cracks extracted in this way, the boundary of the cracked portion becomes clearer, and the progress of the crack can be easily inspected.

Further, as described in claim 4 , when a crack is extracted from the photographed image, the image processing means performs the above-described image correlation. On the other hand, when a crack is not extracted from the photographed image, the image processing means It can be considered not to perform the above-described image correlation. In this way, by not performing extra processing, the processing load on the image processing means can be reduced, and the processing time of the entire image processing can be shortened.

By the way, as described above, when inspecting the presence or absence of the progress of cracks from the extraction results of the differences, the effect can be obtained by using the extraction results twice. When the presence / absence of progress is inspected, there is a possibility that the photographed image may appear to have progressed or may be shrunk. In view of this, it is conceivable that the image processing means inspects whether or not the crack has progressed from the tendency of the difference in the inner wall based on the extraction results three or more times. In this way, for example, when the presence or absence of crack progress is inspected with two images, the photographed image appears to have progressed depending on the case, or conversely appears to be shrunk. Therefore, the accuracy of inspecting whether or not cracks have progressed can be increased.

Embodiments of the present invention will be described below with reference to the drawings.
[First embodiment]
FIG. 1 is a schematic configuration diagram of a crack inspection apparatus 1 for tunnel lining.

[Description of the configuration of the tunnel lining crack inspection apparatus 1]
As shown in FIG. 1, a tunnel lining crack inspection apparatus 1 is an apparatus that is mounted on a railway vehicle and extracts cracks generated on the inner wall of the tunnel lining.

  Specifically, the tunnel lining crack inspection apparatus 1 includes a camera 10, a camera power supply 20, a control device 30, a hard disk (HDD) 40, a metal halide illumination 50, and a light guide 60. The camera 10 and the HDD 40 are connected to the control device 30.

  The camera 10 has a photographing element such as a CCD, continuously shoots the entire inner wall of the tunnel lining from a vehicle on which the tunnel lining crack inspection device 1 is mounted, and the captured image is transferred to the control device 30. Send. Note that a fisheye lens is used for the camera 10 because the side window of the vehicle traveling in the tunnel lining and the inner wall of the tunnel lining are close to each other. For this reason, the resolution tendency of the end portion is significantly lower than that at the center of the angle of view, but the resolution of 1.5 mm / pixel is obtained in the direction perpendicular to the line at the end of the angle of view. As a result of measuring the distance from the vehicle window to the inner wall surface of each tunnel lining in advance, the distance to the inner wall surface of the tunnel lining is the depth of field in all tunnels by reducing the aperture to F4 in this embodiment. The focus of the lens is fixed because it can be confirmed that the lens fits inside.

  Further, the camera 10 is driven by a rotary encoder pulse attached to the axle of the vehicle in order to prevent the captured image from being distorted in the track direction due to a difference in train speed at the time of shooting. In addition, this rotary encoder pulse generates one pulse every time the vehicle moves 1.5 mm. Further, the maximum line rate of the camera 10 is 68 kHz, and a 1.5 mm pitch scan is possible even when the vehicle travels at 270 km / h.

The camera power supply 20 supplies power to the camera 10. In the present embodiment, the camera power supply 20 supplies DC 12V power to the camera 10.
The control device 30 includes a grabber board 31, a motherboard 32, and a SCSI RAID card 33. Since each component is in accordance with a known technique, detailed description thereof is omitted here.

  The control device 30 includes a well-known CPU (not shown), ROM, RAM, I / O as an input / output circuit, a bus line connecting these components, a signal processing circuit for processing a signal from an input operation, A signal output circuit for controlling the display unit is provided. The CPU performs control according to control programs and data stored in the ROM and RAM. The ROM has a program storage area and a data storage area. A control program is stored in the program storage area, and data necessary for the operation of the control program is stored in the data storage area. In addition, the control program operates on the RAM in a form in which the work memory is a work area.

Moreover, the control apparatus 30 analyzes the picked-up image received from the camera 10 by performing the crack inspection process mentioned later, and performs an image process.
The HDD 40 is composed of a nonvolatile memory and is used for storing various data. Note that data is stored in the HDD 40 using the Raid0 method in order to enable high-speed processing.

The metal halide illumination 50 is illumination means for photographing with the camera 10, and the irradiation light is guided by the light guide 60.
The camera 10 corresponds to a photographing unit. The control device 30 corresponds to an image processing unit. The HDD 40 corresponds to a storage unit.

[Description of crack inspection processing]
Next, a crack inspection process executed by the control device 30 of the tunnel lining crack inspection apparatus 1 will be described with reference to the flowchart of FIG.

  First, in step (hereinafter S) 1, image alignment is performed. Specifically, the position of the image is adjusted by searching for the area where the cutout image is the smallest by comparing the gray levels between the current captured image and the previous captured image or the reference captured image.

  First, as illustrated in FIG. 3A, each image is divided into a set of square cells (pixels). Then, the upper left pixel of the image is the origin, the vertical pixels are y = 0, 1, 2,... From the top, and the horizontal pixels are x = 0, 1, 2,.

  Then, the average value of the shade of the entire image is calculated. For example, the average value of the intensity of the first image is “2”, and for example, the average value of the intensity of the second image is “7” (see FIG. 3A).

Subsequently, the average value of the intensity of the entire image is divided from the intensity of each pixel of the image (see FIG. 3B). As a result, the uneven brightness at the time of shooting is removed.
Subsequently, in the two images, a part having the same characteristics is searched and specified. That is, in order to compare the shades of two images under the same conditions in which the above-described brightness deviation is eliminated, the shade of the same position of the second image is divided from the shade of the first image. FIG. 3C shows a case where the calculation is performed with the pixel at the position (y = 1, x = 1) as the center. Then, a value obtained by squaring and adding the calculated shades of each pixel is set as a value of the position. In the example of FIG. 3C, the calculation is as shown in the following equation (A).

なお、比較するピクセルが存在しない場合には、仮に値「０」として計算を行う（図３（ｄ）および図３（ｅ）参照）。この図３（ｅ）の例では次の式（Ｂ）のような計算となる。

If there is no pixel to be compared, the calculation is performed assuming that the value is “0” (see FIG. 3D and FIG. 3E). In the example of FIG. 3E, the calculation is as shown in the following equation (B).

以上のように全ピクセルについて値を計算する（図３（ｆ）参照）。

As described above, values are calculated for all the pixels (see FIG. 3F).

  In order to search for a part having the same feature, the center (x = 1, y = 1) of the second image is sequentially moved to each pixel of the first image, and the second image and When only overlapping pixel portions are compared, as a result, it is possible to specify that the pixel having the smallest difference in density is centered on the closest feature. In this case, as a result of performing the comparison by shifting the pixels one pixel at a time in the x direction and the y direction, the case where the center of the first image (x = 1, y = 1) is the second image (See FIG. 4 (a) and FIG. 4 (b)). And when it is set as deviation | shift amount (dx, dy), the deviation | shift amount (dx, dy) which makes the variable W the minimum value is calculated in following Formula (C) (refer FIG.4 (c)).

続くＳ２では、特徴抽出アルゴリズムにより、撮影画像中に存在する形状・大きさ等の特徴のある箇所をパターンマッチング手法により抽出して削除する。この際、ケーブルの輝度が緩やかに変化することに着目して、ケーブルやパネルの継目、照明などのトンネル内で規則的に配置されている物体を排除する。そして、細くて長いものを「ひび割れらしきもの」と推定して残す。

In subsequent S2, a feature extraction algorithm extracts and deletes a portion having a feature such as shape and size existing in the captured image by a pattern matching method. At this time, paying attention to the gradual change in the luminance of the cable, the objects regularly arranged in the tunnel such as the joint of the cable and the panel and the illumination are excluded. Then, the thin and long object is presumed to be a "cracked object" and left.

  Below, an example is given and demonstrated about a feature extraction algorithm. Focusing on the vertical and horizontal lines, the joints of the concrete panel are extracted (see FIG. 5A), and the vertical and horizontal lines are erased (see FIG. 5B). In addition, for a cable as illustrated in FIG. 5C, a portion having an extremely large differential value of an image is extracted by paying attention to the fact that one of the upper boundary, the lower boundary, and the center line of the cable shines strongly. Then (see FIG. 5D), the extracted portion is expanded (see FIG. 5E). Then, the expanded area is erased. In addition, with respect to the metal fittings illustrated in FIG. 6 (a), a portion that shines strongly is extracted by focusing on the fact that the metal fittings partially shine (see FIG. 6 (b)), and the extracted portion is expanded. (See FIG. 6 (c)). Then, the expanded area is erased. For moss and dirt illustrated in FIG. 7 (a), a portion darker by the threshold value than the average luminance in the horizontal direction is extracted by paying attention to the fact that the illuminance becomes dark over a wide area in the part where the moss or dirt is attached. Then (see FIG. 7B), the extracted portion is contracted and then expanded to extract a region of a certain area or more (see FIG. 7C). Then, the extracted portion is expanded (see FIG. 7D). Then, the expanded area is erased.

  In subsequent S3, digital filter processing is executed. Specifically, objects that are regularly arranged in a tunnel, such as cables, construction seams by placing concrete that is generated at equal intervals, and lighting, are excluded by a digital filter. Then, the thin and long object is presumed to be a "cracked object" and left.

  Below, an example is given and demonstrated about a digital filter process. A repetitive pattern and noise are erased from the image using the two-dimensional FIR filter represented by the following equation (D) from the image as illustrated in FIG. 8A (see FIG. 8B).

なお、Ｓ２の特徴抽出アルゴリズムおよびＳ３のデジタルフィルタが互いに補完し合うことで、規則的に配置されている物体をより効果的に排除するとともに、「ひび割れらしきもの」をより効果的に推定する。

Note that the feature extraction algorithm in S2 and the digital filter in S3 complement each other, so that the regularly arranged objects are more effectively removed and “probable cracks” are more effectively estimated.

  In subsequent S4, cracks are extracted using a learning function based on artificial intelligence such as a neural network or a support vector machine. Here, the shape of a general crack is stored, and “crack” is extracted in comparison with the “like crack” in the image. Note that the crack-shaped pattern can be additionally registered. Here, if no crack is extracted from the captured image by the learning function of S4, the present process is terminated without executing the subsequent steps. On the other hand, if a crack is extracted from the captured image by the learning function in S4, the process proceeds to S5.

  In S5, line segment enhancement filter processing is executed. Specifically, in this line segment enhancement filter process, only the line segment having a certain range of thickness in the image as illustrated in FIG. 9A is emphasized to emphasize the crack extracted in S4 ( (See FIG. 9B).

  In subsequent S6, image correlation is executed. Specifically, so-called dot matching is performed, and an image correlation is achieved between the current captured image (see FIG. 10A) and the previous captured image or the reference captured image (see FIG. 10B). Alternatively, the positional deviation correction is performed based on the image correlation between the front and rear images, and the difference is extracted and calculated using the cross-correlation function (see FIG. 10C). At this time, the difference is emphasized, and the outline of the crack is matched while ignoring the contrast of the entire image.

Note that the binarization process (S11) may be executed instead of S6. Information other than necessary information can be deleted by binarization.
In subsequent S7, the cracks are connected by executing the expansion process.

In subsequent S8, the cracking process is connected by executing the contraction process.
In subsequent S9, the vectorization process is executed to digitize the crack area, length, curvature, fractal dimension, and the like.

  In subsequent S10, the numerical value is graphed and the developed crack is read. In the present embodiment, the progress of cracking is determined using the extraction results of three or more times. This is due to the following reason. In other words, it is still possible to confirm the presence or absence of crack growth from the tendency of crack propagation using the two extraction results. In this way, the presence or absence of crack growth is inspected using two images. Then, the photographed image may appear to be developed depending on the case or may be shrunk on the contrary. Therefore, in the present embodiment, the presence or absence of crack progress is confirmed from the tendency of the progress of differences based on the extraction results three times or more (see FIG. 11). FIG. 11 (a) shows a case where cracks are progressing, and FIG. 11 (b) shows a case where cracks are not progressing. Moreover, you may confirm the presence or absence of a crack progress from the tendency of the progress of a crack using the extraction result 4 times or more. The graph may be displayed on a display device (not shown).

Then, this process ends.
[About test results]
FIG. 12 is a table in which the horizontal axis is the stripe interval and the vertical axis is the luminance difference of the black and white portion of the stripe pattern from the captured images of the six resolution charts affixed to the inner wall of the tunnel lining in advance. In order to automatically extract cracks existing in the photographed image, it has been separately found that a luminance difference of 25 or more is necessary. Therefore, the resolution of the resolution chart corresponding to the luminance difference 25 is 2.5 mm to It can be seen that it is between 3.0 mm. The figure includes data having a larger luminance difference than the others, which is due to the resolution chart paper having a high reflectance. This result is considered to suggest the possibility that the resolution is improved by further increasing the illuminance.

  In addition, in order to determine whether or not the progress of cracks can be extracted by overlaying the images after automatic crack extraction, the following examination was conducted to determine whether or not the vibration of the vehicle would affect. . FIG. 13 shows the result of performing Fourier transform by taking the coordinates of the component orthogonal to the traveling direction caused by the vibration of the vehicle with respect to the linear object in the traveling direction of the train with a cable or the like included in the captured image. The vertical axis represents the intensity of vibration, and the horizontal axis represents the vibration period. From this figure, it can be seen that the influence of the vibration of the vehicle is concentrated on the long period side longer than 30 mm. Note that long-period vibration is separately found to be relatively easy to process at the stage of image analysis, and the influence of train vibration on the captured image is reduced.

[Effect of the first embodiment]
(1) As described above, according to the tunnel lining crack inspection apparatus 1 of the first embodiment, cracks are extracted from captured images, and positions are aligned between consecutively captured images and image correlation is performed. The difference between cracks is extracted, and based on the extracted difference, the presence or absence of progress of cracks on the inner wall of the tunnel lining is inspected. As a result, it is possible to inspect the progress of cracks occurring in the tunnel lining with high accuracy and shorten the inspection cycle. Therefore, it is possible to improve the safety against the concrete peeling of the tunnel lining. In addition, since photography can be performed during business hours, the inspection cycle is shortened in this respect, and safety is increased. Moreover, it is not necessary to prepare a dedicated vehicle for extracting cracks in the tunnel lining, and an existing vehicle can be used.

  (2) Further, according to the tunnel lining crack inspection apparatus 1 of the first embodiment, the feature extraction algorithm is used to extract a characteristic part such as shape and size existing in the photographed image by the pattern matching method. (S2), and objects that are regularly arranged in the tunnel, such as cables, construction seams by placing concrete that are generated at equal intervals, and lighting, are excluded by a digital filter (S3). As a result, it is possible to further improve the accuracy of inspecting the presence or absence of the progress of cracks.

  (3) Moreover, according to the tunnel lining crack inspection apparatus 1 of the first embodiment, a crack location is extracted using a learning function based on artificial intelligence such as a neural network or a support vector machine (S4). Specifically, the shape of a general crack is stored, and “crack” is extracted in comparison with “a crack-like thing” in an image. As a result, it is possible to further improve the accuracy of inspecting the presence or absence of the progress of cracks.

  (4) Moreover, according to the crack lining inspection apparatus 1 of the tunnel lining of the first embodiment, since it is possible to additionally register the crack shape pattern used for the learning function by the artificial intelligence described above, the progress of the crack The accuracy of inspecting the presence or absence can be further increased.

  (5) Further, according to the tunnel lining crack inspection apparatus 1 of the first embodiment, the process of emphasizing the extracted crack is performed by the line segment emphasis filter process (S5). By emphasizing the cracks extracted in this way, the boundary of the cracked portion becomes clearer, and the progress of the crack can be easily inspected.

  (6) According to the tunnel lining crack inspection apparatus 1 of the first embodiment, when no crack is extracted from the photographed image by the learning function of S4, the present process is terminated without executing the subsequent steps. On the other hand, if a crack is extracted from the captured image by the learning function in S4, the process proceeds to S5 to execute a line segment enhancement filter process. Thus, by not performing extra processing, the processing burden on the image processing means can be reduced, and the processing time of the entire image processing can be shortened.

  (6) Moreover, according to the tunnel lining crack inspection apparatus 1 of the first embodiment, in the process of S10, the presence or absence of crack progress is confirmed from the tendency of the progress of differences based on the extraction results three times or more. . This makes it possible for the photographed image to appear to have developed in some cases, or to appear to be shrunk on the contrary, for example, when inspecting the presence or absence of crack growth in two images. And the accuracy of inspecting the presence or absence of crack growth can be increased.

[Other Embodiments]
As mentioned above, although one Embodiment of this invention was described, this invention is not limited to the said embodiment, It is possible to implement in the following various aspects.

  (1) In the above embodiment, in the crack inspection process, the image alignment process (S1), the feature extraction algorithm process (S2), the digital filter process (S3), the learning function (S4), and the line segment enhancement filter process (S5) And the image correlation process (S6 or binarization process (S11)) are executed in this order, but the present invention is not limited to this, and as shown in FIG. 14, these processes are executed in other orders. It may be. For example, as illustrated in FIG. 14A, feature extraction algorithm processing, digital filter processing, learning function, line segment enhancement filter processing, image registration processing, image correlation processing, and binarization processing are executed in this order. And so on. Various other orders are possible.

It is a schematic block diagram of the crack inspection apparatus of a tunnel lining. It is a flowchart explaining a crack test process. It is explanatory drawing (1) explaining position alignment of an image. It is explanatory drawing (2) explaining position alignment of an image. It is explanatory drawing (1) explaining a feature extraction algorithm. It is explanatory drawing (2) explaining a feature extraction algorithm. It is explanatory drawing (3) explaining a feature extraction algorithm. It is explanatory drawing explaining a digital filter process. It is explanatory drawing explaining a line segment emphasis process. It is explanatory drawing explaining an image correlation process. It is a table | surface which shows the progress of a crack, (a) shows the case where the crack has progressed, (b) shows the case where the crack has not progressed. It is a table | surface which made the horizontal axis | shaft the space | interval of a striped pattern from the picked-up image of the six resolution charts previously stuck on the inner wall of the tunnel lining, and made the brightness | luminance difference of the black-and-white part of a striped pattern vertical axis | shaft. The result of performing Fourier transform by taking the coordinates of the component orthogonal to the traveling direction caused by the vibration of the vehicle with respect to the linear subject in the traveling direction of the train with a cable or the like included in the captured image is shown. It is a flowchart which shows other embodiment of a crack test | inspection process.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Tunnel lining crack inspection apparatus, 10 ... Camera, 20 ... Camera power supply, 30 ... Control device, 31 ... Grabber board, 32 ... Mother board, 33 ... Raid card, 40 ... HDD, 50 ... Metal halide lighting, 60 ... Light guide

Claims (4)

A tunnel lining crack inspection device for inspecting the presence or absence of cracks generated on the inner wall of the tunnel lining,
Photographing means for continuously photographing the entire circumference of the inner wall in a plane;
Storage means for storing a photographed image photographed by said photographing means,
When extracting cracks from the captured images stored in the storage unit and determining the progress of cracks in the captured images stored in the storage unit or between the captured images stored in the storage unit and the inner wall At least one of the reference images to be used as a reference is aligned and image correlation is performed to extract crack differences, and based on the extracted differences, whether or not the inner wall cracks have progressed is determined. Image processing means for inspection;
Equipped with a,
The image processing means extracts an image of an object regularly arranged on the inner wall of the tunnel lining from the captured image, deletes the extracted image from the captured image, and cracks a thin and long object. Estimate that it looks like it, and extract the crack from the captured image by comparing the shape of the general crack to be stored with the shape of the crack that seems to be cracked,
Further, the image processing means inspects whether or not the inner wall is cracked from the tendency of the progress of the difference based on the extraction results three times or more, and the crack lining inspection apparatus for tunnel lining.   In the crack inspection apparatus for tunnel lining according to claim 1,
  The tunnel lining crack inspection apparatus, wherein the image processing means is capable of additionally registering a general crack shape.   In the crack inspection apparatus for tunnel lining according to claim 1 or claim 2,
  The crack inspection apparatus for tunnel lining, wherein the image processing means performs processing for emphasizing the extracted crack.   In the crack inspection apparatus for tunnel lining according to any one of claims 1 to 3,
  The image processing means performs the image correlation when a crack is extracted from the photographed image, and does not perform the image correlation when a crack is not extracted from the photographed image. Tunnel lining crack inspection equipment.

Images