JP6905956B2 - Crack change determination device and method - Google Patents

Description

The present invention relates to a crack change determination device and a method, and more particularly to a crack change determination device and a method for determining a change in a crack in a structure.

Examples of techniques for detecting cracks in structures include Patent Document 1 and Patent Document 2. Patent Document 1 discloses a technique for color-coding a cracked line segment with a color according to the degree of cracking of the cracked line segment extracted by the feature point treatment. The crack is binarized, and the drawn line after binarization is represented as a line having the same thickness as a thick line and a thin line.

Patent Document 2 discloses a technique for detecting crack elongation in two images recorded in different periods by a two-dimensional convolutional neural network (CNN). Patent Document 2 uses trained CNNs to identify changes in a structure regardless of the presence of image differences that are independent of changes to the structure.

Patent Document 3 discloses a technique for estimating the width of cracks in a structure.

Japanese Unexamined Patent Publication No. 2001-099784 JP-A-2017-062776 Japanese Unexamined Patent Publication No. 2017-227466

When detecting changes in cracks compared to images taken at different times (for example, after 5 years) even for the same structure, the shooting conditions of the image (shooting position, shooting angle, focus, weather conditions, digital camera model) There is a problem that the same crack cannot be identified as the same because the shade of the cracked part of the image changes due to a slight change in performance.

For example, when comparing a crack photographed at a certain time point as shown in FIG. 8 and a crack photographed at the same place at another time point as shown in FIG. 9, the same crack has different shades of pixels. Comparing these crack images, it may not be possible to accurately detect changes in the same crack.

The present invention has been made in view of such problems, and an object of the present invention is to provide a crack change determination device and a method capable of easily and quickly determining a change in the length direction of a crack in a structure. do.

The crack change determination device according to the first aspect of the present invention includes a first image acquisition unit that acquires a first image taken with the surface of the structure as a subject, and a first image acquisition unit that takes the surface of the structure as a subject and has the first aspect. It has a start point, end point, and length of cracks on the surface of the structure from the second image acquisition unit that acquires the second image taken after the first image and the first image acquired by the first image acquisition unit. From the first crack information detection unit that detects the first crack information, which is the vectorized information of the line segment, and the second image acquired by the second image acquisition unit, the start point, end point, and length of the crack on the surface of the structure. The second crack information detection unit that detects the second crack information, which is the information obtained by vectorizing the lines having a sword, and the first, which is the information obtained by processing the shape of the crack of the structure with respect to the first crack information for comparison. The first crack comparison information creation unit that creates crack comparison information and the second crack comparison information creation that creates the second crack comparison information, which is the information obtained by processing the shape of the crack of the structure for comparison with the second crack information. Changes in cracks on the surface of the structure according to the overlap between the part, the first crack comparison information created by the first crack comparison information creation unit, and the second crack comparison information created by the second crack comparison information creation unit. It is provided with a crack change determination unit for determination.

In the crack change determination device according to the second aspect of the present invention, the first crack comparison information is information in which the crack line segment is rectangularized by giving the crack width of the structure to the first crack information. The second crack comparison information is information in which the line segment of the crack is rectangularized by giving the width of the crack of the structure to the second crack information.

The crack change determination device according to the third aspect of the present invention includes a first crack width estimation unit that estimates the width of cracks on the surface of the structure from the first crack information, and a crack on the surface of the structure from the second crack information. A second crack width estimation unit for estimating the width of the structure, and a first crack comparison information creation unit assigns a crack width on the surface of the structure based on the estimated crack width of the first crack width estimation unit. Then, the second crack comparison information creation unit assigns the width of the crack on the surface of the structure based on the width of the crack estimated by the second crack width estimation unit.

In the crack change determination device according to the fourth aspect of the present invention, the first crack width estimation unit estimates the crack width on the surface of the structure based on the concentration distribution in the width direction of the first crack information, and the second crack width is estimated. The crack width estimation unit estimates the crack width on the surface of the structure based on the concentration distribution in the width direction of the second crack information.

In the crack change determination device according to the fifth aspect of the present invention, the first crack comparison information creation unit expands the width of cracks on the surface of the structure estimated by the first crack width estimation unit by dilation processing, and the first crack width is increased. 2 The crack comparison information creation unit expands the width of cracks on the surface of the structure estimated by the second crack width estimation unit by dilation processing.

In the crack change determination device according to the sixth aspect of the present invention, the first crack comparison information creating unit expands the width of the crack on the surface of the structure according to the density of the crack on the surface of the structure, and the second crack. The comparative information creation unit expands the width of cracks on the surface of the structure according to the density of cracks on the surface of the structure.

In the crack change determination device according to the seventh aspect of the present invention, the crack change determination unit compares the first crack comparison information with the second crack by local projective conversion from one of the first image and the second image to the other. After aligning with the information, it is determined whether or not the length of the crack on the surface of the structure has changed.

In the eighth aspect of the present invention, the projective transformation includes an affine transformation.

The crack change determination method according to the ninth aspect of the present invention includes a step in which the computer acquires a first image taken with the surface of the structure as the subject, and a first image taken with the surface of the structure as the subject. The step of acquiring the second image taken after the image, and the first crack information which is the vectorized information of the line having the start point, the end point, and the length of the crack on the surface of the structure from the first image. The step of detecting the second crack information, which is the vectorized information of the line having the start point, the end point, and the length of the crack on the surface of the structure from the second image, and the first crack information. The step of creating the first crack comparison information, which is the information obtained by processing the shape of the crack of the structure for comparison, and the information obtained by processing the shape of the crack of the structure for comparison with respect to the second crack information. 2. The step of creating the crack comparison information and the step of determining the change of the crack on the surface of the structure according to the overlap of the first crack comparison information and the second crack comparison information are executed.

An aspect of the present invention also includes a crack change determination program for causing a computer to execute the above crack change determination method. A computer-readable recording medium (non-temporary tangible medium) on which a crack change determination program is recorded is also included in the aspect of the present invention.

According to the present invention, the width of cracks in images taken at different times is processed, and the change in the length direction of the cracks is determined from the overlap of these cracks. Since the width of the crack is processed, the probability that the change in the width direction of the crack is erroneously determined as the change in the length direction of the crack is reduced, and the change in the length direction of the crack can be determined more accurately.

Block configuration diagram of crack change determination device Diagram that conceptually shows how to estimate the arrangement relationship of cracks The figure which shows an example of the start point, end point, length of the detected crack vector, and the width of the crack estimated corresponding to each crack vector. The figure which shows an example of the crack vector in the partial region R1 of the 1st image I1 and the crack vector in the partial region R2 of the 2nd image I2. An example of a crack vector whose width has been expanded by crack width processing. Diagram showing how the aligned crack vectors overlap Flow chart of crack change determination process A diagram showing the shading of cracked pixels in an image taken at a certain point in time. A diagram showing the shading of cracked pixels in an image taken at another time point.

FIG. 1 shows a block configuration of a crack change determination device 100 according to a preferred embodiment of the present invention. The crack change determination device 100 includes an image acquisition unit 11, an arrangement relationship estimation unit 12, a crack detection unit 13, a crack width estimation unit 14, a crack width processing unit 15, and a crack change determination unit 16. Each of these parts is composed of one or more processors. The processor includes a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), a GPU (Graphics Processing Unit), and the like. The crack change determination device 100 is composed of a mobile information terminal such as a smartphone, a tablet terminal, a digital camera, a personal computer, or the like.

The image acquisition unit 11 acquires two or more images in which the structure is photographed at different times. The device that captures an image and the device that is the source of the image are a digital camera, a smartphone, or a device (smartphone, personal computer, etc.) that captures and stores the image captured by the digital camera. For example, the image acquisition unit 11 acquires the first image I1 taken at time t1 and the second image I2 taken at time t2 after time t1. The structure to be photographed is not particularly limited to the floor plate of the bridge, the road surface, the concrete surface of the dam body, the concrete surface of the lining of the tunnel, the outer wall of the building, and the like.

The arrangement relationship estimation unit 12 estimates the relevance of the crack positions of the image acquired by the image acquisition unit 11. For example, the arrangement relationship estimation unit 12 estimates the relationship (arrangement relationship) of the cracks detected in the first local region included in the first image I1 and the second local region included in the second image I2. That is, the arrangement relationship estimation unit 12 estimates the correspondence between the positions of the cracks that were originally the same between the local region R1 of the first image and the local region R2 of the second image. The first local region is a rectangular region that is the smallest region including continuous cracks and is included in the first image. The second local region is a region that is the smallest rectangular region including continuous cracks and is included in the second image. For convenience of explanation, the first local region R1 and the second local region R1 are rectangular. However, other shapes may be used.

Specifically, the arrangement relationship estimation unit 12 extracts feature points from both the first image I1 and the second image I2, and associates the feature points between the first image I1 and the second image I2. Then, by a projective transformation such as an affine transformation, the corresponding feature points are aligned between the first image I1 and the second image I2, and the first image I1 and the second image I2 are superimposed.

However, since the shading of the cracked pixels in the image changes depending on the shooting conditions, even if both the first image I1 and the second image I2 are shot at the same location, both images are the same. The points are not always detected as feature points and associated with each other.

The crack detection unit 13 detects cracks from the image acquired by the image acquisition unit 11 and converts (vectorizes) the detected cracks into coordinate information. Vectorized cracks are called crack vectors. The crack vector includes at least the length and orientation of the crack, or the coordinates of the start and end points of the crack. Further, the crack vector may include various additional information such as identification information of the crack vector and shading information of pixels at the coordinates of the crack vector. Cracks can be detected by known methods such as edge detection and deep learning.

The crack width estimation unit 14 estimates the width of each crack detected by the crack detection unit 13 from the image. Specifically, the crack width estimation unit 14 approximates the density distribution (brightness distribution) in the width direction of the cracks in the image with a curve, estimates the central pixel position with a real value, and pixels, as in Patent Document 3. The entire concentration distribution is shifted in the width direction so that the concentration peaks at position zero. By repeating this for all the acquired concentration distributions, a plurality of centered concentration distributions are obtained, and a probability distribution function (Gaussian function) corresponding to the plurality of centered concentration distributions is calculated. Based on this probability distribution function and the shooting conditions at the time of shooting, a scale image having a line drawing having a brightness distribution equivalent to the brightness distribution of the object in the shot image is generated. The estimated crack width is given to the crack vector. As a result, the crack vector has a main component of start point, end point, and length, as well as an additional component of width.

The crack width processing unit 15 performs a crack width processing process for increasing the estimated crack width of the crack width estimation unit 14. For example, in the crack width processing, the contour is expanded by one pixel by performing a dilation process for searching for the maximum value within a predetermined width centered on the pixel of interest on the contour of the crack detected by the crack detection unit 13. .. Alternatively, the dilation process is repeated an arbitrary number of times of two or more times. The cracking process is not limited to the above, and the cracked pixel may be expanded by a circle having a predetermined radius centered on the pixel of interest. The width of the crack vector is updated with the width of the machined crack. As a result, the crack vector has the main components of the start point, the end point, and the length, as well as the additional components of the processed width.

The degree of processing of the crack width may be changed according to the density of cracks in each image including the first image I1 and the second image I2. For example, the higher the density of cracks in an image, the less the degree of expansion of the width of the cracks may be. In this way, it is possible to prevent adjacent cracks from coming into contact with each other by processing the width of the cracks.

The crack change determination unit 16 compares the length and width of the crack vector in which the correspondence relationship is estimated between the two images different by the arrangement relationship estimation unit 12 to obtain the elapsed time corresponding to the difference in the shooting time of the two images. Determine the change in length and width of the corresponding crack.

For example, the crack change determination unit 16 determines the crack length direction of the corresponding crack vector between the first image I1 taken at time t1 and the second image I2 taken at time t2 after time t1. And the change in the width direction are judged. This change can be determined as a change in the length direction and a change in the width direction of the same crack with the passage of time Δt = t2-t1.

The change in cracks is determined according to the degree of overlap of the crack vectors after processing the width of the cracks. For example, the area ratio r of the overlapping portion and the non-overlapping portion of the crack vectors overlapped by the projective transformation is calculated, and if the area ratio r exceeds a certain threshold value (for example, 50%), the non-overlapping portion is cracked. Is determined to be the changed (grown) part.

FIG. 2 conceptually shows how the arrangement relationship estimation unit 12 estimates the arrangement relationship of cracks between the first image and the second image. For example, feature points are extracted from both the first image I1 and the second image I2, and the feature points of the first image I1 and the second image I2 are associated with each other. Then, by a projective transformation such as an affine transformation, the corresponding feature points are aligned between the first image I1 and the second image I2, and the first image I1 and the second image I2 are superimposed. In the case of affine transformation, the corresponding feature between the first image I1 and the second image I2 by any combination of enlargement, reduction, shearing, translation, and rotation of the first image I1 and / or the second image I2. The points are aligned with each other.

The feature points extracted from the first image I1 and the second image I2 are, for example, the shape of the concrete surface formed by the seams and surface patterns of the formwork plywood (so-called control panel). In FIG. 2, feature points f1, f2, f3, f4 are extracted from image I1, and feature points g1, g2, g3, g4 are extracted from image I2. In addition, the feature points f1, f2, f3, f4 and the feature points g1, g2, g3, g4 correspond to each other as the same position, and the state in which these are aligned and overlapped by the projective transformation is shown in the image G. It is shown.

FIG. 3 shows an example of the start point, end point, length of the crack vector detected by the crack detection unit 13, and the crack width estimated by the crack width estimation unit 14 corresponding to each crack vector. An identification number may be assigned to each individual crack.

FIG. 4 is an example of a crack vector in the partial region R1 of the first image I1 and a crack vector in the partial region R2 of the second image. V1-1, V1-2, V1-3, and V1-4 are shown as crack vectors in the partial region R1. Further, as the crack vector in the partial region R2, V2-1, V2-2, V2-3, V2-4, V2-5, V2-6, V2-7 are shown.

FIG. 5 is an example of a crack vector in the first image I1 and a crack vector in the second image whose width has been expanded by the crack width processing of the crack width processing portion 15. Vd1-1, Vd1-2, Vd1-3, Vd1-4 are shown as the expanded crack vectors having the widths of the crack vectors V1-1, V1-2, V1-3, and V1-4 in the first image I1, respectively. Has been done. Further, as the crack vector with the width of the crack vector V2-1, V2-2, V2-3, V2-4, V2-5, V2-6, V2-7 in the second image I2, Vd2-1 respectively. , Vd2-2, Vd2-3, Vd2-4, Vd2-5, Vd2-6, Vd2-7 are shown.

FIG. 6 shows the crack vectors Vd1-1, Vd1-2, Vd1-3, Vd1-4 and the crack vectors Vd2-1, Vd2-2, Vd2-3, Vd2-4 superposed by the arrangement relationship estimation unit 12. The overlapping state of Vd2-5, Vd2-6, and Vd2-7 is shown.

Since the degree of overlap between Vd1-1 and the set of Vd2-1, Vd2-2, and Vd2-3 exceeds a certain threshold value (for example, area ratio 50%), the same crack (no change). It is judged to be cracked). Further, since the degree of overlap between Vd1-2 and the set of Vd2-4 and Vd2-5 exceeds a certain threshold value, it is determined that they are the same crack (crack without change).

On the other hand, Vd2-7 does not overlap with any of the cracks in the first image I1, and is determined to be a change in the cracks (a portion where the cracks have grown). Therefore, the length of Vd2-7 is the length of cracks grown during Δt.

FIG. 7 is a flowchart of the crack change determination process. The crack change determination program for causing various computers (processors) to execute this process is recorded on various computer (processor) readable recording media such as SDRAM, ROM, and CD-ROM.

In S1, the image acquisition unit 11 acquires two or more images taken at different times.

In S2, the arrangement relationship estimation unit 12 estimates the relationship (arrangement relationship) of the cracks detected in the first local region included in the first image and the local region included in the second image.

In S3, the crack detection unit 13 detects cracks from each of the first image and the second image whose arrangement relationship is estimated, and converts (vectorizes) the detected cracks into coordinate information.

In S4, the crack width estimation unit 14 estimates the crack width detected from each of the first image and the second image.

In S5, the crack width processing unit 15 performs a crack width processing process for increasing the estimated crack width of the crack width estimation unit 14.

In S6, the crack change determination unit 16 compares the length and width of the crack vector of the first image and the second image after processing by the crack width processing unit 15 to change the crack in the length direction and the width direction. Judge the change of.

In S7, the crack change determination unit 16 superimposes two or more images taken at different times, calculates the degree of overlap between the vectors, and finds that the same crack vector does not exist. The crack vector is determined to be the change in the length direction.

In S8, the crack change determination unit 16 superimposes two or more images taken at different times, and as a result, if it is determined that the same crack vector exists, the crack width is set for each same crack vector. To compare. Then, when the widths of the cracks are different as a result of the comparison, the difference between the different widths is taken as the change in the thickness direction. If the change in the width of the crack is not determined, S4 and S8 may not be executed.

In this way, even in an image in which the feature points of the same location are not necessarily associated with each other, the cracks with increased widths are overlapped, and the non-overlapping portion is determined to be a change in the cracks, thereby changing the cracks in the width direction. Is reduced in the probability of being erroneously determined as a change in the length direction. Therefore, the change in the length direction of the crack can be determined more reliably.

Further, since the width of the cracks in the images taken at different times is processed and the change in the length direction of the cracks is determined from the overlap of these cracks, the change in the length direction of the cracks can be determined at high speed.

11 Image acquisition unit 12 Arrangement relationship estimation unit 13 Crack detection unit 14 Crack width estimation unit 15 Crack width processing unit 16 Crack change determination unit

Claims (10)

The first image acquisition unit that acquires the first image taken with the surface of the structure as the subject, and
A second image acquisition unit that acquires a second image taken with the surface of the structure as a subject and taken after the first image.
A first crack that detects first crack information, which is information obtained by vectorizing a line segment having a start point, an end point, and a length of a crack on the surface of the structure from the first image acquired by the first image acquisition unit. Information detector and
A second crack that detects second crack information, which is information obtained by vectorizing a line segment having a start point, an end point, and a length of a crack on the surface of the structure from the second image acquired by the second image acquisition unit. Information detector and
A first crack comparison information creation unit that creates first crack comparison information, which is information obtained by processing the shape of the crack of the structure for comparison with respect to the first crack information.
A second crack comparison information creation unit that creates second crack comparison information, which is information obtained by processing the shape of the crack of the structure for comparison with respect to the second crack information.
The change of the crack on the surface of the structure is changed according to the overlap between the first crack comparison information created by the first crack comparison information creation unit and the second crack comparison information created by the second crack comparison information creation unit. The crack change judgment unit to be judged and
A crack change determination device including. The first crack comparison information is information in which a line segment of the crack is rectangularized by imparting a crack width of the structure to the first crack information, and the second crack comparison information is the first crack comparison information. 2. The crack change determination device according to claim 1, which is information in which the line segment of the crack is rectangularized by imparting the width of the crack of the structure to the crack information. A first crack width estimation unit that estimates the width of a crack on the surface of the structure from the first crack information,
A second crack width estimation unit that estimates the width of cracks on the surface of the structure from the second crack information,
With
The first crack comparison information creating unit imparts a crack width on the surface of the structure based on the estimated crack width of the first crack width estimation unit.
The crack change determination device according to claim 2, wherein the second crack comparison information creating unit imparts a crack width on the surface of the structure based on the crack width estimated by the second crack width estimation unit. The first crack width estimation unit estimates the crack width on the surface of the structure based on the concentration distribution in the width direction of the first crack information.
The crack change determination device according to claim 3, wherein the second crack width estimation unit estimates the crack width on the surface of the structure based on the concentration distribution in the width direction of the second crack information. The first crack comparison information creation unit expands the width of cracks on the surface of the structure estimated by the first crack width estimation unit by dilation processing.
The crack change determination device according to claim 3 or 4, wherein the second crack comparison information creating unit expands the crack width on the surface of the structure estimated by the second crack width estimation unit by a dilation process. The first crack comparison information creating unit expands the width of cracks on the surface of the structure according to the density of cracks on the surface of the structure.
The crack change determination device according to claim 5, wherein the second crack comparison information creating unit increases the width of cracks on the surface of the structure according to the density of cracks on the surface of the structure. The crack change determination unit aligns the first crack comparison information and the second crack comparison information by local projective conversion from one of the first image and the second image to the other, and then the crack change determination unit. The crack change determination device according to any one of claims 1 to 6, which determines whether or not the length of cracks on the surface of the structure has changed. The crack change determination device according to claim 7, wherein the projective transformation includes an affine transformation. The computer
The step of acquiring the first image taken with the surface of the structure as the subject,
A step of acquiring a second image taken with the surface of the structure as a subject and taken after the first image, and
From the first image, a step of detecting first crack information, which is information obtained by vectorizing a line segment having a crack start point, end point, and length on the surface of the structure,
From the second image, a step of detecting second crack information, which is information obtained by vectorizing a line segment having a crack start point, end point, and length on the surface of the structure,
A step of creating first crack comparison information, which is information obtained by processing the shape of the crack of the structure for comparison with respect to the first crack information.
A step of creating second crack comparison information, which is information obtained by processing the shape of the crack of the structure for comparison with respect to the second crack information.
A step of determining a change in cracks on the surface of the structure according to the overlap between the first crack comparison information and the second crack comparison information.
Crack change judgment method to execute. A crack change determination program for causing a computer to execute the crack change determination method according to claim 9.

Images