JP5705711B2 - Crack detection method - Google Patents

Description

  The present invention relates to a crack detection method for detecting cracks generated on a concrete surface, and in particular, it is possible to homogenize image quality in a plurality of photographed images, and to easily perform high-accuracy crack detection. The present invention relates to a crack detection method.

  As a method for detecting cracks on the concrete surface, conventionally, a method in which an investigator visually observes using a scale and measures the width and length of the crack has been common. However, this visual observation method requires a large amount of labor and time to accurately process a large amount of information when there are large variations in accuracy due to the measurement skill of the investigator and when there are a large number of cracks. There was a problem.

  In order to solve the above problem, an image processing technique is applied in which a captured image of a concrete surface is taken into a computer, and the image is binarized into a cracked area and other areas. The image binarization process is to express the image density to 0 or 1 with respect to a certain density value. For example, 2 obtained by the binarization process for the input image f (i, j). The binarized image b (i, j) is b (i, j) = 1 (f (i, j)> k) and 0 (f (i, j) ≦ k). Here, k is a threshold value for binarization, and therefore it can be said that the quality of the binarized image is determined by the selection of the threshold value k.

  As a conventional method for obtaining a threshold value, there is a processing method using a fixed threshold value or a variable threshold value. Examples of the processing method using a fixed threshold include a P tile method, a mode method, and a method using a correlation ratio. The processing method using a fixed threshold creates a density histogram of the target image, and a bimodal histogram is obtained in which a clear valley appears between the density value of the background (concrete surface) of the image and the density value of the crack. This is an effective method in some cases.

  On the other hand, the processing method using a variable threshold is an effective method in the case where shooting unevenness occurs due to illumination conditions or the like and the background density value and the density value of the target portion are not constant in the entire image. This variable threshold processing method is a method in which an average density value of a local region centered on a pixel of interest is used as a threshold. A drawback of this method is that, for example, an image having a lot of noise other than cracks is generated in accordance with a subtle change in shading of the background area.

  In the conventional image processing method, a threshold is determined for a photographed input image, and cracks are extracted while performing a binarization process. That is, the general processing flow is as follows. 1) Capture the captured image into a computer and create an input image. 2) Correct the density of the input image. 3) Perform binarization and extract cracks. 4) Smooth the cracks and track the contour lines. 5) Calculate the characteristic amount of the specified crack.

  The conventional image processing method described above can detect a crack with relatively high accuracy in detecting a crack on a concrete surface having a uniform concentration. However, the surface of an actual concrete structure contains various stains, and the density of cracks generally varies depending on the width and depth of the cracks. When a conventional image processing method is used for the concrete surface, various problems may occur when extracting cracks. For example, in the case of the fixed threshold processing, if the dirt region and the crack region on the concrete surface have similar density values, it is extremely difficult to binarize them. Even if the density histogram exhibits bimodality and the threshold value can be easily determined, there is a very high possibility that the area that is determined to be a cracked area includes a dirty area. On the other hand, if a new threshold value is set so as not to include the dirt region around the crack, another crack region will be excluded this time.

  In the case of variable threshold processing, as the dirt on the concrete surface increases, more noise other than cracks will be included in the cracked extracted image. It is impossible to determine at all whether or not the region is a cracked region.

  Therefore, performing high-accuracy crack detection is an important solution in the technical field.

  On the other hand, as a final drawing in a crack investigation, a development view of a crack distribution is generally created. This crack development view is for creating a crack occurrence state and crack width on a map.

  As a method of creating a crack development map, 1) A method in which an inspector visually sketches the crack situation on the site, or measures the crack width of a representative location on a crack scale, and 2) a digital camera, etc. The method of tracing the crack condition and the crack width can be cited with reference to the image data taken in step 1.

  However, because these methods often involve the subjective and qualitative judgments of investigators and tracers, the assessment of cracks is likely to vary among investigators, and there is a possibility of oversight and misperception. .

  In order to solve these problems, it is important to evaluate cracks objectively and quantitatively without intervention of human judgment as much as possible.

  As a measure for this, the present inventors have invented an invention that has been used for image analysis using wavelet transform on images taken with digital devices such as digital cameras and video cameras to precisely evaluate cracks. 1 to 4 etc. By applying the crack detection method disclosed in these patent documents, it is possible to perform the above-described problem, that is, crack detection with high accuracy even when there is a lot of dirt on the concrete surface.

  By the way, in image analysis, it is possible to detect a crack of 0.2 mm or more from an image taken with a spatial resolution of 0.8 mm / pixel, but it is difficult to identify how much the crack width is actually. For example, when the crack width is detected in pixel units, it is necessary to photograph with a spatial resolution of 0.2 mm / pixel to identify the crack width of 0.2 mm.

  Further, in order to take a picture with a spatial resolution of 0.2 mm / pixel, close-up photography is often required, and accordingly, the number of pictures to be taken increases, and the on-site photography and analysis processing takes time and effort. . Therefore, in order to improve the work efficiency, it is necessary to perform remote shooting to reduce the number of shots. For this reason, image analysis based on low-resolution shot images is unavoidable.

  In addition, as the shooting target becomes wider, shooting will be performed separately for each section, but due to differences in conditions such as weather, time of day, sunlight, etc., the contrast, brightness, hue, Differences in quality such as blur can occur.

  In order to ensure the accuracy of the crack width estimation formula for images with different quality in this way, a crack scale is attached to each image, and the actual scale value (crack width) of the crack scale is taken. It is necessary to calibrate.

  For example, the crack detection method disclosed in Patent Document 1 is a method for estimating the crack width below the spatial resolution from low-resolution captured image data. More specifically, a positive correlation between the wavelet coefficient and the crack width is high. Using this, a relational expression with the crack width is created from the wavelet coefficient, and the crack width is estimated from the value of the wavelet coefficient obtained from the captured image data.

  And as a feature of this crack detection method, when photographing cracks occurring on concrete, two cases are considered, when a crack scale is applied to concrete and when it is not applied.

  By sticking a crack scale and calibrating the cracks detected based on the analysis result of this crack scale, it is possible to eliminate differences in quality such as contrast, brightness, hue and blur, for each captured image, If the pasted crack scale is floating from the concrete surface, the accuracy of the estimation formula will be greatly affected. Further, when the surface to be photographed is at a high place or a place where access is dangerous, it is difficult to attach the crack scale itself.

  On the other hand, if the crack scale cannot be applied, the crack width is estimated using the crack width estimation formula, which corresponds to the maximum and minimum values of the wavelet coefficient, which is one of the factors in this estimation formula. There is a problem that it is necessary to set the crack width for each spatial resolution of the captured image.

  From the above, it is possible to specify the crack width with high accuracy by pasting the crack scale on the concrete surface to be photographed and calibrating the detected cracks. Therefore, there is a demand in the technical field for a technique that can specify the crack width with high accuracy without attaching a crack scale to the concrete surface to be photographed.

JP 2008-267943 A JP 2010-121992 A JP 2008-185510 A JP 2006-162583 A

  The crack detection method of the present invention has been made in view of the above-described problems, and provides a crack detection method capable of accurately identifying the crack width without attaching a crack scale to the concrete surface to be photographed. It is an object.

  A wavelet means a small wave, and a basic unit of a localized wave can be expressed by an expression using a wavelet function. By enlarging or reducing the wavelet function, it is possible to simultaneously analyze time information, spatial information, and frequency information. A characteristic of the wavelet coefficient when the wavelet coefficient is applied to a concrete surface having cracks is that the coefficient depends on the concentration of the concrete surface, the concentration of cracks, and the crack width. For example, the value of the wavelet coefficient tends to increase as the crack width increases, and the value of the wavelet coefficient tends to increase as the density of cracks increases (closer to black).

  An algorithm for detecting cracks using wavelet coefficients calculated by wavelet transform is as follows. First, the wavelet coefficient is obtained from the inner product of the captured image of the concrete surface and the wavelet function. By converting this wavelet coefficient to 256 gradations, a wavelet image having a continuous amount can be created.

  As described above, the wavelet coefficient varies depending on the crack width, crack density, and concrete surface density. As a result, the wavelet coefficient related to the crack density and concrete surface density for each gradation is created using simulated data. The wavelet coefficient table is created in advance. The wavelet coefficient for each gradation in the wavelet coefficient table serves as a threshold for crack detection. For example, wavelet coefficients (thresholds) corresponding to two contrasted concentrations (one concentration can be assumed to be a concrete surface concentration and the other concentration to be a crack concentration) are unique if the wavelet coefficient table is referenced. To be determined. Therefore, as will be described later, when the wavelet coefficient between the two densities to be compared in the captured image is calculated, if this wavelet coefficient is larger than the threshold value of the wavelet coefficient table, it can be determined that it is a crack, and the threshold value If it is smaller than that, it can be determined that it is not cracked.

  The pseudo data when creating the wavelet coefficient table is not particularly limited. For example, the crack width is one pixel (one pixel) to five pixels (five pixels), and each pixel width is cracked. For each, the wavelet coefficient corresponding to the gradation of the concrete surface and the gradation of the crack is calculated. When setting the threshold value, for example, the wavelet coefficient corresponding to the crack is selected from the wavelet coefficients when the crack width is one pixel, and the wavelet coefficient of the portion that is not the crack area is selected among the wavelet coefficients when the crack width is five pixels. A coefficient is selected, and an average value of these two wavelet coefficients can be used as a threshold value in an arbitrary gradation.

  In other words, by using the existing captured images of the crack scale, it is possible to eliminate sticking of the crack scale to the object to be imaged. The problem that the crack scale cannot be pasted in the case of a difficult place, and the problem that the accuracy of the crack width estimation formula can be greatly affected when the pasted crack scale is floating from the concrete surface are effectively solved.

  Next, for example, a crack is specified on each of a plurality of concrete surfaces to be photographed constituting a crack development view finally created.

  Specifically, as a second step, for each of a plurality of photographing objects, a crack density and a concrete surface density are set in a pseudo manner, wavelet coefficients corresponding to two contrasted densities are calculated, and the A wavelet coefficient table is created by calculating each wavelet coefficient when each of the two concentrations is changed, and a captured image of the concrete surface to be detected for cracking is input to a computer as an input image. By converting, a wavelet image of each concrete surface is created for each subject.

  The wavelet image is created in the computer as follows. First, wavelet coefficients are calculated for an appropriately set wide area (for example, an area of 30 × 30 pixels). Next, it is the same in a wide area moved by one pixel from this wide area (in the same way, for example, a 30 × 30 pixel area and most of the pixels are the same as the 30 × 30 pixel area before the movement). The wavelet coefficient is calculated as follows. By repeating this operation for the entire input image, a wavelet image composed of continuous amounts of wavelet coefficients is created inside the computer.

  Next, as a third step, in the wavelet coefficient table, wavelet coefficients corresponding to the average density of neighboring pixels in the local region assumed to be the density of the concrete surface and the density of the target pixel assumed to be the crack density If the wavelet coefficient of any pixel of interest in any neighboring pixel is greater than the threshold, the pixel of interest in that neighboring pixel is determined to be cracked, and the wavelet coefficient of any pixel of interest in any neighboring pixel is the threshold If it is smaller than that, it is determined that the target pixel in the neighboring pixels is not cracked, and the wavelet coefficient of the target pixel is compared with the threshold value while changing the local region and the target pixel, and noise other than cracks is removed. (Binarization processing), and further thinning processing is performed to configure the center line. That, to create a crack of the thinned image of the concrete surface to be imaged.

  In a wavelet image consisting of continuous wavelet coefficients, the threshold (wavelet coefficient) in the wavelet coefficient table is compared with the wavelet coefficients that make up the wavelet image, and if the wavelet coefficients that make up the image are larger than the threshold, it is judged as cracked. If it is smaller than the threshold value, it is determined that it is not cracked (eg black on the screen). By performing this operation on the entire wavelet image, a crack extraction image in which white cracks are drawn in a black background color is created.

  In creating the crack extraction image, noise is removed in addition to the binarization process. As a method for removing this noise, there are methods such as using a known image editing software to remove a dot portion or to remove a line segment having a length less than a predetermined length as a non-crack portion. Further, as one of the algorithms for the noise removal method, there is a contour line tracking process. This contour tracking process starts from an arbitrary pixel (a pixel that has been determined to be cracked), connects to the starting pixel if the adjacent pixel is a cracked location, and further if the adjacent pixel is a cracked location Are further connected to each other and finally closed to the starting pixel (for example, the first pixel, the second pixel,..., The n−1th pixel, the nth pixel, and the first pixel are connected in this order). Alternatively, the process is terminated when there is no longer any crack connected to the next. According to this contour line tracking process, an appropriate crack line such as a crack line that closes in a loop shape or a crack line that has a plurality of bent portions and extends linearly is created. At this time, by setting the minimum number of connected pixels in advance, it is possible to delete all pixels equal to or less than the set number from the crack display on the screen, assuming that they are not cracked.

  Moreover, you may perform an outline tracking process after performing a smoothing process. Here, the smoothing process calculates an average value within an appropriately set number of pixels. For example, among the plurality of pixels, a pixel having a darker density than the average value is a cracked pixel, and the average value is calculated. This is a technique for determining that pixels having a lighter density are not cracked.

  Next, thinning processing is performed on an image from which only cracked pixels are extracted, thereby creating a thinned image that is formed by the centerline of the crack, for example, the entire crack has one pixel width (one pixel width). . According to the inventor's verification, it has been found that the density of the center line portion is the highest in a crack with an arbitrary width, and therefore, a crack that forms a crack with an arbitrary width and an arbitrary length by thinning processing. In, the crack concentration of each part can be set.

  As described above, in the second and third steps, a cracked thinned image is created for each concrete surface to be imaged.

  The crack width estimation formula specified by the regression analysis changes depending on the amount of data of the accumulated crack scale.

  As described above, in the fourth step, the fine lines of the photographed images of the crack scale in the database accumulated in the same manner as the image analysis on the concrete surface to be photographed in the second and third steps. This is to specify a digitized image. Therefore, in the fourth step, a method may be used in which the process up to specifying the thinned image of the crack scale photographed image is executed in the first step.

  As a result of the regression analysis, it is possible to create a database composed of crack width at crack scale, wavelet coefficient, luminance of target pixel, average luminance of surrounding pixels, spatial resolution and correlation coefficient.

  According to the verification by the present inventors, it has been proved that the crack width can be specified with extremely high accuracy by applying the above estimation formula.

  Further, in a preferred embodiment of the crack detection method according to the present invention, as a result of the regression analysis, the crack width estimation formula is obtained by using only a photographed image of a crack scale having a correlation coefficient of a certain height or higher. It is.

  The accuracy of the estimation formula largely depends on the correlation coefficient obtained by the regression analysis. More specifically, the regression analysis is performed using only the photographed image of the crack scale having a certain high correlation coefficient. It is desirable to specify an estimation formula for the crack width. As a result, it is possible to remove a blurred image that depends on a problem of focus when the crack scale is photographed or a sticking method of the crack scale is floating, and as a result, the accuracy of the estimation formula is improved.

  As can be understood from the above description, according to the crack detection method of the present invention, the already captured image of the crack scale is set as a database, and wavelet transform, 2 Create a database consisting of wavelet coefficients, threshold values, actual crack width values, etc. by executing value processing and thinning processing, and substituting the crack scale wavelet coefficients into a predetermined relational expression to estimate crack width And the crack width can be specified with high accuracy by applying the data of the thinned image of the crack on the concrete surface to be photographed to this estimation formula.

It is the schematic diagram which showed the relationship between an input image and a local area | region. It is the schematic diagram which showed the relationship between a local area | region and an attention pixel. It is a flowchart of one embodiment of the crack detection method of the present invention. It is the figure which showed the pseudo image. FIG. 5 is a bird's-eye view of wavelet coefficients of the pseudo image of FIG. 4. It is the figure which showed one Embodiment of the wavelet coefficient table. It is a flowchart of other embodiment of the crack detection method, Comprising: It is the figure which extracted the flow after step S110 in the flow of FIG. (A), (b) is the figure which showed the picked-up image of a crack scale, (c), (d) is the image which extracted only the image of the crack scale from (a), (b), respectively. (E) and (f) are diagrams showing thinned images of crack scale images of (a) and (b), respectively. (A) is the figure which showed the estimated value of the crack width of the image of FIG. 8c, (b) is the figure which showed the estimated value of the crack width of the image of FIG. 8d, (c) is a figure. It is the figure which compared the result estimated with the regression equation of 9a, b on the same drawing. (A), (b) is the figure which each showed the picked-up image of the crack scale from which brightness differs extremely, (c), (d) is the image of a crack scale from (a), (b), respectively. (E) and (f) are diagrams showing thinned images of crack scale images of (a) and (b), respectively. (A) is the figure which showed the estimated value of the crack width of the image of FIG. 10c, (b) is the figure which showed the estimated value of the crack width of the image of FIG. 10d, (c) is a figure. It is the figure which compared the result estimated by the regression equation of 10a, b on the same drawing. (A) is a figure which shows the measurement result which calculated | required the ratio in which the estimated crack width and the crack width measured by the crack scale are suitable in the range of error +/- 0.05mm, (b), It is a figure which shows the measurement result which calculated | required the ratio in which the estimated crack width and the crack width measured in the crack scale are suitable within the range of error +/- 0.1mm (in the figure, Example (high resolution) and In the examples (low resolution), the crack width was estimated by the method of the present invention, and in the comparative example, the crack width was estimated from the low resolution image using the method disclosed in Patent Document 1. It is the figure which showed distribution of the crack width estimated for every crack width measured value in the case of low resolution, (a) is a case where a measured value is 0.08 mm, (b) is a measured value is 0.1 mm. (C) is the case where the measured value is 0.15 mm, (d) is the case where the measured value is 0.2 mm, (e) is the case where the measured value is 0.25 mm, and (f) is the measured value is 0. FIG. It is the figure which showed distribution of the crack width estimated for every crack width measured value in each case of low resolution and high resolution, (a) is a case where a measured value is 0.08 mm, (b) is a measured value. Is a case where the measured value is 0.15 mm, (d) is a case where the measured value is 0.2 mm, (e) is a case where the measured value is 0.25 mm, and (f) is It is a figure of a case where a measured value is 0.3 mm.

  Hereinafter, embodiments of a crack detection method of the present invention will be described with reference to the drawings.

(Embodiment 1 of crack detection method)
Before explaining the flow chart of the crack detection method, an outline until a wavelet image is created will be explained. FIG. 1 is a schematic diagram showing a relationship between an input image and a local region. In the crack detection method of the present invention, wavelet transform is performed in the wide area 2 in the input image 1, and cracks are detected in the local area 3 that is the center of the wide area 2. The wide area 2 is translated vertically and horizontally throughout the input image 1 to detect cracks in the input image 1. By this method, it is possible to detect cracks with higher accuracy as compared with the method of determining one threshold value in the input image 1 as in the conventional fixed threshold method.

  FIG. 2 is an enlarged view of the local region 3. In the illustrated embodiment, for example, nine pixels of 3 × 3 (eight neighboring pixels 31, 31... And a target pixel 32 located in the center) are targeted. As a crack judgment. The wavelet coefficient is calculated for the wide area 2 in FIG.

  Here, a calculation formula for calculating a wavelet coefficient by performing wavelet transform using a wavelet function (mother wavelet function) is shown below.

Here, f (x, y) is an input image (where x and y are arbitrary coordinates in the two-dimensional input image), ψ is a mother wavelet function (Gabor function), and (x ₀ , y ₀ ) is the translation amount of ψ, a ^k is the enlargement or reduction of ψ (where a ^k is the reciprocal of the frequency, and the frequency width for calculating several frequency regions is indicated by an integer k) value), rotating the f ₀ is the center frequency, the σ is the standard deviation of the Gaussian function, the rotation angle theta is representing the traveling direction of the wave, (x ', y') is the angle θ (x, y) Each coordinate is shown.

Here, the cumulative value C (x ₀ , y ₀ ) of the wavelet coefficient Ψ is obtained for a plurality of θ, k calculated using Formula 1 as Formula 4.

The above parameters can be set arbitrarily. For example, σ can be set to 0.5 to 2, a ^k to 0 to 5, f ₀ to 0.1, and the rotation angle to 0 to 180 degrees. .

The parallel movement amount (x ₀ , y ₀ ) in Equation 4 corresponds to the position of the target pixel, and the continuous amount of wavelet coefficients (C (x ₀ , y ₀ ) is obtained by sequentially moving the position of the target pixel. )) Can be calculated, and a wavelet image can be created by illustrating this continuous quantity.

  After calculating the wavelet coefficients for all the pixels constituting the wide area 2 based on the above calculation formula, the wavelet coefficients are similarly applied to all the pixels in the wide area 2 that can be obtained by moving one pixel to the left or right or up and down. Is calculated. By performing this wavelet coefficient calculation on the entire input image, it is possible to create a wavelet image (an image made up of a continuous amount of wavelet coefficients) in which the constituent pixels within an appropriate range have respective wavelet coefficients.

  Next, Embodiment 1 of the crack detection method will be described with reference to FIG.

  Unlike the concrete surface to be photographed this time, it is a crack scale photographed image that has been pasted on the concrete surface that was photographed previously, and the already accumulated crack scale photographed image is set as a database. deep. More specifically, a crack scale is affixed to the concrete surface, and only the crack scale is extracted from a photographed image photographed by a digital camera such as a CCD camera to create a photographed image of only the crack scale, which is made into a database (step S10). .

  The crack scale photographed image database contains multiple actual size values from the minimum width to the maximum width of the crack scale, and is composed of multiple image data with different photographing distance, lens focal length, and spatial resolution. It is what is done.

  Next, an input image is created by taking in a computer a plurality of photographed images of the concrete surface to be photographed this time (step S20). With step 10 described above, there is no need to stick a crack scale to the concrete surface during step S20.

  Next, wavelet coefficients are calculated for a pseudo image having two contrasts that have no relationship with the input image. For example, as shown in FIG. 4, a background color a assumed to be a concrete surface (for example, R, G, B of the background color is assumed to be 255, 255, 255) and line segments b1 to b1 assumed to be cracks. The wavelet coefficient of the pseudo image consisting of b5 is obtained. Here, the line segments b1 to b5 change in order from 1 pixel (1 pixel) to 5 pixels (5 pixels) in order, and each line segment has three types of density (for example, In the line segment b1, it changes to b11 (black), b12 (light black), and b13 (gray) in descending order of density). FIG. 5 shows a bird's-eye view of wavelet coefficients calculated by performing wavelet transform on this pseudo image. In FIG. 5, the X axis indicates the line segment width, the Y axis indicates the color density of the line segment, and the Z axis indicates the wavelet coefficient. The line width may be set according to the maximum crack width to be finally extracted. For each pixel width, the wavelet coefficient of the crack region and the wavelet coefficient other than the crack region can be calculated.

  In this embodiment, when calculating the arbitrary density (gradation) assumed to be a concrete surface and the threshold value (wavelet coefficient) corresponding to the arbitrary density (gradation) assumed to be a crack, for example, the crack width is Threshold value corresponding to the target gradation within the set crack width range, with the average value of the wavelet coefficient of the crack area in the case of 1 pixel width and the wavelet coefficient other than the crack area in the case of the crack width of 5 pixels. It is said. Of course, this threshold value may be arbitrarily set.

  A wavelet coefficient table as shown in FIG. 6 is created (step S40) by combining the two densities to be compared with 256 gradations of 0 to 255, respectively. Note that this operation does not have to be performed at the illustrated flow position, and may be performed, for example, before creation of the input image.

  A wavelet image is created (step S30) by wavelet transforming the input image.

  As described above, the wavelet image is a continuous amount in which each pixel has a unique wavelet coefficient, and by comparing the wavelet coefficient of each pixel with the wavelet coefficient (threshold) of the corresponding wavelet coefficient table, A binarized image is created (step S50). For example, the wavelet coefficient of an arbitrary pixel is the density of the pixel (in the wavelet coefficient table, this pixel density corresponds to the crack density) and the average density of neighboring pixels in the local area (in the wavelet coefficient table, in the local area). If the average density of neighboring pixels is larger than the wavelet coefficient (threshold value) that is uniquely determined by (corresponding to the concrete density), this pixel is determined to be cracked.

  By performing the same comparison for the wavelet coefficients of each pixel in the computer, for example, a binary image in which a white line segment (crack) is drawn on a black screen (concrete surface) is created. Noise is further removed from the binarized image.

  As a method for removing noise, there are a method of removing a dot portion using a known image editing software, and a method of removing a line segment having a length less than a predetermined length as a non-crack portion.

  As another method, there is a method of performing smoothing processing and contour tracking processing. In the smoothing process, the average value (for example, median value) of the density in the local area is set as the density value of the pixel of interest in the local area, thereby removing noise from the binarized image and narrowing the cracked portion. By performing the smoothing process, a smoothed image is created.

  Next, the contour line tracking process is performed on the smoothed image. The contour tracking process starts from an arbitrary cracked pixel in each cracked area (first pixel), for example, pays attention to a pixel that is adjacent counterclockwise from the first pixel, and the adjacent pixel (second pixel) is In the case of a cracked pixel, the first pixel and the second pixel are connected. Thereafter, similarly, the second pixel, the third pixel,..., The (n−1) th pixel, the nth pixel and the cracked pixel are tracked, and when the first pixel that is the starting point comes after the n pixel, One pixel to n-th pixel are determined as one crack location (crack line). Alternatively, when there is no crack pixel following the n pixel, the first pixel to the nth pixel are determined as one crack location (crack line). In addition, the crack line which branches into two or more branches in the middle is also included in the crack line. The setting of the order n is arbitrary, and a crack extraction image is created by determining that the tracking number from the first pixel is greater than the set order n as a crack.

  In the binarized image, a thinned image is formed which is composed of the center line of the crack, for example, the entire crack has one pixel width (one pixel width) (step S60).

  In this way, a thinned image for each concrete surface is created by executing Steps S20 to S60 with the captured images of the concrete surfaces to be imaged.

  On the other hand, the image of each crack scale stored in the database in step S10 is input to a computer to create an input image of each crack scale, and the crack scale input image is converted by wavelet transformation of each crack scale input image. A wavelet image is created (step S70), and a wavelet coefficient of each pixel of the wavelet image is compared with a wavelet coefficient (threshold value) of a corresponding wavelet coefficient table to create a binary image of a crack scale. Then, thinning processing is executed on this to create a crack-scale thinned image composed of the center line (step S90).

  If the crack width of each concrete surface is specified, the development of the adjacent concrete surface image can be created (step S120).

  In FIG. 3, the flow from the crack scale database to the identification of the crack width estimation formula and the creation of a thinned image of the concrete surface to be imaged are shown in a parallel flow diagram. The method is not limited to this flowchart. For example, steps S10 to S100 are performed in advance, an estimation formula for crack width is specified in advance, and then a thin line on the surface of the concrete to be photographed. It may be a flow for creating a digitized image.

  Here, as a result of the regression analysis, it is desirable to perform a regression analysis using only a crack scale image having a relatively high correlation coefficient to specify an estimation formula for a crack width.

(Embodiment 2 of crack detection method)
FIG. 7 shows a flowchart of the second embodiment of the crack detection method, and more specifically, most of the parts common to the first embodiment shown in FIG. 3 are omitted. The method has a new step S130 after step S110.

[Verification of accuracy of crack detection method of the present invention]
The present inventors tried to verify the accuracy of the crack detection method of the present invention by the following method.

  In this verification, the accuracy of the crack width estimation formula specified by the present invention is verified by taking an image taken with standard quality as an example. Here, “images taken with standard quality” are images that are brightly shot, in focus, and can identify cracks to some extent by visual inspection. When a regression analysis is performed alone, the correlation coefficient Is an image corresponding to 0.85 or more.

  FIGS. 8a and 8b are stored images including two crack scales (image A and image B, respectively), and the crack scale images extracted from the respective images are shown in FIGS. 8c and 8d. Further, these thinned images are shown in FIGS.

  The crack scale used has an actual size value having a minimum width of 0.04 mm and a maximum width of 2.2 mm.

  The database of crack scale images accumulated by the present inventors so far includes a shooting distance of 3.1 to 11.5 m, a lens focal length of 40 to 175 mm, and a spatial resolution of 0.25 mm / pixel to There are 238 pieces of image data taken at 0.62 mm / pixel, and this database is updated as needed.

  8e and f, it is possible to detect a crack having a width of 0.04 mm or more. FIG. 9a is a diagram showing an estimated value of the crack width of the image of FIG. 8c, FIG. 9b is a diagram showing an estimated value of the crack width of the image of FIG. 8d, and FIG. It is the figure which compared the result estimated by these regression equations on the same drawing.

  From FIG. 9a, the correlation coefficient of the regression equation obtained from the data of the image A alone is 0.880, and it can be confirmed that the estimated crack width and the crack width measured with the crack scale are in good agreement.

  On the other hand, from FIG. 9b, the correlation coefficient of the regression equation obtained from the data of image B alone is 0.899, and the crack width estimated in the same manner as in FIG. 9a and the crack width measured on the crack scale agree well. Can be confirmed.

  And from FIG. 9c, it can also confirm that both have the same result.

  Next, as shown in FIGS. 10a and 10b, the accuracy of the crack width estimation formula based on the photographed image having the crack scale photographed under different brightness is verified.

  FIGS. 10a and 10b are stored images including two crack scales (image C and image D, respectively), and crack scale images extracted from the respective images are shown in FIGS. 10c and 10d. Further, these thinned images are shown in FIGS.

  10e and f, it is possible to detect a crack having a width of 0.04 mm or more. FIG. 11a is a diagram showing an estimated value of the crack width of the image of FIG. 10c, FIG. 11b is a diagram showing an estimated value of the crack width of the image of FIG. 10d, and FIG. It is the figure which compared the result estimated by these regression equations on the same drawing.

  Since the correlation coefficient of the regression equation of image C is 0.728 and the correlation coefficient of the regression equation of image D is as low as 0.481, the estimated crack width and the crack width measured at the crack scale are not necessarily good. It cannot be said that it is in agreement. However, it has been proved that the crack width estimated by applying the crack width estimation formula by the crack detection method of the present invention and performing the image equalization process specifies a crack width equal to or greater than the individual regression analysis. Yes.

  Next, the crack detection method of the present invention was applied to data photographed with a spatial resolution of 0.8 mm / pixel and a spatial resolution of 0.3 mm / pixel of the bridge floor slab, which is an actual structure, to verify the estimation accuracy. As a comparative example, a crack width estimated by applying the crack detection method shown in Patent Document 1 will be taken up.

  FIG. 12a is a diagram showing a measurement result obtained by calculating a ratio in which the estimated crack width and the crack width measured at the crack scale are suitable within an error of ± 0.05 mm, and FIG. 12b is estimated. It is a figure which shows the measurement result which calculated | required the ratio which the crack width measured by the crack width and the crack scale is suitable within the range of error +/- 0.1mm. In the figure, the example (high resolution) and the example (low resolution) are obtained by estimating the crack width by the method of the present invention, and the comparative example is a low resolution using the method disclosed in Patent Document 1. The crack width is estimated from the image.

  Here, the appropriateness ratio is calculated by (number of appropriateness) / (number of appropriateness + number of inappropriateness + number of undetected) × 100.

  12A and 12B, the comparative example (Patent Document 1) has a crack width of 0.15 mm or less compared to the example (both high resolution (0.3 mm / pixel) and low resolution (0.8 mm / pixel)). On the other hand, it has been proved that the predictive value is extremely deteriorated.

  Moreover, it has been demonstrated that both the comparative example (Patent Document 1) and the example have a high predictive value for crack widths of 0.2 mm or more.

  From this fact, the method of the comparative example has a good predictive value for a crack width of 0.2 mm or more, whereas the method of the present invention is not limited to 0.2 mm or more. It can be said that it has a high predictive value even for a crack width of 15 mm or less.

  Next, the distribution of crack widths estimated by the method of the comparative example and the method of the present invention is obtained for the crack measurement values 0.08 mm to 0.3 mm for a low resolution (spatial resolution 0.8 mm / pixel) photographed image. It was. 13a is a case where the measured value is 0.08 mm, FIG. 13b is a case where the measured value is 0.1 mm, FIG. 13c is a case where the measured value is 0.15 mm, FIG. 13d is a case where the measured value is 0.2 mm, and FIG. Is a case with a measured value of 0.25 mm, and FIG. 13f is a diagram of a case with a measured value of 0.3 mm.

  From the above figures, the crack width estimated by the method of the present invention is 0.08 mm to 0.3 mm and is normally distributed around the crack measurement value, and the error is within ± 0.05 mm.

  On the other hand, the crack width estimated by the method of the comparative example (Patent Document 1) tends to be 0.1 mm larger than the measured value below 0.15 mm, and 0.05 mm larger than the measured value above 0.2 mm. It has been proven.

  Furthermore, regarding the method of the present invention, the distribution of crack width estimated at a spatial resolution of 0.8 mm / pixel (low resolution) and a spatial resolution of 0.3 mm (high resolution) was obtained. 14a is a case where the measured value is 0.08 mm, FIG. 14b is a case where the measured value is 0.1 mm, FIG. 14c is a case where the measured value is 0.15 mm, FIG. 14d is a case where the measured value is 0.2 mm, and FIG. Is a case with a measured value of 0.25 mm, and FIG. 14 f is a diagram of a case with a measured value of 0.3 mm.

  From the above figures, the same results are obtained for both the low resolution and the high resolution, and it can be confirmed that a normal distribution is exhibited for each crack measurement value.

  The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and there are design changes and the like without departing from the scope of the present invention. However, they are included in the present invention.

DESCRIPTION OF SYMBOLS 1 ... Input image, 2 ... Wide area | region, 3 ... Local area | region, 31 ... Neighboring pixel, 32 ... Interesting pixel

Claims (3)

A crack detection method for detecting cracks occurring on the concrete surface of a plurality of photographing objects,
A first step of setting, as a database, a captured image of the crack scale that has already been accumulated in a captured image of a crack scale pasted on the concrete surface of the imaged object different from the imaged object;
For each of the plurality of photographing objects, a crack density and a concrete surface density are set in a pseudo manner, wavelet coefficients corresponding to two contrasted densities are calculated, and the two densities are respectively changed. Each wavelet coefficient is calculated to create a wavelet coefficient table, and a captured image of the concrete surface that is the target of crack detection is input to a computer as an input image, and the input image is subjected to wavelet transform for each shooting target. A second step of creating a wavelet image of the concrete surface;
In the wavelet coefficient table, the average density of neighboring pixels in the local area assumed to be the density of the concrete surface and the wavelet coefficient corresponding to the density of the target pixel assumed to be the crack density are used as threshold values, and in any neighboring pixels If the wavelet coefficient of any pixel of interest is larger than the threshold, the pixel of interest in the neighboring pixel is determined to be cracked, and if the wavelet coefficient of any pixel of interest in the neighboring pixel is smaller than the threshold, The target pixel is determined not to be cracked, the local area and the target pixel are changed, the wavelet coefficient of the target pixel is compared with a threshold value, noise other than cracks is removed, thinning processing is performed, and the center A thin line of cracks on the concrete surface to be photographed, composed of lines A third step of creating an image, composed,
A crack scale photographed image in the database is input to a computer as a crack scale input image, and a crack scale wavelet image is created by wavelet transforming the crack scale input image. A thinning image of a crack scale composed of the center line by executing thinning processing on the crack scale is created, and in the thinning image of the crack scale,

A crack detection method, further comprising a fourth step of specifying a crack width of each concrete surface for each photographing target.   The crack detection method according to claim 1, wherein as a result of the regression analysis, the crack width estimation formula is obtained using only a photograph image of a crack scale having a correlation coefficient equal to or higher than a certain height. In order to standardize the quality of captured images that vary from subject to subject, based on a certain standard,

The crack detection method according to claim 1 or 2.

Images