Information processing device, method for setting a determination area, and program for setting a determination area

Abstract

To allow for automatically setting an appropriate region in an image of an inspection target.SOLUTION: An information processing device (1) is provided, comprising a detection unit (101) configured to detect feature points with known position relationships with a determination region to be inspected from an image (111) of an inspection target object, and a setting unit (102) for setting the determination unit based on the feature points detected by the detection unit (101).SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an information processing apparatus that sets a determination area in an image, etc.

Background Art

[0002] Conventionally, various determinations using images have been performed. For example, Patent Document 1 below discloses an ultrasonic flaw detection method using the phased array TOFD (Time Of Flight Diffraction) method. In this ultrasonic flaw detection method, an ultrasonic beam is transmitted from a phased array flaw detection element and focused on a stainless steel welded part, and a flaw detection image generated based on the diffracted wave is displayed. Thereby, it becomes possible to detect welding defects generated inside the stainless steel welded part.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the technique of Patent Document 1, since a flaw detection image is visually confirmed to detect welding defects, there is a problem that the human and time costs required for inspection are high. As a means for solving such a problem, for example, it is conceivable to automatically determine the presence or absence of welding defects by analyzing the flaw detection image with a computer.

[0005] However, flaw detection images contain various echoes and noise in addition to echoes of welding defects. Therefore, in order to improve the accuracy of the judgment, it is desirable to first extract the areas where welding defects are likely to occur from the image area of ​​the flaw detection image as the judgment area. Furthermore, it is desirable that this extraction of the judgment area be performed automatically without human intervention. This is a common challenge not only for inspecting for the presence or absence of defects using flaw detection images, but also for any inspection using any image.

[0006] One aspect of the present invention aims to realize an information processing device, etc., that can automatically set a reasonable judgment area in an image of an object to be inspected. [Means for solving the problem]

[0007] To solve the above problems, an information processing device according to one aspect of the present invention includes a detection unit that detects feature points from an image of an object to be inspected whose positional relationship with a determination area to be inspected is known, and a setting unit that sets the determination area based on the feature points detected by the detection unit.

[0008] To solve the above problems, a method for setting a determination region according to one aspect of the present invention is a method for setting a determination region executed by one or more information processing devices, and includes a detection step of detecting a feature point from an image of an object to be inspected whose positional relationship with the determination region to be inspected is known, and a setting step of setting the determination region based on the feature point detected in the detection step. [Effects of the Invention]

[0009] According to one aspect of the present invention, it becomes possible to automatically set a reasonable judgment area in an image of an object to be inspected. [Brief explanation of the drawing]

[0010] [Figure 1] This is a block diagram showing an example of the main components of an information processing device according to Embodiment 1 of the present invention. [Figure 2]This figure shows an overview of the inspection system, including the information processing device described above. [Figure 3] This is a diagram showing a cross-section of the pipe end weld. [Figure 4] This figure shows an example of setting the judgment area. [Figure 5] This diagram shows an inspection of a pipe where the weld is located near the end. [Figure 6] This figure shows a flaw detection image generated by an ultrasonic flaw detection device equipped with a propagation-inhibiting material. [Figure 7] This flowchart shows an example of the process performed by the above-mentioned information processing device. [Figure 8] This is a block diagram showing an example of the main components of an information processing device according to Embodiment 2 of the present invention. [Figure 9] This flowchart shows an example of the process performed by the above-mentioned information processing device. [Figure 10] This figure shows an example of a CR image of a pipe joint weld, which is an example of an object to be inspected. [Figure 11] This figure shows an example of an object to be inspected, namely a piston crown and its surrounding structure, as well as an image of the piston crown. [Modes for carrying out the invention]

[0011] [Embodiment 1] (Device configuration) The configuration of an information processing device 1 according to one embodiment of the present invention will be described with reference to Figure 1. Figure 1 is a block diagram showing an example of the main components of the information processing device 1. As shown in the figure, the information processing device 1 includes a control unit 10 that controls all parts of the information processing device 1, and a storage unit 11 that stores various data used by the information processing device 1. The information processing device 1 also includes a communication unit 12 for the information processing device 1 to communicate with other devices, an input unit 13 that receives input of various data to the information processing device 1, and an output unit 14 for the information processing device 1 to output various data.

[0012] The control unit 10 also includes a detection unit 101, a setting unit 102, and a determination unit 103. The storage unit 11 stores an image 111, a detection model 112, a determination model 113, and a determination result 114. The determination result 114 is stored after being determined by the determination unit 103.

[0013] The image 111 is an image of the inspection object. The image 111 may be according to the purpose and method of inspection, etc. For example, an image obtained by photographing the inspection object with a general optical camera may be used as the image 111, or a flaw detection image used in ultrasonic inspection as described later, or a CR (Computed Radiography) image used in a radiation transmission test (RT) may be used as the image 111.

[0014] The image area of the image 111 can be divided into an area where at least the inspection target part of the inspection object is imaged and its background area. Also, the image 111 has feature points whose positional relationship with the determination area is known. Although details will be described later, the information processing device 1 sets a determination area including the area where the inspection target part is imaged based on the above feature points.

[0015] The detection model 112 is a detection model obtained by machine learning the appearance of feature points whose positional relationship with the determination area is known. By machine learning using an image in which the determination area and the feature points are imaged as teacher data, a detection model 112 for detecting the feature points imaged in the image 111 can be constructed.

[0016] Any machine learning algorithm can be applied. For example, the detection model 112 may be a convolutional neural network model with high object detection accuracy from images. Also, for example, the detection model 112 may be a model that performs both object detection and classification, such as YOLO (You Only Look Once).

[0017] The determination model 113 is a learned model for determining whether an inspection object is good or bad. By machine learning using images of good inspection objects and images of defective inspection objects as teacher data, a determination model 113 can be constructed that outputs an output value indicating whether the inspection object shown in the input image is good or defective. Although details will be described later, a partial image obtained by cutting out the portion of the determination region from the image 111 is input to the determination model 113.

[0018] The determination result 114 is data indicating the determination result of the information processing apparatus 1 regarding the inspection object. For example, when the information processing apparatus 1 determines whether an inspection object is good or bad, the determination result 114 in which the identification information of the inspection object and the result of the good / bad determination are associated is stored in the storage unit 11.

[0019] The detection unit 101 detects feature points whose positional relationship with the determination region is known from the image 111 of the inspection object. More specifically, the detection unit 101 detects feature points from the image 111 using the detection model 112. Then, the setting unit 102 sets the determination region based on the feature points detected by the detection unit 101. Details of the detection of the feature points and the setting of the determination region will be described later.

[0020] The determination unit 103 determines whether the inspection object is good or bad for the determination region set by the setting unit 102. Specifically, the determination unit 103 cuts out the portion of the determination region set by the setting unit 102 in the image 111 to generate a partial image, and inputs the generated partial image to the determination model 113. Then, the determination unit 103 determines whether the inspection object is good or bad based on the output value of the determination model 113.

[0021] Note that the determination unit 103 only needs to perform determination on the determination region set by the setting unit 102, and the determination method is not limited to the above example. For example, the determination unit 103 may perform determination by analyzing the determination region. Also, the content of the determination is not limited to determining whether the inspection object is good or bad, and may be classification of the inspection object (determination of which classification the inspection object belongs to), etc.

[0022] [[ID=*19]] As described above, the information processing device 1 includes a detection unit 101 that detects feature points whose positional relationship with the determination area to be inspected is known from an image 111 of the object to be inspected, and a setting unit 102 that sets the determination area based on the feature points detected by the detection unit 101.

[0023] According to the above configuration, feature points are detected from the image 111 of the object to be inspected, and the judgment area is set based on the detected feature points. Since the positional relationship between these feature points and the judgment area to be inspected is known, it becomes possible to set a reasonable judgment area by using these feature points as a reference. Therefore, according to the above configuration, it becomes possible to automatically set a reasonable judgment area in the image of the object to be inspected.

[0024] Furthermore, the detection unit 102 detects feature points from the image 111 using a detection model 112 that has been trained on the appearance of feature points. This makes it possible to detect feature points with distinctive appearances with high accuracy.

[0025] (System Overview) An overview of an inspection system according to one embodiment of the present invention will be described with reference to Figure 2. Figure 2 is a diagram showing an overview of the inspection system 100. The inspection system 100 is a system that inspects whether or not there are defects in an object to be inspected from an image of that object, and includes an information processing device 1 and an ultrasonic flaw detection device 7.

[0026] The following describes an example of using the inspection system 100 to inspect for defects in the pipe end welds of a heat exchanger. A pipe end weld is the joint where multiple metal pipes constituting the heat exchanger are welded to a metal tube sheet that bundles these pipes together. A defect in a pipe end weld is a defect where a void forms inside the weld. The pipes and tube sheet may be made of non-ferrous metals such as aluminum, or of resin. Furthermore, the inspection system 100 can also be used to inspect for defects in the welds (base welds) between pipe supports and pipes in boiler equipment used in facilities such as waste incineration plants. Of course, the inspection area is not limited to welds, and the inspection target is not limited to heat exchangers.

[0027] During inspection, as shown in Figure 2, a probe coated with a coupling medium is inserted from the end of the pipe. Ultrasonic waves are propagated from the inner wall of the pipe towards the pipe end weld using this probe, and the resulting echoes are measured. If a defect has occurred that creates a void within the pipe end weld, the echoes from that void will be measured, and these can be used to detect the defect.

[0028] For example, in the magnified view of the area around the probe shown in the lower left of Figure 2, the ultrasonic wave indicated by arrow L3 propagates to a part of the pipe end weld that has no void. Therefore, the echo of the ultrasonic wave indicated by arrow L3 is not measured. On the other hand, the ultrasonic wave indicated by arrow L2 propagates toward a part of the pipe end weld that has a void, so the echo of the ultrasonic wave reflected by this void is measured.

[0029] Furthermore, since ultrasonic waves are reflected from the periphery of the pipe end weld, the echoes of ultrasonic waves propagating to the periphery are also measured. For example, the ultrasonic wave indicated by arrow L1 propagates towards the pipe end rather than the pipe end weld, so it does not hit the pipe end weld, but is reflected from the pipe surface on the pipe end side of the pipe end weld. Therefore, the echo from the pipe surface is measured by the ultrasonic wave indicated by arrow L1. Also, the ultrasonic wave indicated by arrow L4 is reflected from the pipe surface on the inner side of the pipe end weld, so that echo is measured.

[0030] Since pipe end welds exist around the entire 360 ​​degrees of the pipe, measurements are repeatedly taken while rotating the probe by a predetermined angle (e.g., 1 degree) at a time. The data showing the measurement results from the probe is then transmitted to the ultrasonic flaw detection device 7. For example, the probe may be an array probe consisting of multiple array elements. With an array probe, by arranging the array elements so that their orientation coincides with the pipe's extension direction, pipe end welds that are wide in the pipe's extension direction can be efficiently inspected. The above array probe may also be a matrix array probe in which multiple array elements are arranged both vertically and horizontally.

[0031] The ultrasonic flaw detection device 7 generates an ultrasonic image by using data indicating the measurement results from the probe to visualize the ultrasonic echoes propagated to the pipe and pipe end weld. Figure 2 shows flaw detection image 111a, which is an example of an ultrasonic image generated by the ultrasonic flaw detection device 7. Alternatively, the information processing device 1 may generate the flaw detection image 111a. In this case, the ultrasonic flaw detection device 7 transmits data indicating the measurement results from the probe to the information processing device 1.

[0032] In flaw detection image 111a, the measured echo intensity is represented as the pixel value of each pixel. Furthermore, the image area of ​​flaw detection image 111a can be divided into a pipe area ar1 corresponding to the pipe, a weld area ar2 corresponding to the pipe end weld, and peripheral echo areas ar3 and ar4 where echoes from around the pipe end weld appear.

[0033] As described above, ultrasonic waves propagated from the probe in the direction indicated by arrow L1 are reflected by the pipe surface on the pipe end side of the pipe end weld. These ultrasonic waves are also reflected by the inner surface of the pipe, and these reflections occur repeatedly. Therefore, repeated echoes a1 to a4 appear in the peripheral echo region ar3 along arrow L1 in flaw detection image 111a. Similarly, ultrasonic waves propagated from the probe in the direction indicated by arrow L4 are repeatedly reflected by the outer and inner surfaces of the pipe. Therefore, repeated echoes a6 to a9 appear in the peripheral echo region ar4 along arrow L4 in flaw detection image 111a. These echoes appearing in peripheral echo regions ar3 and ar4 are also called bottom echoes.

[0034] The ultrasonic waves propagated from the probe in the direction indicated by arrow L3 do not have any reflectors, so no echoes appear in the region along arrow L3 in the flaw detection image 111a. On the other hand, the ultrasonic waves propagated from the probe in the direction indicated by arrow L2 are reflected by the void in the pipe end weld, i.e., the defective area, and as a result, echo a5 appears in the region along arrow L2 in the flaw detection image 111a.

[0035] As will be explained in detail below, the information processing device 1 analyzes such flaw detection images 111a to inspect the quality of the pipe end welds. Specifically, pipe end welds with internal voids are judged as defective, and pipe end welds without voids are judged as good. Furthermore, the information processing device 1 may also automatically determine the type of defect in pipe end welds that it has determined to be defective. Known types of defects that can occur in pipe end welds include, for example, poor initial penetration, poor fusion between weld passes, and undercuts. In addition, the location of occurrence and the echo shape are generally determined for each type of defect. Therefore, the type of defect can be determined based on the flaw detection images 111a.

[0036] (Details of pipe end welds) Figure 3 shows a cross-section of a pipe end weld. When a pipe and a tube sheet are welded, the weld metal penetrates into both the tube sheet and the pipe. In Figure 3, the position of the outer surface of the pipe before welding is shown by a dashed line, and the position of the surface of the tube sheet before welding is shown by a dashed line. As shown by this dashed line, the tube sheet in Figure 3 has a notch extending to the part that contacts the pipe. This notch is flat. Of course, the shape of the notch is arbitrary; for example, the notch could be curved.

[0037] Figure 3 shows that the pipe end weld penetrates into the pipe by a width w1 and into the tube sheet by a width w3. w2 is the groove width. Therefore, the thickness W of the pipe end weld can be expressed as (w1 + w2 + w3). In Figure 3, the height of the weld (the distance from the pipe end side to the pipe end side of the pipe end weld) is indicated by H.

[0038] Here, the groove width w2 is predetermined in the design. Also, the penetration depth (w1 + w3) into the inside of the pipe and the inside of the tube sheet usually falls within a certain range (for example, a range of 1 to 2 mm). For this reason, the thickness W of the pipe end weld can be calculated based on the groove width and the general penetration depth without having to measure it. As will be described in detail later, the setting unit 102 can set the judgment area using the thickness W calculated in this way.

[0039] (Example of setting the judgment area) Figure 4 shows an example of setting the judgment area. Image 111A in Figure 4 is a flaw detection image of a pipe end weld, and image 111B is a flaw detection image of another pipe end weld. Images 111A and 111B are images of ultrasonic echoes propagated from the back surface, which is the surface opposite to the surface where the weld is located, onto the weld of the object being inspected.

[0040] Image 111A shows the first reflected echoes A1 and A2 and the second reflected echoes A3 and A4 from the pipe surface. These are all bottom echoes. For example, such an image can be generated using an ultrasonic flaw detection device 7 equipped with a linear array transducer (a transducer whose elements are arranged in a single row).

[0041] In image 111A, the straight line L5 passing through the first reflected echoes A1 and A2 corresponds to the position on the pipe surface. The distance D from the first reflected echo A2 to the second reflected echo A4 corresponds to the thickness of the pipe.

[0042] Furthermore, in Image 111A, curve L6 indicates the position corresponding to the surface of the weld metal, straight line L7 indicates the position corresponding to the tube sheet, and curve L8 indicates the position corresponding to the boundary between the penetration area and the tube sheet. In other words, the area enclosed by L5 to L8 corresponds to the tube end weld.

[0043] In Image 111A, the distance from the straight line L5 to the ends of curves L6 and L8 is longer than the distance D from the first reflected echo A2 to the second reflected echo A4. In other words, this pipe end weld is thicker than the pipe.

[0044] In this case, the detection unit 101 only needs to detect the first reflected echoes A1 and A2 as feature points of the image 111A. Then, the setting unit 102 should set a determination area having a width of W or more than the thickness of the pipe end weld, based on the first reflected echoes A1 and A2 detected by the detection unit 101.

[0045] More specifically, the detection unit 101 simply needs to input image 111A into a detection model 112 that has been trained to detect the welded end in the first reflected echoes A1 and A2. This yields the detection results shown by rectangles A6 and A7 in Figure 4.

[0046] Furthermore, when constructing the detection model 112, only the edges of the reflected echoes need to be trained, thus keeping the training time and cost low. Also, since the edges of the reflected echoes have a similar shape regardless of the model of the ultrasonic flaw detector 7 or the object being inspected, the detection model 112 is highly versatile.

[0047] The setting unit 102 then sets a rectangular region A8 with a width W, where one side is the line segment connecting point A61 at the upper left end of rectangle A6 and point A71 at the upper right end of rectangle A7, as the determination region. The value of W can be, for example, the groove width plus the penetration width on the pipe end side and the pipe side, as explained with reference to Figure 3. Alternatively, the setting unit 102 may set a determination region with a width obtained by adding or multiplying a predetermined margin to the W calculated in this way.

[0048] The detection model 112 for detecting the first reflected echo can be constructed using machine learning with training data that associates ground truth data indicating the position of the first reflected echo in an image containing the first reflected echo. For example, if image 111A is used as training data, then information indicating the position and range of rectangle A6 and information indicating the position and range of rectangle A7 can be associated with image 111A as ground truth data. For example, the width, height, and representative coordinates of the rectangle can be used as information indicating the position and range of the rectangle.

[0049] On the other hand, Image 111B shows the first reflected echoes B1 and B2 from the pipe surface, and the second reflected echoes B3 and B4. These are all bottom echoes. In Image 111B, the area enclosed by L5 to L8 corresponds to the pipe end weld. In Image 111B, the distance from the straight line L5 to the ends of the curves L6 and L8 is shorter than the distance D from the first reflected echo B2 to the second reflected echo B4. In other words, this pipe end weld is thinner than the pipe.

[0050] In this case, the detection unit 101 only needs to detect the first reflection echoes B1 and B2 and the second reflection echoes B3 and B4 as feature points of image 111B. Then, the setting unit 102 only needs to set the region between the first reflection echoes B1 and B2 and the second reflection echoes B3 and B4 detected by the detection unit 101 as the determination region.

[0051] More specifically, the detection unit 101 simply needs to input image 111B into a detection model 112 that has been trained to detect the weld-side edges in the first reflection echoes B1 and B2 and the second reflection echoes B3 and B4. This yields the detection results shown by rectangles B6 to B9 in Figure 4. Note that the weld-side edges in the four reflection echoes (B1 to B4) may each be detected by a different detection model.

[0052] The setting unit 102 then sets a rectangular region B10, whose four vertices are point B61 at the upper left corner of rectangle B6, point B71 at the upper right corner of rectangle B7, point B91 at the lower right corner of rectangle B9, and point B81 at the lower left corner of rectangle B8, as the determination region. The detection model 112 for performing such detection can be constructed by machine learning using training data that associates the image showing the first and second reflected echoes with ground truth data indicating the positions of the first and second reflected echoes in the image.

[0053] As described above, in inspection using flaw detection images obtained by imaging ultrasonic echoes propagated around the welded area of ​​an object under inspection, the detection unit 101 may detect the first reflected echo among the echoes reflected by ultrasonic waves around the welded area of ​​the object under inspection as a feature point. The setting unit 102 may then set a judgment area used for determining the quality of the welded area based on the first reflected echo detected by the detection unit 101.

[0054] According to the above configuration, a reasonable judgment area can be set for determining the quality of the weld. This is because the reflected echo appears at a predetermined position around the weld in the flaw detection image. In the above example, the first reflected echo is detected at two locations (one on the inner side of the pipe and one on the outer side of the pipe, flanking the weld), but if the height range of the weld is known, the judgment area can be set by detecting only one of the first echoes. In this case, the setting unit 102 should set a judgment area with a height greater than or equal to the height of the weld, based on the first reflected echo.

[0055] As described above, by changing the method of setting the judgment area according to the relationship between the pipe thickness (determined based on the interval between the first and second reflected echoes) and the thickness of the pipe end weld (determined based on the groove width, etc.), the entire pipe end weld can be included in the judgment area. This prevents situations where a part of the pipe end weld falls outside the judgment area and defects are overlooked.

[0056] This detection of feature points and judgment areas is not limited to the inspection of pipe end welds, but is also effective for any inspection using flaw detection images, which are images of ultrasonic echoes propagated around the welded area of ​​the object being inspected.

[0057] For example, a quality inspection method is known in which the quality of a joint weld is confirmed using a flaw detection image generated based on measurement results obtained by propagating ultrasonic waves while a probe is in contact with the surface of the joint weld of a flat plate. Another quality inspection method is known in which flaw detection images are generated based on measurement results obtained by propagating ultrasonic waves at an oblique angle from the base material surface at the same height as the surface of the weld in a joint weld. In the flaw detection images used in these inspections, the first reflected echo of the ultrasonic waves reflected around the weld appears at a predetermined position around the weld, so the judgment area can be set based on this reflected echo.

[0058] Furthermore, the method of setting the region between the first and second reflected echoes as the judgment region is applicable to flaw detection images other than those shown in images 111A and 111B. For example, in flaw detection images generated based on measurements taken by applying a probe to the surface of a weld in a flat plate joint weld, the weld appears within the region between the first and second reflected echoes. The same is true for flaw detection images generated based on measurements taken by propagating ultrasonic waves obliquely from the base material surface at the same height as the surface of the weld in a joint weld. Therefore, the method of setting the region between the first and second reflected echoes as the judgment region is effective even in these types of flaw detection images.

[0059] (Example of setting the judgment area: When no bottom surface echo appears on the end of the tube) As shown in the example in Figure 4, when bottom echoes appear on both the pipe end side and the pipe end side of the pipe end weld, the judgment area can be set based on these echoes. However, when the weld is located near the end of the pipe, the bottom echo on the pipe end side may not appear. Examples of setting the judgment area in such cases will be explained based on Figures 5 and 6.

[0060] Figure 5 shows an inspection of a pipe where the weld is located near the end. In the example shown at 1001 in Figure 5, the pipe end and the surface of the pipe sheet are on almost the same plane, and the pipe end weld is also on this plane. When inspecting such a pipe end weld, as described below, the echo is measured with the propagation inhibiting material 8 attached near the probe of the ultrasonic flaw detection device 7.

[0061] Figure 5, section 1002 schematically shows the inspection of a pipe with a weld located near its end. Note that the propagation inhibitor 8 is omitted in this figure. As shown in the figure, the pipe end weld is formed to cover from the groove of the pipe sheet to the end face of the pipe. Ultrasonic waves are transmitted from the probe toward this pipe end weld. In this case, as shown in the figure, the ultrasonic waves may be propagated perpendicular to the extension direction of the pipe, or they may be propagated at an inclined angle.

[0062] The propagation inhibitor 8 either eliminates the transmitted ultrasonic waves or delays their propagation. For example, rubber materials such as silicone rubber, nitrile rubber, chloroprene rubber, ethylene rubber, fluororubber, butyl rubber, and silobutyl rubber can be used as the propagation inhibitor 8. The propagation inhibitor 8 may also eliminate the reflected ultrasonic waves or delay the propagation of the reflected waves, or it may eliminate both the reflected and transmitted waves, or it may delay the propagation of both the reflected and transmitted waves.

[0063] As shown in Figures 5, 1003 and 1004, the propagation obstruction material 8 is ring-shaped. The propagation obstruction material 8 has a diameter such that it contacts the entire circumference of the pipe end weld and can be attached to the base of the probe.

[0064] When the probe of the ultrasonic flaw detector 7, equipped with the propagation-inhibiting material 8, is inserted into the pipe to be inspected, the propagation-inhibiting material 8 comes into contact with the pipe end weld, as shown in 1001 of Figure 5. When measurements are taken in this state, the propagation of the transmitted wave sent from the probe is delayed at the pipe end side of the pipe end weld by the propagation-inhibiting material 8.

[0065] The effect of this delay appears as feature points in the flaw detection image, so the detection unit 101 can detect these feature points, and the setting unit 102 can set the judgment area based on these feature points. This will be explained with reference to Figure 6. Figure 6 shows a flaw detection image 111C generated by an ultrasonic flaw detection device 7 equipped with a propagation-inhibiting material 8. For comparison, Figure 6 also shows a flaw detection image 111D generated by an ultrasonic flaw detection device 7 without the propagation-inhibiting material 8.

[0066] In flaw detection image 111C, the first reflected echo C1 from the pipe surface and echo C2, which contains a mixture of transmitted and reflected waves and noise, are visible. More specifically, echo C2 appears in flaw detection image 111C in a width Wc area above echo C1. In flaw detection image 111C, the right side corresponds to the inner side of the pipe, and the left side corresponds to the outer side of the pipe.

[0067] In this case, in flaw detection image 111C, echo C2 is interrupted. Here, the region where echo C2 is interrupted is called the echo disappearance region C3. As is clear from comparing the echo disappearance region C3 in flaw detection image 111C with region D1 in flaw detection image 111D, the echo disappearance region does not appear in flaw detection image 111D generated by the ultrasonic flaw detection device 7 without the propagation-inhibiting material 8. In other words, the echo disappearance region C3 is caused by the presence of the propagation-inhibiting material 8, and the position where the echo disappearance region C3 appears corresponds to the position of the propagation-inhibiting material 8.

[0068] As explained with reference to Figure 5, the flaw detection image 111C is generated based on the measurement results when the propagation-inhibiting material 8 is in contact with the pipe end weld. Therefore, the position of the echo disappearance area C3, which appears depending on the position of the propagation-inhibiting material 8, serves as a criterion for identifying the position of the pipe end weld. More specifically, the position of the boundary line L9 between the echo disappearance area C3 and echo C2, that is, the interrupted portion of echo C2, indicates the position of the pipe end side of the pipe end weld.

[0069] Therefore, the detection unit 101 only needs to detect the first reflected echo C1 and the end of echo C2 on the side of the echo disappearance C3 as feature points of the flaw detection image 111C. Then, the setting unit 102 only needs to set a determination area having a width of W or more than the thickness of the pipe end weld, based on the first reflected echo C1 detected by the detection unit 101 and the end of echo C2 on the side of the echo disappearance C3.

[0070] More specifically, the detection unit 101 should input the flaw detection image 111C into a detection model 112 that has been trained to detect the weld-side end of the first reflected echo C1 and the echo-disappearing end C3 of echo C2. This yields the detection results shown by rectangles C4 and C5 in Figure 6. Note that the weld-side end of the reflected echo C1 and the echo-disappearing end C3 of echo C2 may be detected by separate detection models.

[0071] The setting unit 102 then sets a rectangular region C9 (a rectangular region with points C41, C6, C7, and C8 as its four vertices) as the determination region. This region has a side made up of a line segment connecting point C41, the upper right corner of rectangle C4, and point C6, the intersection of a line segment drawn to the left from point C41 and the boundary line L9.

[0072] The setting unit 102 may also define an arc-shaped curve L10 connecting points C7 and C41, and set a sector-shaped determination area enclosed by line segments C6-C7, C6-C41, and curve L10. As shown in Figure 3, a notch may be provided in the tube sheet at the welding location, and weld metal is filled into this notch (called the groove). In Figure 6, the groove is represented by a line segment connecting points C6, C41, and C7 in order. The area around point C8 is further inside the tube than the groove and corresponds to the inside of the tube sheet, so by excluding the area around point C8, an appropriately narrowed determination area can be set.

[0073] Of course, the shape of the determination area set by the setting unit 102 can be any shape, not limited to a rectangle or a sector, as long as it is determined based on the feature points detected by the detection unit 101. For example, the setting unit 102 may set a triangular determination area with vertices at points C41, C6, and C7. In addition, the setting unit 102 may set a determination area that is trapezoidal, parallelogram-shaped, or a combination of multiple types of shapes.

[0074] As described above, the flaw detection image may be an image of echoes measured with an ultrasonic wave propagation inhibiting material 8 placed adjacent to the weld. The propagation inhibiting material 8 may either eliminate at least one of the ultrasonic transmitted waves and ultrasonic reflected waves, or delay the propagation of at least one of the ultrasonic transmitted waves and ultrasonic reflected waves. In this case, the detection unit 101 may detect the interrupted portion where the ultrasonic wave propagation is interrupted in the flaw detection image as a feature point, and the setting unit 102 may set the determination area based on the first reflected echo detected by the detection unit 101 and the interrupted portion.

[0075] As described above, in flaw detection images generated based on measurement results taken with ultrasonic wave propagation inhibiting material placed adjacent to the weld, interruptions occur at the location of the propagation inhibiting material. Therefore, with the above configuration, these interruptions are detected as feature points, and the judgment area is set based on the interruptions and the first reflected echo. This makes it possible to set a reasonable judgment area according to the height of the weld (which can also be said to be the thickness along the axial direction of the pipe).

[0076] The method for detecting interruptions is not limited to the examples described above. For example, the detection unit 101 may detect interruptions in the ultrasonic wave propagation in the flaw detection image based on the echo intensity at each position in a band-shaped region with a width Wc corresponding to the thickness of the object being inspected. With this configuration, interruptions can be detected without prior learning or other procedures.

[0077] For example, the detection unit 101 may divide the pixels constituting the above-mentioned region of width Wc in the flaw detection image 111C into rows of pixels arranged in a vertical direction, and for each row, calculate the sum of the echo intensities corresponding to each pixel included in that row. The detection unit 101 may then detect the position of a row where the sum of the echo intensities falls below a threshold, or the position of the row immediately preceding the sum of the echo intensities falling below a threshold, as an interruption section. In the echo disappearance section C3, the echo intensity is significantly reduced compared to the region where echo C2 is present, so this configuration allows for the appropriate detection of the interruption section.

[0078] [Processing flow] The flow of processing (method of setting the judgment area) executed by the information processing device 1 will be explained with reference to Figure 7. Figure 7 is a flowchart of an example of processing executed by the information processing device 1. At the start of the processing in this flowchart, the generation of a flaw detection image by the ultrasonic flaw detection device 7 for the pipe end weld to be inspected has been completed, and the generated flaw detection image is stored in the storage unit 11 as image 111.

[0079] In S11, the detection unit 101 acquires image 111, which is a flaw detection image generated for the pipe end weld to be inspected. Image 111 is the image used to determine whether the pipe end weld is good or bad.

[0080] In S12 (detection step), the detection unit 101 detects feature points from the image 111 acquired in S11 that have known positional relationships with the judgment area to be used for determining the quality of the pipe end weld. For example, the detection unit 101 may detect the first and second reflected echoes from the pipe surface surrounding the pipe end weld (two locations each on the inner and outer sides of the pipe) as feature points. Specifically, the detection unit 101 can detect feature points based on the output values ​​obtained by inputting the image 111 into the detection model 112.

[0081] In S13 (setting step), the setting unit 102 sets the determination area based on the feature points detected in S12. For example, if the first and second reflected echoes from the pipe surface around the pipe end weld (two locations each on the inner and outer sides of the pipe) were detected as feature points, the setting unit 102 sets the area enclosed by these four feature points as the determination area.

[0082] In S14, the determination unit 103 determines whether the pipe end weld shown in the image 111 acquired in S11 is good or bad. Specifically, the determination unit 103 extracts the portion of the determination area set in S13 from the image 111, inputs the extracted image into the determination model 113, and makes a good or bad determination based on the output value obtained. The determination unit 103 then stores the result of the good or bad determination as the determination result 114 in the storage unit 11, and the process in Figure 7 is completed.

[0083] As described above, the method for setting the judgment region executed by the information processing device 1 includes a detection step (S12) in which a known positional relationship with the judgment region to be inspected is detected from the image 111 of the object to be inspected, and a setting step (S13) in which the judgment region is set based on the feature points detected in the detection step. This setting method makes it possible to automatically set a reasonable judgment region in the image of the object to be inspected.

[0084] [Embodiment 2] Other embodiments of the present invention are described below. For the sake of clarity, components having the same function as those described in the above embodiments will be denoted by the same reference numerals, and their descriptions will not be repeated. This also applies to Embodiment 3 and subsequent embodiments.

[0085] This embodiment describes an information processing device 2 that automatically switches the method of setting the determination area depending on whether the object to be inspected or the welded area is thicker. Figure 8 is a block diagram showing an example of the main components of the information processing device 2. The information processing device 2 includes a control unit 20, which includes a thickness determination unit 201. In addition, the information processing device 2 includes a setting unit 202 instead of the setting unit 102 described in Embodiment 1.

[0086] The thickness determination unit 201 determines which is thicker, the object under inspection or the welded area. For example, if the object under inspection is a welded pipe and the welded area is a pipe end weld at the end of the pipe, the thickness determination unit 201 determines which is thicker, the pipe or the pipe end weld.

[0087] The method for determining which is thicker, the object being inspected or the welded area, is not particularly limited. For example, as mentioned above, the groove width, which affects the thickness of the pipe end weld, is determined by design, and the pipe thickness is also predetermined. Therefore, the thickness determination unit 201 may obtain these design values ​​by having the user input them from the input unit 13, calculate the thickness of the pipe end weld from the acquired groove width design value, and determine which is thicker, the calculated value or the acquired pipe thickness design value.

[0088] The detection unit 101 detects the first and second reflected echoes of the ultrasonic waves reflected from the surface surrounding the welded area of ​​the object being inspected as feature points. The detection model 112 is used for this detection.

[0089] If the thickness determination unit 201 determines that the thickness of the object to be inspected is equal to or greater than the thickness of the welded area, the setting unit 202 sets the region between the first and second reflected echoes as the determination region.

[0090] On the other hand, if the thickness determination unit 201 determines that the thickness of the object to be inspected is less than the thickness of the weld, the setting unit 202 sets a determination area having a width equal to or greater than the thickness of the weld, based on the first reflected echo. This width may be calculated, for example, based on the sum of the penetration width on the pipe side, the groove width, and the penetration width on the tube sheet side, similar to Embodiment 1.

[0091] As described above, if the thickness of the object to be inspected is greater than or equal to the thickness of the weld, the information processing device 2 sets the region between the first and second reflected echoes as the determination region. Since the interval between the first and second reflected echoes is approximately equal to the thickness of the object to be inspected, the weld is included within the determination region set in this way.

[0092] Furthermore, with the above configuration, if the thickness of the object being inspected is less than the thickness of the weld, a judgment area with a width greater than or equal to the thickness of the weld is set based on the first reflected echo. This makes it possible to set a judgment area that covers the entire weld, which is thicker than the object being inspected.

[0093] Thus, with the above configuration, an appropriate width for the judgment area can be automatically set whether the thickness of the object to be inspected is greater than or equal to the thickness of the weld, or whether the thickness of the object to be inspected is less than the thickness of the weld.

[0094] Furthermore, when the welding location is near the end of the object to be inspected, it is preferable for the detection unit 101 to acquire an image generated based on the measurement results obtained with an ultrasonic wave propagation inhibiting material placed near the welding location. In this case, it is preferable for the detection unit 101 to detect the interrupted portion where the propagation of ultrasonic waves is interrupted as a feature point in the image. In this case, the setting unit 202 can determine the position of the end of the pipe end of the judgment area based on the interrupted portion.

[0095] [Processing flow] The flow of processing (method of setting the determination area) executed by the information processing device 2 will be explained based on Figure 9. Figure 9 is a flowchart of an example of processing executed by the information processing device 2. Note that S21 and S26 are the same as S11 and S14 in Figure 7, respectively, so the explanation will not be repeated here.

[0096] In S22 (detection step), the detection unit 101 detects the first and second reflected echoes of the ultrasonic waves reflected from the surface around the weld area of ​​the object to be inspected as feature points from the image 111 acquired in S21. The detection model 112 can be used to detect these reflected echoes. If the image 111 is generated by an ultrasonic flaw detector 7 equipped with a propagation-inhibiting material, the detection unit 101 may also detect interrupted portions where the propagation of ultrasonic waves was interrupted due to the propagation-inhibiting material as feature points, in addition to the first and second reflected echoes.

[0097] In S23, the thickness determination unit 201 determines whether the thickness of the object to be inspected is equal to or greater than the thickness of the weld. If it is determined that the thickness of the object to be inspected is equal to or greater than the thickness of the weld (YES in S23), the process proceeds to S24. On the other hand, if it is determined that the thickness of the object to be inspected is less than the thickness of the weld (NO in S23), the process proceeds to S25. Note that the determination in S23 may be performed in parallel with S22, or before S22.

[0098] In S24 (setting step), the setting unit 202 sets a determination area with a width greater than or equal to the thickness of the welded area, based on the first reflected echo detected in S22. If an interrupted portion where ultrasonic wave propagation was interrupted due to a propagation-inhibiting material was detected as a feature point in S22, the setting unit 202 sets the determination area based on a straight line (L9 in Figure 6) defined based on that interrupted portion and the first reflected echo.

[0099] In S25 (setting step), the setting unit 202 sets the region between the first and second reflected echoes detected in S22 as the determination region. If an interruption in the propagation of ultrasound caused by a propagation-inhibiting material was detected as a feature point in S22, the setting unit 202 sets the determination region based on a straight line (L9 in Figure 6) defined based on that interruption, and the first and second reflected echoes.

[0100] [Embodiment 3] In this embodiment, an example of performing a radiographic test (RT) of a pipe joint weld using a CR (Computed Radiography) image will be explained with reference to Figure 10. Figure 10 shows the configuration of a pipe 2000 including a joint weld, which is an example of an object to be inspected, and its CR image 111D.

[0101] Figure 10 shows a perspective view and a side view of the pipe 2000. As shown in the figure, the pipe 2000 is configured in which a first pipe section 2001 and a second pipe section 2002 are connected by a welded joint 2003. Radiation is irradiated from the RT device 3000 onto the welded joint 2003 of the pipe 2000, and the radiation that passes through the pipe 2000 is imaged by the IP (Imaging Plate) 4000. The IP 4000 is positioned so that its back side is in contact with the pipe 2000. When inspecting the portion of the annularly formed welded joint 2003 that is farther from the RT device 3000 (closer to the IP 4000), the radiation is incident on the annularly formed welded joint 2003 at an inclined angle, as shown in the figure.

[0102] Figure 10 shows a side view of pipe 2000, viewed from the side where the IP4000 is located. As shown in the figure, the pipe 2000 is marked with numbers 1 to 4 (5001 to 5004). These numbers are placed to surround the area to be inspected in the welded joint 2003, serving as a reference when setting the judgment area. For example, such markings can be made by attaching numbered stickers to pipe 2000. Of course, the markings on pipe 2000 are not limited to numbers; letters, symbols, or anything else can be used.

[0103] In CR image 111D, the amount of radiation absorbed by the object being inspected is shown as varying shades of gray. For example, the welded section 2003 is made of a different material than the first pipe section 2001 and the second pipe section 2002, and therefore absorbs different amounts of radiation. For this reason, in CR image 111D, the welded section 2003 is shown as images D1 and D2. As mentioned above, because the angle of incidence of the radiation was tilted, the half of the annular welded section 2003 closer to the RT device 3000 is shown as an arc-shaped image D2, and the half of the annular welded section 2003 closer to the IP4000 is shown as a roughly linear image D1. The width of image D1 is Wd.

[0104] Furthermore, Figure 10 shows a step wedge (film density comparator) D3 superimposed on the CR image 111D. By comparing it with the step wedge D3, the amount of radiation absorbed in each part of the CR image 111D can be determined. It is desirable that the step wedge D3 be made of the same material as the test subject, the pipe 2000. For example, if the pipe 2000 is made of austenitic stainless steel SUS304, it is desirable that the step wedge D3 also be made of SUS304.

[0105] When inspecting a weld using the CR image 111D, it is necessary to set a judgment area that includes image D1. In the example in Figure 10, the numbers 1 to 4 that appear in the CR image 111D are detected as feature points, and the judgment area is set based on these feature points. In Figure 10, the detection results of the numbers 1 to 4 by the detection unit 101 are shown by rectangles D4 to D7. Note that the numbers in the CR image 111D are not horizontally inverted because the IP4000 is positioned so that its back side is in contact with the pipe 2000. Also in Figure 10, the judgment area set by the setting unit 102 is shown by rectangle D8. Rectangle D8 is a rectangle with four vertices: the top left vertex D41 of rectangle D4, the top right vertex D51 of rectangle D5, the bottom right vertex D61 of rectangle D6, and the bottom left vertex D71 of rectangle D7.

[0106] Since rectangle D8 includes the image D1 of the weld, rectangle D8 is an appropriate judgment area. When performing a quality inspection of the weld 2003, it is preferable to set the judgment area to a position slightly away from the opening tip of the weld 2003. For example, in the example in Figure 10, the judgment area is set to a range from 5 mm above the upper opening tip of the image D1 of the weld with width Wd to 5 mm below the lower opening tip of the weld with width Wd.

[0107] As described above, for objects to be inspected that do not have suitable feature points to use as a reference, a mark that appears in the image of the object to be inspected may be attached to the object. This makes it possible to set an appropriate judgment area based on the mark that appears in the image. Furthermore, as described above, the information processing device 1 can be used not only for non-destructive testing using ultrasound as described in Embodiment 1, but also for non-destructive testing using radiation.

[0108] [Embodiment 4] In this embodiment, an example of performing a condition inspection of an object to be inspected using an image captured by an optical camera with the information processing device 1 will be explained with reference to Figure 11. Figure 11 shows a piston crown 6001, which is an example of an object to be inspected, and its surrounding structure, as well as an image 111E of the piston crown 6001.

[0109] Figure 11 shows the piston crown 6001 and its surrounding structure. The piston crown 6001 is used in engines and the like, and Figure 11 shows the piston crown 6001 of a marine engine. As shown in the figure, the piston crown 6001 is provided at the end of the piston rod 6003. A piston ring 6002 is also provided around the piston crown 6001. The piston crown 6001 reciprocates within the combustion chamber 6004 when the engine is running.

[0110] Furthermore, the combustion chamber 6004 is provided with a scavenging port 6005 for drawing air into the combustion chamber 6004. Part of the piston crown 6001 is visible through the scavenging port 6005. Image 111E shows the piston crown 6001 as seen through the scavenging port 6005.

[0111] Image 111E shows the scavenging port 6005, the piston crown 6001, and the piston ring 6002. When inspecting the piston crown 6001 using image 111E, it is necessary to set a judgment area that includes the image of the piston crown 6001. In the example in Figure 11, the four corners of the scavenging port 6005 shown in image 111E are detected as feature points, and the judgment area is set based on these feature points.

[0112] More specifically, Figure 11 shows the detection results of the corners of the scavenging port 6005 by the detection unit 101 as rectangles E01 to E12. Also in Figure 11, the determination area set by the setting unit 102 is shown as rectangle E13. Rectangle E13 is a rectangle with four vertices: the upper left vertex E011 of rectangle E01 detected at the upper left corner, the upper right vertex E061 of rectangle E06 detected at the upper right corner, the lower right vertex E121 of rectangle E12 detected at the lower right corner, and the lower left vertex E071 of rectangle E07 detected at the lower left corner.

[0113] Since rectangle E12 contains an image of the piston crown 6001, rectangle E12 is a suitable determination region. The determination unit 103 can determine the state of the piston crown 6001 (for example, the state of oil adhesion or the presence or absence of notched scratches) by targeting the determination region indicated by rectangle E12. When setting such a determination region, rectangles E02~E05 and E08~E11 do not need to be detected as feature points. When rectangles E01~E12 are detected as feature points, a determination region may be set for each scavenging port 6005. In other words, the setting unit 102 may set the determination region based on rectangles E01, E02, E07, and E08, as well as based on rectangles E03, E04, E09, and E10, and further based on rectangles E05, E06, E011, and E12.

[0114] As described above, the information processing device 1 can also be used for inspections using images captured by an optical camera. Furthermore, as described above, the information processing device 1 may detect as feature points those objects in the image of the object to be inspected that are clearly visible and easy to detect, among those objects that appear around the judgment area to be inspected. This makes it possible to detect these feature points with high accuracy and to set a judgment area with high accuracy.

[0115] [Variation] The entity executing each process described in the above embodiments is arbitrary and is not limited to the examples given. In other words, an information processing system having the same functions as information processing device 1 or information processing device 2 can be constructed using multiple information processing devices that can communicate with each other. For example, in the method for setting the determination area shown in Figure 7, processes S11 and S12, process S13, and process S14 may be executed by different information processing devices, respectively. In other words, the entity executing this method for setting the determination area may be a single information processing device 1 or multiple information processing devices. The same applies to the method for setting the determination area shown in Figure 9.

[0116] [Examples of implementation using software] The functions of the information processing devices 1 and 2 (hereinafter referred to as "devices") can be realized by programs that cause the devices to function as computers, and by programs that cause the computers to function as each control block of the devices (particularly each part included in the control unit 10 or 20) (decision area setting programs).

[0117] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., memory) as hardware for executing the program. By executing the program using this control device and storage device, the functions described in each of the embodiments are realized.

[0118] The above program may be recorded on one or more computer-readable recording media, not temporary ones. These recording media may or may not be provided by the above device. In the latter case, the program may be supplied to the above device via any wired or wireless transmission medium.

[0119] Furthermore, some or all of the functions of each of the above control blocks can also be realized by logic circuits. For example, an integrated circuit in which logic circuits functioning as each of the above control blocks are formed is also included in the scope of the present invention. In addition, it is also possible to realize the functions of each of the above control blocks by, for example, a quantum computer.

[0120] The present invention is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention. [Explanation of Symbols]

[0121] 1. Information Processing Device 101 Detection unit 102 Settings Section 111 images 112 detection models 2. Information Processing Device 201 Thickness determination unit 202 Settings Section

Claims

1. A detection unit that detects characteristic points whose positional relationship with the judgment area to be inspected is known from an image of the object to be inspected, The system includes a setting unit that sets the determination area based on the feature points detected by the detection unit, The detection unit detects the feature points from the image using a detection model that has learned the appearance of the feature points through machine learning. The aforementioned image is a flaw detection image obtained by imaging the ultrasonic echo propagated around the welded area of ​​the object being inspected. The part of the object to be inspected is the pipe end weld, which is the welded joint between the metal pipe and the metal tube sheet that bundles the pipes together. The above image was obtained by propagating ultrasonic waves from the inner wall surface of the pipe toward the pipe end weld and measuring the resulting echo. The detection model is an information processing device that has been trained using machine learning to detect, for each of the sets of first reflected echoes from a set of echoes in which ultrasonic waves are reflected multiple times near both ends of the pipe end weld, the portion of each reflected echo on the pipe end weld side as a feature point, rather than the entire reflected echo.

2. The detection unit detects the portion of the end on the pipe end weld side of the first reflected echo of the ultrasonic waves reflected around the welded area of ​​the object to be inspected as the characteristic point. The information processing apparatus according to claim 1, wherein the setting unit sets the determination area used for determining the quality of the welded portion of the pipe end weld in the first reflected echo detected by the detection unit.

3. A detection unit that detects characteristic points whose positional relationship with the judgment area to be inspected is known from an image of the object to be inspected, The system includes a setting unit that sets the determination area based on the feature points detected by the detection unit, The aforementioned image is a flaw detection image obtained by imaging the ultrasonic echo propagated around the welded area of ​​the object being inspected. The detection unit detects the first reflected echo from the echoes of the ultrasonic waves reflected around the welded area of ​​the object to be inspected as the feature point. The setting unit sets the judgment area used for determining the quality of the welded area, based on the first reflected echo detected by the detection unit. The system includes a determination unit that determines which of the object to be inspected or the welded area is thicker. The detection unit also detects the second reflected echo of the ultrasonic waves reflected from the surface surrounding the welded area of ​​the object to be inspected as a characteristic point. The aforementioned setting unit is, If the determination unit determines that the thickness of the object to be inspected is equal to or greater than the thickness of the welded area, the region between the first reflected echo and the second reflected echo is set as the determination region. If the determination unit determines that the thickness of the object to be inspected is less than the thickness of the welded area, the information processing device sets the determination area having a width equal to or greater than the thickness of the welded area, based on the first reflected echo.

4. A detection unit that detects characteristic points whose positional relationship with the judgment area to be inspected is known from an image of the object to be inspected, The system includes a setting unit that sets the determination area based on the feature points detected by the detection unit, The aforementioned image is a flaw detection image obtained by imaging the ultrasonic echo propagated around the welded area of ​​the object being inspected. The detection unit detects the first reflected echo from the echoes of the ultrasonic waves reflected around the welded area of ​​the object to be inspected as the feature point. The setting unit sets the judgment area used for determining the quality of the welded area, based on the first reflected echo detected by the detection unit. The aforementioned flaw detection image is an image of the echo measured while a propagation-inhibiting material is placed adjacent to the welded area, which eliminates at least one of the transmitted ultrasonic waves and the reflected ultrasonic waves, or delays the propagation of at least one of the transmitted ultrasonic waves and the reflected ultrasonic waves. The detection unit detects the interrupted portions in the flaw detection image where the propagation of ultrasonic waves is interrupted as feature points. The setting unit is an information processing device that sets the determination area based on the first reflected echo detected by the detection unit and the interrupted portion.

5. The information processing apparatus according to claim 4, wherein the detection unit detects the interrupted portion based on the echo intensity at each position in a band-shaped region corresponding to the thickness of the object to be inspected.

6. A method for setting a determination area to be executed by one or more information processing devices, A detection step involves detecting characteristic points whose positional relationship with the judgment area to be inspected is known from an image of the object to be inspected, The set step includes setting the determination region based on the feature points detected in the detection step, In the detection step, the feature points are detected from the image using a detection model that has been trained to recognize the appearance of the feature points. The aforementioned image is a flaw detection image obtained by imaging the ultrasonic echo propagated around the welded area of ​​the object being inspected. The part of the object to be inspected is the pipe end weld, which is the welded joint between the metal pipe and the metal tube sheet that bundles the pipes together. The above image was obtained by propagating ultrasonic waves from the inner wall surface of the pipe toward the pipe end weld and measuring the resulting echo. The method for setting the determination area is a machine learning model in which the detection model is selected to detect, for each of the sets of first reflected echoes from a set of echoes in which ultrasonic waves are reflected multiple times near both ends of the pipe end weld, the portion of the end of each reflected echo on the pipe end weld side, rather than the entire reflected echo, as the feature point.

7. A determination area setting program for causing a computer to function as an information processing device according to claim 1, wherein the detection unit and the setting unit are the same as the computer.

Images