JP7784949B2 - Reinforcement inspection device, reinforcement inspection method, and program - Google Patents

Description

This disclosure relates to a reinforcement inspection device, a reinforcement inspection method, and a program.

In the construction of reinforced concrete structures, after the rebars have been assembled and arranged, a rebar arrangement inspection is performed to check whether the rebars have been arranged as designed. For example, Patent Document 1 describes a rebar arrangement confirmation support system that uses photographed images of arranged rebars to determine the diameter of the rebars. The system obtains a one-dimensional distribution of brightness values of pixels aligned along the length of the rebars from the photographed images, and determines the node spacing of the rebars using a waveform obtained by frequency analysis of the one-dimensional distribution. The system then determines the diameter of the rebars by using the name corresponding to the identified node spacing from the names (rebar diameters) associated with each of the multiple node spacings in the JIS standard.

Japanese Patent Application Laid-Open No. 2021-21622

However, because there is not necessarily a one-to-one correspondence between rebar diameter and node spacing, the reinforcement bar confirmation support system described in Patent Document 1 had the problem of being unable to accurately determine the rebar diameter. For example, in the case of threaded rebar, where the nodes are formed in a screw-like shape, different names may be assigned to the same node spacing (thread pitch).

The present disclosure aims to solve the above problem by providing a reinforcing bar inspection device, reinforcing bar inspection method, and program that can determine the diameter of reinforcing bars even if they have the same node spacing.

The reinforcing bar inspection device according to the present disclosure comprises a three-dimensional information acquisition unit that acquires three-dimensional information of an inspection area in which reinforcing bars are arranged; a plane identification unit that uses the three-dimensional information to identify the reinforcing bar arrangement plane of the inspection target from the inspection area; an image conversion unit that converts the captured image of the inspection area into an orthogonal image that is oriented squarely to the reinforcing bar arrangement plane of the inspection target; a position detection unit that uses the orthogonal image to detect reinforcing bar position information in the reinforcing bar arrangement plane; an image extraction unit that extracts a reinforcing bar image from the orthogonal image based on the reinforcing bar position information in the reinforcing bar arrangement plane of the inspection target; and a scanning line consisting of pixel values of multiple pixels aligned along the longitudinal direction of the reinforcing bar in the orthogonal image. The system includes a feature calculation unit that acquires images for each pixel in a direction perpendicular to the longitudinal direction of the rebar, calculates feature values by frequency-converting the pixel values of the scanning lines, identifies the boundary position between the rebar and the background based on the results of comparing the calculated feature values with reference feature values, and identifies the first rebar diameter using the boundary position; an inference unit that infers the second rebar diameter using a learning model that outputs the rebar diameter when a rebar image is input; a rebar diameter determination unit that determines the rebar diameter of the measurement results using the first rebar diameter and the second rebar diameter; and a measurement result information generation unit that generates and outputs measurement result information for displaying the rebar diameter of the measurement results.

According to the present disclosure, a scanning line consisting of pixel values of multiple pixels aligned along the longitudinal direction of the rebar is acquired for each pixel in a direction perpendicular to the longitudinal direction of the rebar, the pixel values of the scanning line are frequency-converted to calculate feature values, the calculated feature values are compared with reference feature values, and the boundary position between the rebar and the background is identified based on the results. The boundary position is used to identify the first rebar diameter, a learning model is used to infer the second rebar diameter, and the first rebar diameter and second rebar diameter are used to determine the rebar diameter of the measurement results. As a result, the rebar arrangement inspection device according to the present disclosure can determine the rebar diameter even for rebars with the same node spacing.

1 is a block diagram showing the configuration of a reinforcement bar arrangement inspection device according to a first embodiment. FIG. 1 is a flowchart showing a reinforcement bar arrangement inspection method according to the first embodiment. FIG. 10 is an explanatory diagram showing an overview of the photographing process of the reinforcement inspection area. 4A, 4B, and 4C are graphs showing the relationship between a function indicating the plane of each step in a structure in an inspection area and outliers and inliers. FIG. 10 is an explanatory diagram showing a process of converting a captured image into a normal image. FIG. 10 is an explanatory diagram showing a process for detecting the position of a reinforcing bar in a normal image. 7A, 7B, and 7C are explanatory diagrams showing the process of identifying line segments corresponding to reinforcing bar portions in a mask image. FIG. 10 is an explanatory diagram showing a process of extracting a partial image of a reinforcing bar from a normalized image. 9A and 9B are explanatory diagrams showing a process of determining the diameter of a reinforcing bar using a normalized image. 10A and 10B are explanatory diagrams showing a process of calculating the correlation between a frequency conversion spectrum obtained from a normalized image and a reference spectrum. FIG. 10 is an explanatory diagram showing a lap joint in a reinforcement inspection area. 12A and 12B are block diagrams showing a hardware configuration for realizing the functions of the bar arrangement inspection device according to the first embodiment.

Embodiment 1.
FIG. 1 is a block diagram showing the configuration of a reinforcement inspection device 1 according to a first embodiment. The reinforcement inspection device 1 inspects rebars arranged in a reinforcement plane based on an image of the reinforcement plane to be inspected captured by a stereo camera 2, and displays the inspection results on a display unit 3. The reinforcement plane to be inspected is a plane included in an architectural or civil engineering structure constructed with a plurality of rebars arranged as a framework. For example, the reinforcement inspection device 1 determines the diameter (type; name) of rebars in the reinforcement plane, and inspects the number of rebars of the determined diameter arranged and the spacing between adjacent rebars. A tablet terminal, smartphone, or personal computer (PC) is used as the reinforcement inspection device 1.

There are 16 types of rebars based on the JIS standard. These 16 types of rebars are given "designations": D4, D5, D6, D8, D10, D13, D16, D19, D22, D25, D29, D32, D35, D38, D41, and D51. The designation indicates the rounded diameter of the rebar's nominal diameter. For example, the nominal diameter of a rebar designated D10 is 9.53 mm, the nominal diameter of a rebar designated D13 is 12.7 mm, and the nominal diameter of a rebar designated D16 is 15.9 mm. Note that rebars generally used as the framework of buildings are D10 or higher.

A "maximum average knot spacing" is specified for each of the 16 types of rebar. For example, the maximum average knot spacing for D10 rebar is 6.7 mm, the maximum average knot spacing for D13 rebar is 8.9 mm, and the maximum average knot spacing for D16 rebar is 11.1 mm. As such, only the maximum average knot spacing is specified for the 16 types of rebar, and the actual knot spacing for rebar varies depending on the manufacturer or production lot. Furthermore, threaded rebars may have different diameters even if the knot spacing is the same.

The reinforcing bar inspection device 1 acquires a scanning line consisting of pixel values of multiple pixels aligned along the longitudinal direction of the reinforcing bar in an image of the reinforcing bar plane captured by the stereo camera 2, for each pixel in a direction perpendicular to the longitudinal direction of the reinforcing bar, and calculates feature values by frequency-converting the pixel values of the scanning line. The reinforcing bar inspection device 1 then identifies the boundary position between the reinforcing bar and the background based on the results of comparing the calculated feature values with reference feature values, and uses the boundary position to identify the reinforcing bar diameter. This reinforcing bar diameter is designated as the first reinforcing bar diameter.

Furthermore, the reinforcing bar inspection device 1 uses a learning model that outputs the reinforcing bar diameter when an image of the reinforcing bar is input, and infers the reinforcing bar diameter. This reinforcing bar diameter is designated as the second reinforcing bar diameter. The reinforcing bar inspection device 1 determines the reinforcing bar diameter of the measurement result using the first reinforcing bar diameter and the second reinforcing bar diameter.
In this way, the reinforcing bar inspection device 1 uses images of the reinforcing bars to compare the reinforcing bar diameters determined by the above two methods and determines the reinforcing bar diameter of the final measurement result, so it is possible to accurately determine the reinforcing bar diameter even for reinforcing bars with the same node spacing.

The stereo camera 2 is a camera unit having a left camera unit, a right camera unit, and a three-dimensional information generator (not shown). The left camera unit captures a left viewpoint image seen from the left side. The right camera unit captures a right viewpoint image seen from the right side. The three-dimensional information generator performs stereo matching processing on the left viewpoint image and the right viewpoint image to generate a three-dimensional image as three-dimensional information. The display unit 3 is a display device provided in the reinforcement inspection device 1.

The display unit 3 is, for example, a liquid crystal display (LCD) or an organic electroluminescence (EL) display device. The memory unit 4 stores rebar characteristic information for various rebar diameters. The rebar characteristic information is the aforementioned reference characteristic amount, and is a reference frequency transformation spectrum obtained by frequency transforming the pixel values of a scanning line consisting of multiple pixels aligned along the longitudinal direction of the rebar. For example, the frequency transformation is a fast Fourier transform (hereinafter referred to as FFT).

The reinforcement inspection device 1 includes a three-dimensional information acquisition unit 11, a plane identification unit 12, an image conversion unit 13, a position detection unit 14, an image extraction unit 15, a feature calculation unit 16, an inference unit 17, a rebar diameter determination unit 18, and a measurement result information generation unit 19. The three-dimensional information acquisition unit 11 acquires, as three-dimensional information, three-dimensional images of the reinforcement plane to be inspected taken by the stereo camera 2.

The plane identification unit 12 uses the three-dimensional information acquired by the three-dimensional information acquisition unit 11 to identify the reinforcement plane to be inspected from the inspection area. For example, the plane identification unit 12 detects plane candidates containing multiple three-dimensional points from a three-dimensional image of a structure including a reinforcement plane, and calculates the number of three-dimensional points whose distance from the plane candidate is less than a threshold for each plane candidate. The plane identification unit 12 then identifies the plane candidate with the largest number of three-dimensional points as the reinforcement plane to be inspected from among the multiple plane candidates. The reinforcement plane to be inspected is the plane located at the forefront of the structure.

The image converter 13 converts the captured image of the inspection area into a normal image that faces the reinforcement plane of the inspection target. The normal image is an image in which the distance between the stereo camera 2 and the reinforcement plane of the inspection target is constant and the reinforcement plane of the inspection target is directly facing the stereo camera 2.
For example, the image conversion unit 13 estimates a homography transformation matrix using the position coordinates of four points at the four corners of a rectangle formed by reinforcing bars arranged in a lattice pattern on a reinforcement plane. Then, the image conversion unit 13 converts the captured image into a normal-oriented image based on the homography transformation matrix.
All pixels in the orthogonal image are scaled so that the distance from the stereo camera 2 is constant. As a result, in the orthogonal image, differences in the size of the rebars according to the distance between the stereo camera 2 and the reinforcement plane of the inspection target are corrected.

The position detection unit 14 detects rebar position information in the reinforcement plane of the inspection target using the orientated image generated by the image conversion unit 13. For example, the position detection unit 14 binarizes the orientated image using a threshold value to generate a mask image in which areas other than the rebar are masked, and detects the rebar positions in the reinforcement plane of the inspection target contained in the orientated image by counting the pixels of the rebar areas in the mask image.

The image extraction unit 15 extracts rebar images from the orientated image based on the rebar position information in the reinforcement plane of the inspection target. For example, the image extraction unit 15 identifies the rebar image for each rebar in the orientated image based on the rebar position information in the reinforcement plane of the inspection target, and sequentially extracts multiple partial images of the same size in the longitudinal direction of the rebar from the rebar image. The multiple partial images extracted by the image extraction unit 15 are output to the inference unit 17 for each rebar image.

The feature calculation unit 16 acquires a scanning line consisting of pixel values of multiple pixels aligned along the longitudinal direction of the rebar in the normalized image, for each pixel in a direction perpendicular to the longitudinal direction of the rebar, and calculates feature values by frequency-converting the pixel values of the scanning line using FFT. The feature calculation unit 16 then identifies the boundary position between the rebar and the background based on the result of comparing the calculated feature values with reference feature values, and identifies the diameter of the first rebar using the boundary position.

The inference unit 17 infers the second rebar diameter using a learning model that outputs a rebar diameter when a rebar image is input. For example, the learning model is a machine learning model that infers the second rebar diameter when multiple partial images extracted by the image extraction unit 15 for each rebar image are input.
The learning algorithm may be, for example, deep learning, neural networks, genetic programming, functional logic programming, or support vector machines.

The rebar diameter determination unit 18 determines the rebar diameter of the measurement result using the first rebar diameter and the second rebar diameter. For example, the rebar diameter determination unit 18 determines the rebar diameter of the first rebar diameter and the second rebar diameter that satisfies a predetermined determination condition as the rebar diameter of the measurement result. The determination condition is a condition such that the rebar diameter that is the same as or closest to the rebar diameter specified in the JIS standard is determined to be the measurement result. The rebar diameter determination unit 18 may also determine the rebar diameter obtained by statistically processing the first rebar diameter and the second rebar diameter as the rebar diameter of the measurement result. Statistical processing includes, for example, calculating the average, maximum, or minimum value.

The measurement result information generation unit 19 generates and outputs measurement result information for displaying the rebar diameter of the measurement results. For example, the measurement result information generation unit 19 outputs display control information for displaying the rebar diameter of the measurement results to the display unit 3 as measurement result information. The display unit 3 displays the rebar diameter of the measurement results using the measurement result information.

FIG. 2 is a flowchart showing the reinforcement bar arrangement inspection method according to the first embodiment.
The three-dimensional information acquisition unit 11 acquires three-dimensional information of the inspection area where reinforcing bars are arranged (step ST1). Fig. 3 is an explanatory diagram showing an overview of the photographing process of the reinforcing bar arrangement inspection area, and shows a three-dimensional image 2C of a plurality of reinforcing bar arrangement planes where reinforcing bars 20 are arranged. As shown in Fig. 3, the stereo camera 2 captures a left viewpoint image 2A and a right viewpoint image 2B, and generates a three-dimensional image 2C using the left viewpoint image 2A and the right viewpoint image 2B. The three-dimensional information acquisition unit 11 acquires the three-dimensional image 2C from the stereo camera 2.

The plane identification unit 12 uses the three-dimensional image 2C to identify the reinforcement plane to be inspected from multiple reinforcement planes in the inspection area (step ST2). Here, in the three-dimensional image 2C, the closer the rebar 20 is to the stereo camera 2, i.e., the closer it is to the front, the larger it appears, and the farther it is from the stereo camera 2, the smaller it appears. The fact that the rebar 20 appears larger means that there are more three-dimensional points corresponding to the rebar 20 in the three-dimensional image 2C. Conversely, the fact that the rebar 20 appears smaller means that there are fewer three-dimensional points corresponding to the rebar 20 in the three-dimensional image 2C. Furthermore, the closer the plane is to the front, the fewer obstacles, such as rebars on the rear side, that block the view from the stereo camera 2, making occlusion less likely to occur.

The plane identification unit 12 therefore identifies the foreground plane using, for example, the RANSAC (RANdom Sample Consensus) method. The plane identification unit 12 repeatedly estimates a function indicating plane candidates using a group of 3D points randomly detected from the 3D image 2C. Figures 4A, 4B, and 4C are graphs showing the relationship between the functions P(1), P(2), and P(3) indicating the planes of each layer and the outliers 31 and inliers 32, and show 3D points on the XY coordinate plane. Outliers 31 are 3D points that are not within the allowable range, and inliers 32 are 3D points that are within the allowable range.

In the RANSAC method, the number of 3D points that are inliers 32 is counted for each parameter representing the functions P(1), P(2), and P(3), and the parameter with the highest count is determined to be the optimal parameter. In other words, the plane candidate represented by the function to which the determined parameters are applied is determined to be the estimated foreground plane. As is clear from Figures 4A, 4B, and 4C, the parameters representing function P(3) have the largest number of 3D points that are inliers 32, so the plane identification unit 12 identifies the plane candidate represented by function P(3) as the foreground plane.

Next, the image conversion unit 13 converts the image of the inspection area including the reinforcement plane of the inspection target captured by the stereo camera 2 into a normalized image (step ST3).
5 is an explanatory diagram showing the process of converting a captured image into a normal-oriented image, and shows only the foreground reinforcement plane in the three-dimensional image 2C and the normal-oriented image 2D. The image conversion unit 13 specifies four points at the four corners of any rectangle among the reinforcing bars 20 arranged in a grid pattern on the reinforcement plane to be inspected in the three-dimensional image 2C, and estimates a homography transformation matrix that gives the rectangle a shape viewed from the front of the stereo camera 2. Then, the image conversion unit 13 converts the image shown in the three-dimensional image 2C into the normal-oriented image 2D based on the homography transformation matrix.

The position detection unit 14 uses the orientated image to detect rebar position information on the reinforcement bar arrangement plane (step ST4-1). Figure 6 is an explanatory diagram showing the process of detecting the position of rebar in the orientated image. As shown in Figure 6, the position detection unit 14 converts the orientated image 2D, which the image conversion unit 13 converted from the three-dimensional image 2C, into a mask image 51. The mask image 51 is an image with binary pixel values, where areas representing rebar in the orientated image 2D are represented by "1" and other areas by "0". Areas with a pixel value of "1" are white pixel areas, and areas with a pixel value of "0" are black pixel areas.

For example, the position detection unit 14 uses the left viewpoint image 2A and the right viewpoint image 2B to calculate the amount of misalignment in the image of rebars at the same position as left-right parallax, identifies the image portion of the rebar using the parallax, and generates a mask image 51 by assigning a pixel value of "1" to the identified image portion and a pixel value of "0" to the other image portions. The position detection unit 14 then detects the position in the mask image 51 where the count of white pixels is equal to or greater than a threshold as the position of the rebar. The position detection unit 14 can detect the X-direction position and Y-direction position of the rebar by rotating the mask image 51 and counting the number of white pixels. That is, the positions of rebars lined up vertically and rebars lined up horizontally are detected in the orientated image 2D.

Next, the position detection unit 14 performs a process to identify line segments corresponding to rebars in the mask image 51. Figures 7A, 7B, and 7C are explanatory diagrams showing the process of identifying line segments corresponding to rebar portions in the mask image. As shown in Figure 7A, when mask image 51 has white pixel regions A, B, and C, it is unclear which region corresponds to the image of the rebar. Therefore, the position detection unit 14 calculates the length D(1) of region A, the length D(2) of region B, and the length D(3) of region C on the coordinate axis passing through the white pixel regions A, B, and C, as well as the distance D(4) between region A and region B and the distance D(5) between region B and region C.

As shown in FIG. 7B , the position detection unit 14 determines that areas with a distance equal to or greater than the threshold, between distances D(4) and D(5), are not rebars, and determines that areas with a distance less than the threshold are image areas corresponding to the same rebar. For example, the position detection unit 14 determines that the image area indicated by line segment 52 with length D(6) that passes through area A and area B, where the distance D(4) is less than the threshold, is an image area corresponding to the same rebar. Furthermore, because D(5) is equal to or greater than the threshold, the position detection unit 14 determines that there is no image area corresponding to a rebar between area B and area C in the mask image 51. The position detection unit 14 assigns position information of the portion of the mask image 51 determined to be an image area of a rebar to the orientated image 2D and outputs this orientated image 2D to the image extraction unit 15 and the feature calculation unit 16.

The image extraction unit 15 extracts rebar images from the orientated image based on the rebar position information in the reinforcement bar arrangement plane of the inspection target (step ST4-2). Figure 8 is an explanatory diagram showing the process of extracting partial images of rebars 20 from the orientated image. As shown in Figure 8, the image extraction unit 15 extracts orientated image 2E of rebars 20 along their longitudinal direction from the orientated image. Furthermore, the image extraction unit 15 sequentially extracts partial images 2F(1), 2F(2), and 2F(3) of the same size from the orientated image 2E and outputs them to the inference unit 17. Partial images 2F(1), 2F(2), and 2F(3) are, for example, square images with the same number of pixels in both the vertical and horizontal directions. The image extraction unit 15 extracts partial images for all rebars 20 in the orientated image 2D.

The feature calculation unit 16 identifies the boundary position between the rebar and the background based on the result of comparing the feature calculated from the normalized image with the reference feature, and uses the identified boundary position to identify the first rebar diameter (step ST5-1). Figures 9A and 9B are explanatory diagrams showing the process of determining the rebar diameter using the normalized image. Figure 9A shows a scanning line i (i = 0 to n, 0 < k < n) consisting of the brightness values of multiple pixels aligned along the longitudinal direction of the rebar in the normalized image 2E. As shown in Figure 9A, the feature calculation unit 16 acquires scanning line i for each pixel in the direction perpendicular to the longitudinal direction of the rebar, and calculates feature values (frequency transformation spectrum) obtained by frequency-converting the brightness values of scanning line i using FFT for each pixel in the direction perpendicular to the longitudinal direction of the rebar.

Figure 9B shows the amount of change in luminance value of scanning line i=0 in the background part outside the rebar in the normalized image 2E and a frequency conversion spectrum obtained by frequency converting the luminance value of scanning line i=0; the amount of change in luminance value of scanning line i=k on the rebar and a frequency conversion spectrum obtained by frequency converting the luminance value of scanning line i=k; the amount of change in luminance value of scanning line i=n in the background part outside the rebar and a frequency conversion spectrum obtained by frequency converting the luminance value of scanning line i=n; and a reference frequency conversion spectrum, which is rebar characteristic information. If there is a difference in luminance value on scanning line i, the change in luminance value becomes a peak in the frequency conversion spectrum.

In the orthogonal image 2E, the feature calculation unit 16 acquires scanning lines i for each pixel from scanning line i=0 to scanning line i=n in the portion corresponding to the outside of the rebar in a direction perpendicular to the longitudinal direction of the rebar (from left to right in Figure 9A). Next, the feature calculation unit 16 calculates a frequency conversion spectrum by frequency converting the luminance values of scanning line i for each pixel. Next, the feature calculation unit 16 compares the frequency conversion spectrum by frequency converting the luminance values of scanning line i with the rebar feature information stored in the memory unit 4. The rebar feature information is a frequency spectrum by frequency converting the luminance values of scanning lines on the rebar that were calculated in advance in the rebar image.

For example, the feature calculation unit 16 determines that, among the frequency transformation spectra calculated for each pixel in the orientated image 2E, a spectrum similar to the frequency transformation spectrum that is the rebar feature information corresponds to the scanning line i on the rebar. Next, the feature calculation unit 16 identifies the scanning line i that corresponds to the boundary position between the rebar and the background (the boundary on the left side and the boundary on the right side in FIG. 9A ), and calculates the distance between the identified scanning lines i. The feature calculation unit 16 then converts the distance between the scanning lines i into a length in real space to calculate the first rebar diameter. For example, if the stereo camera 2 is orientated with the rebar plane at a shooting distance of 1.5 meters, and the captured image is 200 opx in the horizontal direction, this corresponds to 0.75 mm/px in real space.

The rebar characteristic information may include a frequency transform spectrum of the luminance values of scan line i calculated using a captured image in which the rebar and shadows are prominently visible due to the effects of sunlight, or may include a frequency transform spectrum of the luminance values of scan line i calculated using an image in which the rebar glows white using flash photography or the like. By using such rebar characteristic information, the feature calculation unit 16 can accurately identify the boundary between the rebar and the background even in rebar images in which shadows or the white glow of the rebar is present.

The starting point for scanning the rebar image and the ending point for ending the scan are determined by, for example, setting a scanning range that is the maximum rebar diameter specified in the JIS standard plus a certain margin, and setting the rebar image within that scanning range. This makes one side of the scanning range the starting point and the other the ending point. The starting point for scanning the rebar image and the ending point for ending the scan may also be set by the user using an input device (not shown).

Figure 10 is an explanatory diagram showing the process of calculating the correlation between a frequency transformation spectrum obtained from a normalized image and a reference spectrum. In Figure 10, the solid line spectrum is the reference frequency transformation spectrum, which is rebar characteristic information, and the dashed line spectrum is the frequency transformation spectrum obtained from the normalized image. For example, the feature calculation unit 16 performs a correlation calculation (1) that evaluates the correlation between these spectra across the entire frequency range, a correlation calculation (2) that evaluates the correlation between the first peaks that appear first in the spectrum, and a correlation calculation (3) that evaluates the correlation between the second peaks that appear next in the spectrum.

In correlation calculation (1), the feature calculation unit 16 calculates a correlation score (1) indicating the correlation between the two spectra within a correlation evaluation range expressed as (min(PeakA, PeakB)-4) to 256, where PeakA is the intensity of the first peak that first appears in the spectrum that is the rebar characteristic information, and PeakB is the intensity of the first peak that first appears in the spectrum obtained from the orientated image. In correlation calculation (2), the feature calculation unit 16 calculates a correlation score (2) indicating the correlation between the two spectra within a correlation evaluation range expressed as PeakA±10. In correlation calculation (3), the feature calculation unit 16 calculates a correlation score (3) indicating the correlation between the two spectra within a correlation evaluation range expressed as (PeakA×2)±10.

The feature calculation unit 16 determines a weighting coefficient according to each of the correlation score (1), correlation score (2), and correlation score (3). The feature calculation unit 16 then multiplies the determined weighting coefficient by the correlation score (1), correlation score (2), and correlation score (3), and then adds them together to determine the overall correlation score. The feature calculation unit 16 compares the overall correlation score with a threshold, and if the overall correlation score is equal to or greater than the threshold, determines that the spectrum obtained from the orientated image is similar to rebar feature information.

The feature calculation unit 16 also masks the portion of scanning line i where a rebar parallel to scanning line i intersects with another rebar, and calculates the feature by frequency converting the pixel values of the masked scanning line i. In the reinforcement plane, the rebars are arranged in a grid pattern, and there are portions where the rebars intersect with each other. In these portions, other rebars are arranged in a direction perpendicular to the longitudinal direction of the rebar along scanning line i, making it impossible to accurately identify the boundary between the rebar and the background. Therefore, the feature calculation unit 16 masks the portion where the rebar parallel to scanning line i intersects with another rebar, and calculates the feature by frequency converting the pixel values of the masked scanning line i. This enables the feature calculation unit 16 to accurately identify the boundary between the rebar and the background.

Furthermore, the feature calculation unit 16 may divide the scan line i into multiple sections along the longitudinal direction of the rebar, identify the boundary position between the rebar and the background based on the result of comparing the feature calculated for each section with a reference feature, and detect a lap splice made of multiple rebars and the length of the lap splice based on the number of rebars having a first rebar diameter identified using the boundary position. Figure 11 is an explanatory diagram showing a lap splice in the reinforcement inspection area. In Figure 11, rebar 20A and rebar 20B form a lap splice. The feature calculation unit 16 divides the scan line i into multiple sections along the longitudinal direction of the rebar. The memory unit 4 stores rebar feature information including a frequency conversion spectrum of the brightness values of the scan line corresponding to the spaces between the rebars in the lap splice.

For example, the area where another rebar intersects with rebar 20A is set as the feature acquisition interval, and the feature is calculated for each interval to calculate the first rebar diameter. At the lap joint between rebar 20A and rebar 20B, the frequency transform spectrum of the luminance values of scan line i at the boundary between rebar 20A and rebar 20B has a waveform that is different from the waveform at the boundary between the rebar and the background. In FIG. 11 , when scanning is started pixel by pixel from the rebar 20A side, a frequency transform spectrum corresponding to the boundary between rebar 20A and the background is acquired, followed by a frequency transform spectrum corresponding to the rebar 20A itself. As scanning continues, a frequency transform spectrum corresponding to the boundary between rebar 20A and rebar 20B is acquired. As scanning continues, a frequency transform spectrum corresponding to rebar 20B is acquired, followed by a frequency transform spectrum corresponding to the boundary between rebar 20B and the background. By comparing these frequency transformation spectra with the reinforcing bar characteristic information, the characteristic amount calculation unit 16 can determine that the reinforcing bar 20A is lap-spliced with the reinforcing bar 20B.
In addition, by determining the first reinforcing bar diameter for each of the above sections, the feature calculation unit 16 can identify the position where the reinforcing bar 20B begins to be detected and the position where only the reinforcing bar 20B begins to be detected, and therefore it is also possible to determine the length of the lap joint based on the identified positions.

In FIG. 2, when the inference unit 17 receives an image of a rebar extracted from the normalized image 2D, it infers the second rebar diameter using a learning model that outputs the rebar diameter (step ST5-2). The learning model is generated using learning data, which is, for example, a set of multiple partial images extracted from the rebar image and correct labels assigned to each partial image indicating the rebar diameter. The inference unit 17 performs inference for each partial image using the learning model, and calculates, for each partial image, the probability that a rebar with the second rebar diameter D is captured and the probability that something other than a rebar is captured.

The learning model calculates an inference result for the rebar image by averaging the inference results for all partial images of the same rebar image. For example, the inference result includes an average value of the probability that each type of rebar appears in the image and an average value of the probability that something other than a rebar appears in the image (NON). The inference unit 17 determines the second rebar diameter based on this inference result.
The learning model may be generated using training data including partial images of rebars in which the rebars and their shadows are prominent due to the influence of sunlight, and partial images in which the rebars are shining white due to flash photography, etc. By generating a learning model using such training data, the learning model can recognize the boundary between the rebars and the background even when there is a shadow or the white light of the rebars, and can accurately infer the diameter of the second rebar.

Next, the rebar diameter determination unit 18 determines the rebar diameter of the measurement result using the first rebar diameter and the second rebar diameter (step ST6). For example, the rebar diameter determination unit 18 determines the rebar diameter of the first rebar diameter and the second rebar diameter that satisfies a predetermined determination condition as the rebar diameter of the measurement result. If the determination condition is to determine that the measurement result is a value that is the same as or closest to the rebar diameter specified in the JIS standard, the rebar diameter determination unit 18 compares the first rebar diameter and the second rebar diameter with the rebar diameter specified in the JIS standard, and if either or both of the first rebar diameter and the second rebar diameter are the same as or closest to the rebar diameter specified in the JIS standard, determines that the rebar diameter is the measurement result. This makes it possible to accurately determine the rebar diameter.

The rebar diameter determination unit 18 may also determine the rebar diameter obtained by statistically processing the first rebar diameter and the second rebar diameter as the rebar diameter of the measurement result. Statistical processing includes, for example, calculating the average, maximum, or minimum value. The rebar diameter determination unit 18 determines the average value of the first rebar diameter and the second rebar diameter, or the maximum or minimum value of the first rebar diameter and the second rebar diameter, as the measurement result. Furthermore, if the rebar diameter obtained by statistically processing the first rebar diameter and the second rebar diameter satisfies the above-mentioned determination conditions, the rebar diameter determination unit 18 may determine that the rebar diameter is the determination result. This method also makes it possible to accurately determine the rebar diameter.

The measurement result information generation unit 19 generates and outputs measurement result information for displaying the rebar diameter of the measurement results (step ST7). For example, the measurement result information generation unit 19 outputs display control information for displaying the rebar diameter of the measurement results on an electronic whiteboard to the display unit 3 as measurement result information. The display unit 3 displays the rebar diameter of the measurement results on the electronic whiteboard, which is screen information. By referring to the rebar diameter of the measurement results displayed on the display unit 3, the inspector can appropriately check the rebar diameter of the measurement results.

The functions of the three-dimensional information acquisition unit 11, plane identification unit 12, image conversion unit 13, position detection unit 14, image extraction unit 15, feature calculation unit 16, inference unit 17, rebar diameter determination unit 18, and measurement result information generation unit 19 provided in the reinforcement bar inspection device 1 are realized by a processing circuit. In other words, the reinforcement bar inspection device 1 is equipped with a processing circuit for executing the processes of steps ST1 to ST7 shown in Figure 2. The processing circuit may be dedicated hardware, or it may be a CPU (Central Processing Unit) that executes a program stored in memory.

Figure 12A is a block diagram showing the hardware configuration that realizes the functions of the reinforcement inspection device 1. Figure 12B is a block diagram showing the hardware configuration that executes software that realizes the functions of the reinforcement inspection device 1. In Figures 12A and 12B, the input interface 100 is an interface that relays the three-dimensional data output from the stereo camera 2 to the reinforcement inspection device 1. The output interface 101 is an interface that relays the inspection results and the like output from the reinforcement inspection device 1 to the display unit 3.

When the processing circuit is the dedicated hardware processing circuit 102 shown in FIG. 12A, the processing circuit 102 may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a combination thereof.
The functions of the three-dimensional information acquisition unit 11, plane identification unit 12, image conversion unit 13, position detection unit 14, image extraction unit 15, feature calculation unit 16, inference unit 17, rebar diameter determination unit 18 and measurement result information generation unit 19 provided in the reinforcement inspection device 1 may be realized by separate processing circuits, or these functions may be realized together by a single processing circuit.

When the processing circuit is the processor 103 shown in FIG. 12B, the functions of the three-dimensional information acquisition unit 11, plane identification unit 12, image conversion unit 13, position detection unit 14, image extraction unit 15, feature calculation unit 16, inference unit 17, rebar diameter determination unit 18, and measurement result information generation unit 19 provided in the reinforcement bar inspection device 1 are realized by software, firmware, or a combination of software and firmware. The software or firmware is written as a program and stored in the memory 104.

The processor 103 reads out and executes the programs stored in the memory 104, thereby realizing the functions of the three-dimensional information acquisition unit 11, plane identification unit 12, image conversion unit 13, position detection unit 14, image extraction unit 15, feature calculation unit 16, inference unit 17, reinforcing bar diameter determination unit 18, and measurement result information generation unit 19 provided in the reinforcement bar inspection device 1. For example, the reinforcement bar inspection device 1 includes the memory 104 for storing a program that, when executed by the processor 103, results in the execution of the processes of steps ST1 to ST7 shown in FIG.
These programs cause the computer to execute the procedures or methods of processing performed by the three-dimensional information acquisition unit 11, plane identification unit 12, image conversion unit 13, position detection unit 14, image extraction unit 15, feature calculation unit 16, inference unit 17, rebar diameter determination unit 18, and measurement result information generation unit 19. The memory 104 may be a computer-readable storage medium that stores programs for causing the computer to function as the three-dimensional information acquisition unit 11, plane identification unit 12, image conversion unit 13, position detection unit 14, image extraction unit 15, feature calculation unit 16, inference unit 17, rebar diameter determination unit 18, and measurement result information generation unit 19.

Memory 104 may be, for example, non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), or EEPROM (Electrically-EPROM), as well as magnetic disks, flexible disks, optical disks, compact disks, minidisks, and DVDs.

Some of the functions of the three-dimensional information acquisition unit 11, plane identification unit 12, image conversion unit 13, position detection unit 14, image extraction unit 15, feature calculation unit 16, inference unit 17, rebar diameter determination unit 18 and measurement result information generation unit 19 provided in the reinforcement inspection device 1 may be realized by dedicated hardware, and other functions may be realized by software or firmware.
For example, the three-dimensional information acquisition unit 11 realizes its functions by a processing circuit 102, which is dedicated hardware, and the plane identification unit 12, image conversion unit 13, position detection unit 14, image extraction unit 15, feature amount calculation unit 16, inference unit 17, rebar diameter determination unit 18, and measurement result information generation unit 19 realize their functions by a processor 103 reading and executing programs stored in a memory 104. In this way, the processing circuit can realize the above functions by hardware, software, firmware, or a combination of these.

The reinforcement inspection device 1 may also be equipped with a learning device that generates a learning model used by the inference unit 17. The reinforcement inspection device 1 may also acquire a learning model stored in an external device through data communication over a network. Furthermore, the learning model may use, as learning data, partial images with correct labels input using an operation unit (not shown in FIG. 1). This allows the inference unit 17 to infer the rebar diameter using a learning model generated for each site, improving the accuracy of reinforcement inspection.

As described above, the reinforcing bar inspection device 1 according to embodiment 1 acquires a scan line consisting of the pixel values of multiple pixels aligned along the longitudinal direction of the reinforcing bar, pixel by pixel in a direction perpendicular to the longitudinal direction of the reinforcing bar, calculates feature values by frequency-converting the pixel values of the scan line, identifies the boundary position between the reinforcing bar and the background based on the results of comparing the calculated feature values with reference feature values, identifies the first reinforcing bar diameter using the boundary position, infers the second reinforcing bar diameter using a learning model, and determines the reinforcing bar diameter of the measurement results using the first reinforcing bar diameter and the second reinforcing bar diameter. This allows the reinforcing bar inspection device 1 to determine the reinforcing bar diameter even for reinforcing bars with the same node spacing.

In the reinforcing bar inspection device 1 according to embodiment 1, the feature calculation unit 16 acquires a scanning line consisting of the brightness values of multiple pixels aligned along the longitudinal direction of the reinforcing bar in the orthogonal image for each pixel in the direction perpendicular to the longitudinal direction of the reinforcing bar, and calculates a feature by frequency-converting the brightness values of the scanning line. This enables the reinforcing bar inspection device 1 to accurately measure the diameter of the first reinforcing bar using the orthogonal image.

In the reinforcing bar inspection device 1 according to embodiment 1, the feature calculation unit 16 masks the portion of the scan line where a reinforcing bar parallel to the scan line intersects with another reinforcing bar, and calculates the feature by frequency-converting the pixel values of the masked scan line. By masking the portion where the reinforcing bars intersect, the feature calculation unit 16 can accurately measure the diameter of the first reinforcing bar using the orthogonal image.

In the reinforcing bar inspection device 1 according to embodiment 1, the feature calculation unit 16 divides the scan line into multiple sections along the longitudinal direction of the reinforcing bars, identifies the boundary position between the reinforcing bars and the background based on the results of comparing the feature calculated for each section with a reference feature, and detects lap splices made of multiple reinforcing bars and the lengths of those lap splices based on the number of reinforcing bars having the first reinforcing bar diameter identified using the boundary position. This allows the feature calculation unit 16 to detect lap splices made of reinforcing bars and the lengths of those lap splices.

In the reinforcing bar inspection device 1 according to embodiment 1, the reinforcing bar diameter determination unit 18 determines the reinforcing bar diameter of the measurement result as the reinforcing bar diameter of the first reinforcing bar diameter and the second reinforcing bar diameter that satisfies predetermined determination conditions. This allows the reinforcing bar diameter determination unit 18 to accurately determine the reinforcing bar diameter of the measurement result using the first reinforcing bar diameter actually measured from the captured image of the inspection area and the second reinforcing bar diameter inferred by the learning model.

In the reinforcing bar inspection device 1 according to embodiment 1, the reinforcing bar diameter determination unit 18 determines the reinforcing bar diameter obtained by statistically processing the first reinforcing bar diameter and the second reinforcing bar diameter as the reinforcing bar diameter of the measurement result. This allows the reinforcing bar diameter determination unit 18 to accurately determine the reinforcing bar diameter of the measurement result using the first reinforcing bar diameter actually measured from the captured image of the inspection area and the second reinforcing bar diameter inferred by the learning model.

In the reinforcement inspection method according to the first embodiment, the three-dimensional information acquisition unit 11 acquires three-dimensional information of an inspection area in which reinforcing bars are arranged, the plane identification unit 12 uses the three-dimensional information to identify the reinforcement plane of the inspection target from the inspection area, the image conversion unit 13 converts the captured image of the inspection area into an orthogonal image that is oriented squarely toward the reinforcement plane of the inspection target, the position detection unit 14 detects reinforcing bar position information in the orthogonal image on the reinforcement plane, the image extraction unit 15 extracts reinforcing bar images from the orthogonal image based on the reinforcing bar position information in the reinforcement plane of the inspection target, and the feature calculation unit 16 calculates the feature amount of the reinforcing bars arranged along the longitudinal direction of the reinforcing bars in the orthogonal image. A scanning line consisting of pixel values of multiple pixels is acquired for each pixel in a direction perpendicular to the longitudinal direction of the rebar, and feature values are calculated by frequency-converting the pixel values of the scanning line. The calculated feature values are compared with reference feature values to identify the boundary position between the rebar and the background based on the results. The boundary position is used to identify the first rebar diameter. An inference unit 17 infers the second rebar diameter using a learning model that outputs the rebar diameter when a rebar image is input. A rebar diameter determination unit 18 determines the rebar diameter of the measurement results using the first and second rebar diameters. A measurement result information generation unit 19 generates and outputs measurement result information indicating the rebar diameter of the measurement results. This provides a rebar arrangement inspection method that can determine the diameter of rebars even if they have the same node spacing.

In addition, any of the components of the embodiments may be modified or omitted.

Aspects of the present disclosure are described in the appendix below.
(Appendix 1)
a three-dimensional information acquisition unit that acquires three-dimensional information of an inspection area in which reinforcing bars are arranged;
a plane specifying unit that specifies a reinforcement plane to be inspected from the inspection area using the three-dimensional information;
an image conversion unit that converts the captured image of the inspection area into a normal image that faces the reinforcement plane of the inspection target;
a position detection unit that detects reinforcing bar position information in the reinforcement bar arrangement plane using the orthogonal image;
a feature calculation unit that acquires a scanning line consisting of pixel values of a plurality of pixels aligned along the longitudinal direction of the rebar in the orthogonal image for each pixel in a direction perpendicular to the longitudinal direction of the rebar, calculates feature values by frequency-converting the pixel values of the scanning line, identifies a boundary position between the rebar and the background based on the result of comparing the calculated feature values with a reference feature value, and identifies a first rebar diameter using the boundary position;
an image extraction unit that extracts a reinforcing bar image from the orthogonal image based on the reinforcing bar position information in the reinforcing bar arrangement plane of the inspection target;
an inference unit that infers a second reinforcing bar diameter using a learning model that outputs a reinforcing bar diameter when the reinforcing bar image is input;
a reinforcing bar diameter determination unit that determines the reinforcing bar diameter of the measurement result using the first reinforcing bar diameter and the second reinforcing bar diameter;
a measurement result information generating unit that generates and outputs measurement result information indicating the reinforcing bar diameter of the measurement result.
(Appendix 2)
The reinforcing bar arrangement inspection device according to claim 1, wherein the feature amount calculation unit acquires a scanning line consisting of brightness values of a plurality of pixels aligned along the longitudinal direction of the reinforcing bar in the opposed image for each pixel in a direction perpendicular to the longitudinal direction of the reinforcing bar, and calculates a feature amount by frequency-converting the brightness values of the scanning line.
(Appendix 3)
The reinforcement bar inspection device according to claim 1 or 2, wherein the feature amount calculation unit masks a portion of the scanning line where a reinforcing bar parallel to the scanning line intersects with another reinforcing bar, and calculates the feature amount by frequency-converting pixel values of the masked scanning line.
(Appendix 4)
The reinforcing bar inspection device according to any one of Supplementary Note 1 to Supplementary Note 3, characterized in that the feature calculation unit divides the scanning line into a plurality of sections along the longitudinal direction of the reinforcing bar, identifies a boundary position between the reinforcing bar and the background based on a result of comparing the feature calculated for each section with a reference feature, and detects lap splices made of a plurality of reinforcing bars and the length of the lap splice based on the number of reinforcing bars having the first reinforcing bar diameter identified using the boundary position.
(Appendix 5)
The reinforcing bar inspection device according to any one of Supplementary Note 1 to Supplementary Note 4, wherein the reinforcing bar diameter determination unit determines the reinforcing bar diameter of the measurement result as being one of the first reinforcing bar diameter and the second reinforcing bar diameter that satisfies a predetermined determination condition.
(Appendix 6)
The reinforcing bar inspection device according to any one of Supplementary Note 1 to Supplementary Note 4, wherein the reinforcing bar diameter determination unit determines the reinforcing bar diameter obtained by statistically processing the first reinforcing bar diameter and the second reinforcing bar diameter as the reinforcing bar diameter of the measurement result.
(Appendix 7)
A reinforcement inspection method using a reinforcement inspection device,
a step in which a three-dimensional information acquisition unit acquires three-dimensional information of an inspection area in which reinforcing bars are arranged;
a step in which a plane specifying unit specifies a reinforcement plane to be inspected from the inspection area using the three-dimensional information;
an image conversion unit converting the captured image of the inspection area into a normal image that faces the reinforcement plane of the inspection target;
A step in which a position detection unit detects reinforcing bar position information in the reinforcement bar arrangement plane using the orthogonal image;
An image extraction unit extracts a reinforcing bar image from the orthogonal image based on the reinforcing bar position information in the reinforcing bar arrangement plane of the inspection target;
a feature calculation unit acquires a scanning line consisting of pixel values of a plurality of pixels aligned along the longitudinal direction of the rebar in the orthogonal image for each pixel in a direction perpendicular to the longitudinal direction of the rebar, calculates a feature by frequency-converting the pixel values of the scanning line, identifies a boundary position between the rebar and the background based on a result of comparing the calculated feature with a reference feature, and identifies a first rebar diameter using the boundary position;
an inference unit inferring a second reinforcing bar diameter using a learning model that outputs a reinforcing bar diameter when the reinforcing bar image is input;
A step in which a reinforcing bar diameter determination unit determines the reinforcing bar diameter of the measurement result using the first reinforcing bar diameter and the second reinforcing bar diameter;
a step of generating and outputting measurement result information indicating the reinforcing bar diameter of the measurement result by a measurement result information generating unit.
(Appendix 8)
A program for causing a computer to function as the reinforcement inspection device according to any one of Supplementary Note 1 to Supplementary Note 6.

1 Reinforcement inspection device, 2 Stereo camera, 2A Left viewpoint image, 2B Right viewpoint image, 2C Three-dimensional image, 2D, 2E Oriented images, 2F(1) to 2F(3) Partial images, 3 Display unit, 4 Memory unit, 11 Three-dimensional information acquisition unit, 12 Plane identification unit, 13 Image conversion unit, 14 Position detection unit, 15 Image extraction unit, 16 Feature calculation unit, 17 Inference unit, 18 Reinforcement diameter determination unit, 19 Measurement result information generation unit, 20, 20A, 20B Reinforcement bar, 31 Outlier, 32 Inlier, 51 Mask image, 52 Line segment, 100 Input interface, 101 Output interface, 102 Processing circuit, 103 Processor, 104 Memory.

Claims (8)

a three-dimensional information acquisition unit that acquires three-dimensional information of an inspection area in which reinforcing bars are arranged;
a plane specifying unit that specifies a reinforcement plane to be inspected from the inspection area using the three-dimensional information;
an image conversion unit that converts the captured image of the inspection area into a normal image that faces the reinforcement plane of the inspection target;
a position detection unit that detects reinforcing bar position information in the reinforcement bar arrangement plane using the orthogonal image;
an image extraction unit that extracts a reinforcing bar image from the orthogonal image based on the reinforcing bar position information in the reinforcing bar arrangement plane of the inspection target;
a feature calculation unit that acquires a scanning line consisting of pixel values of a plurality of pixels aligned along the longitudinal direction of the rebar in the orthogonal image for each pixel in a direction perpendicular to the longitudinal direction of the rebar, calculates feature values by frequency-converting the pixel values of the scanning line, identifies a boundary position between the rebar and the background based on the result of comparing the calculated feature values with a reference feature value, and identifies a first rebar diameter using the boundary position;
an inference unit that infers a second reinforcing bar diameter using a learning model that outputs a reinforcing bar diameter when the reinforcing bar image is input;
a reinforcing bar diameter determination unit that determines the reinforcing bar diameter of the measurement result using the first reinforcing bar diameter and the second reinforcing bar diameter;
a measurement result information generating unit that generates and outputs measurement result information for displaying the reinforcing bar diameter of the measurement result. The reinforcing bar arrangement inspection device according to claim 1, characterized in that the feature calculation unit acquires the scanning line, which is made up of brightness values of a plurality of pixels lined up along the longitudinal direction of the reinforcing bar in the facing image, for each pixel in a direction perpendicular to the longitudinal direction of the reinforcing bar, and calculates feature values by frequency-converting the brightness values of the scanning line. 3. The reinforcement bar inspection device according to claim 1, wherein the feature calculation unit masks a portion of the scanning line where a reinforcing bar parallel to the scanning line intersects with another reinforcing bar, and calculates the feature by frequency-converting pixel values of the masked scanning line. The bar arrangement inspection device according to claim 3, characterized in that the feature calculation unit divides the scanning line into a plurality of sections along the longitudinal direction of the rebar, identifies the boundary position between the rebar and the background based on the result of comparing the feature calculated for each section with a reference feature, and detects lap splices made of a plurality of rebars and the length of the lap splices based on the number of rebars having the first rebar diameter identified using the boundary position. The reinforcing bar inspection device according to claim 1, characterized in that the reinforcing bar diameter determination unit determines, from the first reinforcing bar diameter and the second reinforcing bar diameter, a reinforcing bar diameter that satisfies a predetermined determination condition as the reinforcing bar diameter of the measurement result. The reinforcing bar inspection device according to claim 1, wherein the reinforcing bar diameter determination unit determines the reinforcing bar diameter obtained by statistically processing the first reinforcing bar diameter and the second reinforcing bar diameter as the reinforcing bar diameter of the measurement result. A reinforcement inspection method using a reinforcement inspection device,
a step in which a three-dimensional information acquisition unit acquires three-dimensional information of an inspection area in which reinforcing bars are arranged;
a step in which a plane specifying unit specifies a reinforcement plane to be inspected from the inspection area using the three-dimensional information;
an image conversion unit converting the captured image of the inspection area into a normal image that faces the reinforcement plane of the inspection target;
A step in which a position detection unit detects reinforcing bar position information in the reinforcement bar arrangement plane using the orthogonal image;
An image extraction unit extracts a reinforcing bar image from the orthogonal image based on the reinforcing bar position information in the reinforcing bar arrangement plane of the inspection target;
a feature calculation unit acquires a scanning line consisting of pixel values of a plurality of pixels aligned along the longitudinal direction of the rebar in the orthogonal image for each pixel in a direction perpendicular to the longitudinal direction of the rebar, calculates a feature by frequency-converting the pixel values of the scanning line, identifies a boundary position between the rebar and the background based on the result of comparing the calculated feature with a reference feature, and identifies a first rebar diameter using the boundary position;
an inference unit inferring a second reinforcing bar diameter using a learning model that outputs a reinforcing bar diameter when the reinforcing bar image is input;
A step in which a reinforcing bar diameter determination unit determines the reinforcing bar diameter of the measurement result using the first reinforcing bar diameter and the second reinforcing bar diameter;
a step in which a measurement result information generating unit generates and outputs measurement result information for displaying the reinforcing bar diameter of the measurement result. A program for causing a computer to function as the reinforcing bar inspection device described in claim 1.