JP7398613B2 - Joint surface roughness evaluation device and joint surface roughness evaluation system - Google Patents

Description

Article 30, Paragraph 2 of the Patent Act applies Publication name: Concrete Engineering Annual Papers Volume 41 (2019) Publication date: June 15, 2020 Publisher: Japan Concrete Institute Applicable page: No. Pages 1295-1300 Publication title: Effects of the shape of concrete pouring surfaces on the mechanical properties of structures

The present invention relates to a joint surface roughness evaluation device and a joint surface roughness evaluation system for evaluating the roughness of a joint surface formed on the surface of previously poured concrete.

BACKGROUND ART Conventionally, when constructing a concrete structure, a method is known in which a concrete pouring area is divided into a plurality of blocks or a plurality of layers, and concrete is successively poured. In this case, in order to prevent the joint from becoming a weak point against shearing force, the surface of the concrete that has been previously cast and hardened is treated to create an uneven joint surface. , new concrete is being poured.

However, no index has been defined for evaluating the roughness of the joint surface, and at present only skilled engineers can visually observe the unevenness of the joint surface. For this reason, the quality of the poured surface not only varies due to individual differences among skilled engineers, but also cannot be evaluated if a skilled engineer is not permanently stationed at the construction site. This may cause construction delays, such as not being able to start pouring work.

Under these circumstances, for example, Patent Document 1 discloses a method for evaluating concrete pour joint surfaces. Specifically, a concrete joint surface is photographed using two digital cameras, and the amount of unevenness of the joint surface from a certain reference height is calculated based on the two photographed images. Then, the surface shape of the joint surface is converted into a spectrum from the calculated amount of unevenness, and only the waveform component that contributes to the shear strength is extracted from this spectrum to evaluate the shear strength.

Japanese Patent Application Publication No. 2008-145174

Although the method described in Patent Document 1 is capable of accurately evaluating the shear strength of a concrete joint surface, it is complicated to perform the above calculation based on the amount of unevenness of the joint surface each time an evaluation is performed. This requires a great deal of effort. In addition, although shear strength is influenced by the surface roughness of the joint surface, it is not a direct quantification of surface roughness, so it can be used as an evaluation instead of observing the joint surface by a skilled engineer. However, there are many problems in using the evaluation results of shear strength.

The present invention was made in view of the above problem, and its main purpose is to improve the roughness of the joint surface formed on the surface of previously poured concrete, without being influenced by individual differences among skilled engineers. It is an object of the present invention to provide a concrete joint surface roughness evaluation device and a concrete joint surface roughness evaluation system capable of evaluating concrete joint surface roughness.

In order to achieve such an object, the pour joint surface roughness evaluation device of the present invention is a pour joint surface roughness evaluation device that evaluates the roughness of a pour joint surface formed on the surface of previously poured concrete, and comprises: an evaluation target image acquisition unit that acquires an evaluation target image in which the joint surface is captured; and quantification of the roughness of the joint surface captured in the evaluation target image using a roughness index representing the degree of surface roughness. a roughness determination unit that performs statistical processing to determine the roughness index using difference information between height distribution information of the joint surface and smoothed information obtained by smoothing the height distribution information. The height distribution information is calculated by vertically dividing a group of height measurement values, which is a collection of point cloud data obtained by measuring the height distribution of the soldering surface, at equal intervals. It is characterized in that it is converted into a set of height information that is continuous in the direction and the lateral direction .

Further, in the joint surface roughness evaluation device of the present invention, when the roughness determination section is inputted with the evaluation target image, the roughness determination section is configured to compare a learning image in which the joint surface is imaged and a learning image captured in the learning image. The roughness index of the joint surface captured in the evaluation target image is automatically determined using a machine-learned index estimation model using a known roughness index of the joint surface captured in the evaluation target image as training data. , is characterized by comprising an automatic determination section.

Furthermore, the joint surface roughness evaluation device of the present invention is characterized in that the roughness index is a standard deviation of the difference information. Further, the difference information is obtained by using a two-dimensional image created by using continuous height information in the vertical and horizontal directions of the height distribution information as a pixel value of each pixel, and a moving average filter from the two-dimensional image. It is characterized in that it is a difference from a smoothed image in which the acquired smoothed information is used as the pixel value of each pixel.

According to the joint surface roughness evaluation device of the present invention, since the roughness index is an index that directly quantifies the surface roughness of the joint surface from height distribution information, It becomes possible to use the roughness index as an evaluation index instead of observation.

In addition, by evaluating the joint surface based on this roughness index, it is possible to minimize the variation in evaluation results without being affected by individual differences among specific skilled engineers, and improve the reliability of the evaluation. It becomes possible to increase the

Further, by simply acquiring a captured image of the joint surface, it is possible to automatically determine the roughness index of the joint surface captured in the evaluation target image using an index estimation model created by machine learning. As a result, even in the absence of a skilled engineer, the roughness index obtained through automatic judgment can be used to determine the suitability of the pouring surface before pouring the following concrete, and further improve the treatment of the pouring surface. It becomes possible to judge whether or not there is a need to implement it.

The joint surface roughness evaluation system of the present invention is characterized by comprising the joint surface roughness evaluation device of the present invention and an imaging camera that captures the evaluation target image.

According to the joint surface roughness evaluation system of the present invention, the roughness of the joint surface formed on the surface of previously poured concrete can be evaluated economically and with good workability using a simple device without the need for special equipment. It becomes possible to evaluate.

According to the present invention, since the roughness of the joint surface is quantified by the roughness index, the roughness of the joint surface can be calculated based on the roughness index without being influenced by the judgment of a specific skilled engineer. This makes it possible to minimize the variation in evaluation results and improve the reliability of the evaluation.

FIG. 1 is a diagram showing a joint surface roughness evaluation system in an embodiment of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a figure which shows the structure of the joint surface roughness evaluation apparatus in embodiment of this invention. It is a figure which shows the procedure of quantifying the joint surface by a roughness index in embodiment of this invention. FIG. 3 is a diagram (part 1) showing three-dimensional shape data created from an evaluation target image and height distribution information in an embodiment of the present invention. FIG. 7 is a diagram showing three-dimensional shape data created from the evaluation target image and height distribution information in the embodiment of the present invention (Part 2). FIG. 3 is a diagram illustrating how a two-dimensional image is created from height distribution information and differential information is extracted in an embodiment of the present invention. It is a graph which shows the height distribution information, smoothing information, and difference information of the pouring surface imaged in the evaluation target image in embodiment of this invention. FIG. 3 is a diagram (part 1) showing the distribution of roughness index of the pouring surface captured in the evaluation target image in the embodiment of the present invention. FIG. 7 is a diagram showing the distribution of roughness index of the learning image in the embodiment of the present invention (Part 2). It is a figure showing the procedure for creating an index estimation model in an embodiment of the present invention. It is a figure showing the concept of teacher data L in an embodiment of the present invention.

The present invention is to quantify and evaluate the roughness of a joint surface formed on the surface of previously poured concrete using a roughness index. The details will be explained below with reference to FIGS. 1 to 11.

As shown in FIG. 1, the preceding concrete 1 that is placed in advance of the construction target area is hardened and then the following concrete 2 is placed, but before that, a pouring surface treatment is performed. A concave and convex joint surface 3 with an appropriately adjusted surface roughness is formed.

In addition to so-called laitance treatment, which removes laitance and loose aggregates, joint surface treatments generally include scraping the concrete surface with a wire brush, chipping, and spraying with high-pressure water.

≪Joint surface roughness evaluation system≫
The joint surface roughness evaluation system S quantifies and evaluates the roughness of the joint surface 3 formed in this way using a roughness index R indicating the degree of surface roughness, and uses an imaging method to image the joint surface 3. It includes a camera C, a measuring device H that measures the height distribution of the joint surface 3, and a joint surface roughness evaluation device 100.

The imaging camera C may be any camera, such as a still image camera or a moving image camera, as long as it is capable of recording captured images as digital data. In this embodiment, an imaging camera C images the joint surface 3 to obtain an evaluation target image 4 and a learning image 5, which will be described later.

The measuring device H may be any measuring device that can measure the height distribution of the joint surface 3, and may be a three-dimensional measuring device such as a three-dimensional digital camera or a laser scanner, or a caliper. In this embodiment, a 3D laser scanner is employed as the measuring device H, and point cloud data is acquired as a height measurement value group V obtained by measuring the height distribution of the joint surface 3. Note that the height measurement value group V is a set of data obtained by measuring the height distribution of the joint surface 3 continuously or intermittently.

≪Joint surface roughness evaluation device≫
As shown in FIG. 2, the joint surface roughness evaluation device 100 includes an input section 110, an output section 120, a storage section 130, and an arithmetic processing section 140 such as a CPU, GPU, ROM, RAM, and a hardware interface. It is composed of a computer system.

Note that the input unit 110 supplies information input from input devices such as a keyboard and a mouse in addition to the imaging camera C and the measuring device H to the joint surface roughness evaluation device 100, and the output unit 120 It outputs information supplied from the input section 110, information stored in the storage section 130, etc. to an output device such as a display or a printer.

Then, when the CPU of the arithmetic processing unit 140 executes a predetermined program, the automatic determination unit 150 and the roughness index calculation unit 170 included in the evaluation target image acquisition unit 141, determination method selection unit 142, and roughness index determination unit 143 , and the functions of the index estimation model creation unit 160 are realized.

These functions, together with the procedure for quantifying and evaluating the roughness of the joint surface 3, will be explained below according to the flowchart of FIG. 3.

In addition, when quantifying the roughness of the joint surface 3 using the roughness index R, it can also be calculated based on the measured value of the height distribution of the joint surface 3 (hereinafter referred to as "measurement calculation"). However, it is also possible to perform automatic determination from a captured image of the joint surface 3 to be evaluated without performing measurement.

<Selection of method for determining roughness image R>
First, a user inputs an evaluation target image 4 obtained by capturing an image of the joint surface 3 into the joint surface roughness evaluation apparatus 100 via the input unit 110 . In addition, selection information is input that instructs which method, automatic determination or measurement calculation, should be used to quantify the roughness index R of the joint surface 3 captured in the evaluation target image 4.

Then, in the joint surface roughness evaluation device 100, the evaluation target image acquisition unit 141 acquires the evaluation target image 4 stored in the storage unit 130 via the input unit 110, and supplies it to the determination method selection unit 142 (Step 1 ).

The determination method selection unit 142 acquires the selection information input by the user into the input unit 110, and selects the evaluation target image 4 supplied from the evaluation target image acquisition unit 141 based on the acquired selection information from the roughness index determination unit 143. (Step 2).

<When measuring and calculating the roughness index R>
When the selection information is measurement calculation of the roughness index R, the joint surface roughness evaluation device 100 uses the roughness index calculation unit 170 to calculate the roughness of the joint surface 3 captured in the evaluation target image 4. Quantify the index R (Step 3).

≪Roughness index calculation section≫
As shown in FIG. 2, the roughness index calculation section 170 includes a height information acquisition section 171, a difference information extraction section 172, and an index acquisition section 173. These functions will be explained below together with the calculation procedure of the roughness index R with reference to FIGS. 4 to 9.

First, in the joint surface roughness evaluation device 100, the height information acquisition unit 171 acquires a group of height measurement values V of the joint surface 3 captured in the evaluation target image 4. The height measurement value group V is input by the user to the joint surface roughness evaluation apparatus 100 together with the evaluation target image 4 through the input unit 110.

Note that the height measurement value group V may be one measured by the measurement device H of the joint surface roughness evaluation system S, or may be measured by another device or measurement information measured in the past. good. Further, when a caliper is used as the measuring device H, it is preferable to create a height measurement value group V by combining a set of measured values by the caliper and the position information of each measurement value.

The height information acquisition unit 171 that has acquired the height measurement value group V of the joint surface 3 converts the height measurement value group V into height distribution information U, which is a set of height information continuous at desired equal intervals. This is then supplied to the difference information extraction section 172 and the storage section 130.

In this embodiment, since point cloud data is acquired as the height measurement value group V, sub-sampling is performed at 2 mm intervals in both the vertical and horizontal directions, and the height information is displayed at consecutive heights at equal intervals of 2 mm. The distribution information U was created. As an example, three-dimensional shape data created from height distribution information U is shown in FIGS. 4(a) and 5(b) and 5(a) and (b), respectively, for the joint surface 3 captured in the evaluation target image 4. show.

Figure 4 (a) is an image in which a skilled engineer observed that the joint surface treatment was not appropriate (undercut), and Figure 4 (b) is an image that was observed by a skilled engineer as having improper joint surface treatment (overcut). (cut) and the observed image are shown. In addition, Figure 5(a) is an image in which a skilled engineer observed that the joint surface treatment was appropriate (shallower), and Figure 5(b) is an image in which a skilled technician observed that the joint surface treatment was appropriate (deeper). ) is shown.

Next, in the joint surface roughness evaluation device 100, the difference information extraction section 172 converts the height distribution information U supplied from the height information acquisition section 171 into two-dimensional image information. Thereafter, image processing is performed on this two-dimensional image information to extract smoothed information obtained by smoothing the height distribution information U. Note that the smoothing information is information that can be regarded as unevenness that occurs on the concrete surface after the preceding concrete 1 is placed.

Further, the difference information extraction unit 172 extracts difference information between the height distribution information U and the smoothing information, and extraction of the difference information is generally widely known as image processing that emphasizes blurred contours of the image. This is done in a manner similar to unsharp processing.

Specifically, first, as shown in FIG. 6A, a two-dimensional image is created in which the height distribution information U is the pixel value of each pixel, and this is used as the original image 6. The original image 6 in FIG. 6(a) is created using the height distribution information U corresponding to the evaluation target image 4 (overcut) shown in FIG. 4(b).

Next, as shown in FIG. 6(b), a smoothed image 7 is obtained from the original image 6 using a moving average filter, with smoothing information being the pixel value of each pixel. In this embodiment, the pixel values of each pixel included in the range of 25 to 50 mm are averaged to determine the pixel value of the pixel located at the center.

Thereafter, as shown in FIG. 6(c), by taking the difference between the original image 6 and the smoothed image 7, a difference information image 8 in which the difference information is the pixel value of each pixel is obtained. The difference information extraction unit 172 supplies the smoothed information and difference information acquired through the above processing to the index acquisition unit 173 and the storage unit 130.

In FIG. 7, for each of the four types of evaluation target images 4 shown in FIGS. 4(a)(b) and 5(a)(b), for an arbitrary row, the height distribution information U of the original image 6 is shown. A graph of pixel values, pixel values of a smoothed image 7 that is smoothed information, and pixel values of a difference information image 8 that is difference information is shown.

Finally, in the joint surface roughness evaluation device 100, the index acquisition unit 173 calculates the roughness index R from the extracted difference information. The index acquisition unit 173 performs statistical processing using the difference information supplied from the difference information extraction unit 172 to calculate the roughness index R. However, in this embodiment, the standard deviation of the difference information, that is, the scattering of the difference information The degree of surface roughness is calculated as a roughness index R representing the degree of surface roughness.

The joint surface roughness evaluation device 100 associates the roughness index R calculated in the above procedure with the evaluation target image 4 and supplies it to the storage unit 130, and also outputs it to the output device via the output unit 120. (Step 5).

FIGS. 8A and 8B and 9A and 9B show the roughness index R calculated by the above procedure. In this embodiment, the original image 6 is divided into 10 equal parts in the vertical and horizontal directions to form a mesh, and then the roughness index R is calculated for each mesh. Therefore, FIGS. 8 and 9 show the roughness index R calculated for each mesh.

Note that FIG. 8(a) shows the roughness index R of the joint surface 3 (in the case of undercut) shown in FIG. 4(a), and FIG. 8(b) shows the roughness index R of the joint surface 3 shown in FIG. 4(b). This is the roughness index R of surface 3 (in the case of overcut). FIG. 9(a) shows the roughness index R of the joint surface 3 shown in FIG. 5(a) (when the joint surface treatment is appropriate (shallow)), and FIG. 9(b) shows the roughness index R of the joint surface 3 shown in FIG. ) is the roughness index R of the joint surface 3 (when the joint surface treatment is appropriate (deep)).

Looking at Figures 8(a) and (b), it can be seen that the roughness index R of the joint surface 3, which was observed as an undercut by a skilled engineer, was mostly 0.4 or less, whereas the roughness index R of the joint surface 3, which was observed as an overcut, was mostly 0.4 or less. The roughness index R of the surface 3 shows a value exceeding 1.0 in most cases, and it appears that the surface of the overcut joint surface 3 has a large variation in the uneven state.

On the other hand, looking at FIGS. 9(a) and 9(b), the roughness index R of the joint surface 3, which is observed by skilled engineers as being in an appropriate state (shallow), is often around 0.5; The roughness index R of the joint surface 3 observed as (deepening) is often about 0.5 to 1.0. As a result, the roughness index R of the joint surface 3 observed by a skilled engineer as appropriate (shallow) (deep) is both the roughness index R of the joint surface 3 observed as an undercut, and the roughness index R of the joint surface 3 observed as an overcut. It can be seen that the roughness index R is within the range of the roughness index R of the joint surface 3, and it can be seen that there is a harmony between the observations made by the skilled engineer and the roughness index R.

As mentioned above, the roughness index R to be calculated is an index that directly quantifies the uneven state of the joint surface 3, that is, the roughness, from the height distribution information U, so the roughness index R is calculated by a skilled engineer. It is possible to use this as an evaluation index instead of observing the joint surface 3. Further, since the roughness index R is not influenced by individual differences among specific skilled engineers, it is possible to minimize variations in evaluation results and improve reliability of evaluation.

<When automatically determining roughness image R>
Next, returning to the flowchart of FIG. 3, the case where the user wishes to obtain the roughness index R of the joint surface 3 captured in the evaluation target image 4 by automatic determination will be described below as shown in FIG. , will be explained with reference to FIGS. 10 and 11.

≪Automatic judgment section≫
When the selection information input by the user into the input unit 110 indicates automatic determination, the automatic determination unit 150 determines the roughness index of the joint surface 3 captured in the evaluation target image 4. Obtain R (Step 4).

In order to automatically determine the roughness index R of the joint surface 3, it is necessary to prepare in advance an index estimation model M for determining the roughness index R from image information of the joint surface 3. Therefore, the index estimation model M used by the automatic determination section 150 will be explained below.

<Creation of index estimation model for automatically determining roughness index R>
The index estimation model M is, for example, a learning result obtained by learning applying a neural network, is created by the index estimation model creation unit 160 of the joint surface roughness evaluation device 100, and is stored in the storage unit 130. .

≪Index estimation model creation department≫
As shown in FIG. 2, the index estimation model creation section 160 includes a learning image acquisition section 161, a correct label acquisition section 162, a teacher data creation section 163, and a machine learning section 164. These functions, together with the procedure for creating the index estimation model M, will be explained below according to the flowchart of FIG.

As shown in FIG. 11, in the joint surface roughness evaluation device 100, the learning image acquisition unit 161 uses, for example, evaluated past images as the learning images 5 stored in the storage unit 130 via the input unit 110. The evaluation target image 4 is acquired and supplied to a teacher data creation unit 163, which will be described later. Note that identification information such as an ID is given to the learning image 5 (Step 411).

If the roughness index R of the shot joint surface 3 being imaged is known, the learning image 5 is an evaluated past evaluation target imaged by the imaging camera C of the joint surface roughness evaluation system S. The image 4 may be the image 4 or may be a captured image captured by another method and stored in the storage unit 130 via the input unit 110.

Next, in the joint surface roughness evaluation device 100, the correct label acquisition unit 162 applies the correct label D stored in the storage unit 130 via the input unit 110 to the joint surface 3 captured in the learning image 5. The corresponding roughness index R is acquired and supplied to the teacher data creation unit 163, which will be described later (Step 412).

The correct label D (roughness index R) is given the identification information of the corresponding learning image 5, and the correct label acquisition unit 162 receives the information supplied from the learning image acquisition unit 161 based on this identification information. The correct label D corresponding to the learning image 5 (for example, the past evaluation target image 4) is acquired from the storage unit 130.

Thereafter, in the joint surface roughness evaluation device 100, the teacher data creation unit 163 uses the learning image 5 (for example, the past evaluation target image 4) supplied from the learning image acquisition unit 161, and the correct label acquisition unit 162. Creates teacher data L that is linked with the correct label D (roughness index R) corresponding to the joint surface 3 of the learning image 5 supplied from , and supplies it to the machine learning unit 164 and the storage unit 130 Step 413).

The machine learning unit 164 performs machine learning on the teaching data L supplied from the teaching data creation unit 163 using a known machine learning method (algorithm), creates an index estimation model M as a result, and stores it. 130 (Step 414).

<Automatic determination of roughness image R>
In the joint surface roughness evaluation device 100, the automatic determination unit 150 acquires the index estimation model M created by the index estimation model creation unit 160 and stored in the storage unit 130, and then obtains the index estimation model M that is supplied from the determination method selection unit 142. A value target image 4 is input into this index estimation model M.

Then, the roughness index R, which is estimated to correspond to the roughness of the joint surface 3 captured in the evaluation target image 4, is automatically determined. The roughness index R is associated with the evaluation target image 4, and is supplied to the storage unit 130 and output to the output device via the output unit 120 (Step 5).

As mentioned above, a plurality of evaluation target images 4 taken of the joint surface 3 with known roughness index R are accumulated, and the evaluation target image 4 is used as the learning image 5, and the corresponding roughness index R is calculated. The correct answer label is set as D, and an index estimation model M is created in the index estimation model creation section 160. Thereby, the roughness index R of the joint surface 3 captured in the evaluation target image 4 can be automatically determined by simply acquiring the evaluation target image 4 that captures the joint surface 3 .

Therefore, even if a skilled engineer is absent, the suitability of the joint surface 3 and the joint surface treatment before pouring the trailing concrete 2 can be determined based on the roughness index R obtained by automatic judgment. It will be possible to judge whether there is a need to further implement the above.

In addition, the joint surface roughness evaluation system S does not require special equipment and is a simple device that is economical and has good workability, and can evaluate the roughness of the joint surface 3 formed on the surface of the preceding concrete 1. becomes possible.

The joint surface roughness evaluation system S of the present invention is not limited to the above-described embodiments, and various changes can be made without departing from the spirit of the present invention.

For example, in the present embodiment, when acquiring smoothing information in the roughness index calculation unit 170, a moving average filter is used to acquire a smoothed image 7 in which the smoothing information is used as the pixel value of each pixel from the original image 6. However, other filters may be used as long as they are processing methods that can obtain smoothed information.

In addition, when evaluating the roughness of the joint surface 3 based on the roughness index R, depending on the unevenness of the joint surface 3, undercut (shallow), appropriate (shallow), appropriate (deep), overcut (deep), but any classification item or quantity may be used.

1 Leading concrete 2 Trailing concrete 3 Pour joint surface 4 Evaluation target image 5 Learning image 6 Original image 7 Smoothed image (smoothed information)
8 Difference information image (difference information)
100 Joint surface roughness evaluation device 110 Input section 120 Output section 130 Storage section 140 Arithmetic processing section 141 Evaluation target image acquisition section 142 Judgment method selection section 143 Roughness index judgment section 150 Automatic judgment section 160 Index estimation model creation section 161 Learning Image acquisition unit 162 Correct label acquisition unit 163 Teacher data creation unit 164 Machine learning unit 170 Roughness index calculation unit 171 Height information acquisition unit 172 Difference information extraction unit 173 Index acquisition unit

S Joint surface roughness evaluation target system C Imaging camera H Measuring device R Roughness index M Index estimation model V Height measurement value group U Height distribution information D Correct answer label L Teacher data

Claims (5)

A joint surface roughness evaluation device for evaluating the roughness of a joint surface formed on the surface of previously poured concrete,
an evaluation target image acquisition unit that acquires an evaluation target image in which the soldering surface is imaged;
a roughness determination unit that quantifies the roughness of the joint surface captured in the evaluation target image using a roughness index representing the degree of surface roughness;
The roughness index is calculated by performing statistical processing using difference information between height distribution information of the joint surface and smoothed information obtained by smoothing the height distribution information,
The height distribution information includes height measurement value groups that are a set of point cloud data obtained by measuring the height distribution of the soldering surface, and height information that is continuous in the vertical and horizontal directions at equal intervals. A joint surface roughness evaluation device characterized in that the roughness is converted into a set. The joint surface roughness evaluation device according to claim 1,
When the roughness determination unit receives the evaluation target image,
Using a machine-learned index estimation model using a learning image of a joint surface and a known roughness index of the joint surface captured in the learning image as training data,
A joint surface roughness evaluation device comprising an automatic determination unit that automatically determines the roughness index of the joint surface captured in the evaluation target image. The joint surface roughness evaluation device according to claim 1,
The difference information is
A two-dimensional image created using continuous height information in the vertical and horizontal directions of the height distribution information as a pixel value of each pixel, and the smoothed information obtained from the two-dimensional image using a moving average filter. A joint surface roughness evaluation system characterized in that the pixel value of each pixel is the difference between a smoothed image and a pixel value. In the joint surface roughness evaluation device according to any one of claims 1 to 3,
A joint surface roughness evaluation device characterized in that the roughness index is a standard deviation of the difference information. A joint surface roughness evaluation device according to any one of claims 1 to 4,
an imaging camera that captures the evaluation target image;
A joint surface roughness evaluation system characterized by comprising:

Images