Image-processing device, image-processing method, and image-processing program

Abstract

This image-processing device comprises a processor. The processor acquires a first captured image obtained by irradiating a welded part of two base materials welded to each other with radiation from a first direction. The first direction is a direction inclined with respect to a bisector of an included angle by an angle smaller than half of the included angle, the included angle being formed by weld surfaces of the two base materials.

Description

Image processing apparatus, image processing method, and image processing program

[0001] The present disclosure relates to an image processing apparatus, an image processing method, and an image processing program.

[0002] Japanese Patent Laid-Open No. 8-247749 discloses that X-rays from an X-ray tube are irradiated onto a spot welded portion of an inspection object from a direction perpendicular or oblique to the surface of a panel material, and a transmission image thereof is caused to appear on a fluorescent screen. From this X-ray transmission image, a transmission image of the structure interface between a nugget as a melted portion where a structural change has occurred due to welding heat during welding at the spot welded portion and a base material where no structural change occurs is detected. A welding inspection method is disclosed in which the size, shape, etc. of the nugget are detected from the density, shape and dimensions, etc. of the transmission image of the structural interface, and the quality of the spot welded portion is inspected.

[0003] Japanese Patent Laid-Open No. 2018-190509 discloses an inspection method for a secondary battery including an electrode assembly, a case body that accommodates the electrode assembly, and a case having a lid that closes an opening of the case body, wherein a welded portion where the outer edge of the case body and the outer edge of the lid are welded to each other over the entire continuous area thereof, the method including: an acquisition step of acquiring a fluoroscopic image of the welded portion by irradiating X-rays toward the welded portion and detecting the X-rays transmitted through the welded portion; and a determination step of determining whether the joining state of the welded portion is good or not using the fluoroscopic image of the welded portion acquired in the acquisition step, wherein in the acquisition step, X-rays are irradiated toward the welded portion with the case tilted relative to the X-ray irradiation direction, and the case is relatively moved and relatively rotated to acquire a fluoroscopic image over the entire circumference of the welded portion. An inspection method for a power storage device is disclosed.

[0004] Japanese Patent Laid-Open No. 2020-112524 discloses a welding inspection system including: an X-ray imaging apparatus that irradiates X-rays onto a welded portion of an inspection object (steel pipe) to obtain an X-ray transmission image; an input unit that inputs structural information indicating the structure of the inspection object; a combining unit that combines structural information corresponding to the X-ray transmission image among the structural information with the X-ray transmission image; and a display unit that enables inspection for abnormalities in the welded portion by displaying the combined image combined by the combining unit. The welding inspection system is disclosed.

[0005] In non-destructive testing, radiographic testing is a well-known method that uses radiation, such as X-rays, to image objects being inspected, such as welds, and visualizes the presence of internal defects based on differences in transmission density.

[0006] However, the images obtained from radiographic testing are two-dimensional transmission images, and since information in the depth direction cannot be obtained, it was sometimes impossible to determine the type of defect.

[0007] This disclosure aims to provide an image processing apparatus, an image processing method, and an image processing program capable of improving the accuracy of determining the type of defect.

[0008] To achieve the above objective, the image processing apparatus of the present disclosure comprises a processor which acquires a first image obtained by irradiating the welded joint of two welded base materials with radiation from a first direction, the first direction being a direction inclined with respect to the angle bisector of the angle between the welded surfaces of the two base materials by an angle smaller than half of the angle between the welded surfaces.

[0009] The processor may further acquire a second image obtained by irradiating the weld from a second direction different from the first direction.

[0010] The processor may acquire a second image, in which the second direction is along the angle bisector of the angle between the two sides.

[0011] The processor may acquire a second image, in which the second direction is symmetric to the first direction with respect to the angle bisector.

[0012] The processor may identify at least one of the location and type of defect in the weld based on the first captured image.

[0013] If the processor cannot identify a single type, it may present multiple types in order of likelihood, or in order of least likelihood.

[0014] The processor may perform processing to display the first and second captured images on the display unit.

[0015] The processor may use the first and second captured images to estimate the volume of the defect in the weld and use the estimated volume to determine the severity of the defect.

[0016] The processor estimates at least one of the defect type, severity, depth, and volume of a defect, using the lengths of one or more line segments that cross the defect image, if a defect image corresponding to a defect in the weld is included in the first captured image.

[0017] The processor further acquires a second image obtained by irradiating the weld from a second direction different from the first direction, and if multiple defect images corresponding to multiple defects contained in the weld are included in both the first and second images, it identifies at least one of the location and type of the defect based on the first positional relationship of the multiple defect images in the first image and the second positional relationship of the multiple defect images in the second image.

[0018] The image processing method of the present disclosure involves a computer performing a process to acquire a first image obtained by irradiating the welded joint of two welded base materials with radiation from a first direction, wherein the first direction is a direction inclined with respect to the angle bisector of the angle between the welded surfaces of the two base materials by an angle smaller than half of the angle between the welded surfaces.

[0019] The image processing program of this disclosure causes a computer to perform a process of acquiring a first image obtained by irradiating the welded joint of two welded base materials with radiation from a first direction, wherein the first direction is a direction inclined with respect to the angle bisector of the angle between the welded surfaces of the two base materials by an angle smaller than half of the angle between the welded surfaces.

[0020] According to this disclosure, it is possible to provide an image processing apparatus, an image processing method, and an image processing program that can improve the accuracy of determining the type of defect.

[0021] This is a block diagram showing an example of the overall configuration of the radiation imaging system according to this embodiment. This is a block diagram showing an example of the configuration of the main part of the control device. This is a diagram showing an example of a "crack". This is a diagram showing an example of transmission imaging, in which radiation is incident perpendicularly onto the object to be inspected. This is a diagram showing an example of a type of defect that occurs in welding. This is a diagram showing an example of "failure to fuse". This is a diagram showing an example of frontal and oblique imaging. This is a perspective view showing an example of butt welding. This is a side view showing an example of butt welding. This is a perspective view showing an example of fillet welding. This is a side view showing an example of fillet welding. This is a diagram showing an example of irradiation direction symmetric with respect to the reference direction K. This is a diagram showing an example of two line segments crossing a defect image. This is a diagram showing an example of the case where the variation in the length of the two line segments crossing a defect image is less than a predetermined range and above a predetermined range. This is a diagram for explaining the estimation of the depth of the defect. This is a diagram showing an example of an image with two defect images. This is a diagram showing the positions of the two defects in the depth direction. This is a flowchart showing an example of the processing flow performed in the control unit of the control device in the radiation imaging system according to this embodiment.

[0022] Embodiments of this disclosure will be described in detail below with reference to the drawings. These embodiments are not intended to limit the present invention.

[0023] In this embodiment, the control device included in the radiation imaging system is described as an example of an information processing device. The radiation imaging system according to this embodiment is used for non-destructive inspection (NDI) of objects to be inspected. Figure 1 is a block diagram showing an example of the overall configuration of the radiation imaging system according to this embodiment.

[0024] Non-destructive testing refers to the process of determining whether an object to be inspected meets certain standards, such as passing or failing an object, based on the results of non-destructive testing (NDT), including radiographic testing (RT), and criteria specified in various standards such as JIS (Japanese Industrial Standards) or customer specifications. Non-destructive testing is the inspection related to the technology described in this disclosure and will also be simply referred to as inspection below.

[0025] The object to be inspected W shown in Figure 1 is an example of an object to be inspected according to the technology of this application, and as will be described later, it includes a bead 40, which is a welded joint formed by welding two flat metal base materials together (see Figure 4, etc.). In non-destructive testing, for example, by examining the properties inside the bead 40, it is possible to determine whether the processing quality and deterioration state conform to the standards.

[0026] Non-destructive testing (NDT) is a test that examines the properties of an object W without destroying the object W, including the presence, location, size, shape, and distribution of defects within the bead 40 of the object W. In addition to radiographic testing, other non-destructive testing methods include ultrasonic testing. Radiographic testing (RT) is a test that examines the properties of an object W by using radiation R to photograph the object W and obtaining the resulting image P (see Figure 4). Photographing an object W using radiation R means obtaining an image P by irradiating the object W with radiation R. More specifically, the image P is a radiographic image representing the two-dimensional intensity distribution of radiation R that has passed through the object W, and is a projected image in which the internal structure of the object W is projected two-dimensionally. Image P is an example of an "image" related to the technology of this disclosure.

[0027] Regarding radiographic testing methods for objects W that include welds, various methods are specified in JIS standards, for example, depending on the type of object W. Examples include "Radiographic Testing Method for Steel Welded Joints" (JIS Z 3104), "Radiographic Testing Method for Aluminum Welded Joints" (JIS Z 3105), "Radiographic Testing Method for Stainless Steel Welded Joints" (JIS Z 3106), and "Radiographic Testing Method for Titanium Welds" (JIS Z 3107). Each of these test methods specifies the test procedure and the method for evaluating defects.

[0028] The radiographic imaging system is used to examine the properties of the object W under inspection through radiographic testing. The results of the radiographic testing are used for a comprehensive assessment in non-destructive testing, as will be described later.

[0029] As shown in Figure 1, the radiation imaging system 10 includes a radiation irradiation device 12, a radiation imaging device 16, and a control device 18.

[0030] The radiation irradiation device 12 according to this embodiment includes a radiation source 14 that irradiates the object W to be inspected with radiation R, such as X-rays and gamma rays. The method of instructing the radiation irradiation device 12 to irradiate with radiation R is not particularly limited. For example, if the radiation irradiation device 12 is equipped with an irradiation button, a user such as a radiologist may instruct the radiation irradiation device 12 to irradiate with radiation R by pressing the irradiation button. Alternatively, for example, a user such as a radiologist may instruct the radiation irradiation device 12 to irradiate with radiation R by operating a control device 18.

[0031] When the radiation irradiation device 12 receives an instruction to irradiate with radiation R, it irradiates with radiation R from the radiation source 14 according to the set irradiation conditions such as tube voltage, tube current, and irradiation period. For example, the radiation source 14 emits radiation R radially from a focal point, and the beam of radiation R diverges with the focal point as its apex, spreading out in a cone shape. The radiation source 14 is equipped with an irradiation field limiter (also called a collimator) that limits the irradiation field of radiation R, and for example, the irradiation field of radiation R is limited to a rectangular shape. One end of the central axis A of the beam is the focal point of the radiation source 14, and the other end is the center of the irradiation field.

[0032] The radiation imaging apparatus 16 according to this embodiment includes a radiation detector 20 that detects radiation R irradiated from the radiation irradiation device 12 and transmitted through the object to be inspected W. The radiation detector 20 has a detection surface in which a plurality of pixels that convert the intensity of radiation R into electrical signals are arranged in a matrix, and is also called a flat panel detector or DR cassette (Digital Radiography). The radiation detector 20 outputs an electrical signal corresponding to the intensity of radiation R incident on each pixel of the detection surface. The radiation imaging apparatus 16 generates an image P (see Figure 4) of the object to be inspected W by performing signal processing on the electrical signal output by the radiation detector 20. The generated image P is output to the control device 18. In this way, the object to be inspected W is imaged using radiation R, and the image P is acquired.

[0033] In addition to using a radiation detector 20 as the image receiving unit, the radiation imaging device 16 may also use a CR (Computed Radiography) cassette or a photographic film for radiation as the image receiving unit.

[0034] As shown in Figure 2, the control device 18 includes a control unit 22, a display unit 32, an operation unit 34, and a communication I / F (interface) unit 36. Figure 2 is a block diagram showing an example of the configuration of the main parts of the control device 18.

[0035] The main part of the control unit 22 has a general computer configuration and controls the overall operation of the control device 18.

[0036] The control unit 22 includes a CPU 24, a ROM 26, and a RAM 28. The ROM 26 pre-stores various programs 30 executed by the CPU 24, such as a shooting processing program and a display processing program that are executed when controlling shooting. The RAM 28 temporarily stores various data.

[0037] The display unit 32 displays the screen for operating the radiation irradiation device 12 and the radiation imaging device 16, as well as captured images. For example, the display unit 32 may be a direct-view type electronic display.

[0038] The control unit 34 is used by the user to input instructions and various information related to shooting. The control unit 34 is not particularly limited and can include, for example, various switches, a touch panel, a stylus, and a mouse.

[0039] The communication I / F unit 36 ​​communicates captured images and various information with the radiation irradiation device 12 and the radiation image acquisition device 16 via wireless or wired communication.

[0040] In radiographic testing, it is possible to detect and visualize defects D inside the object W being inspected. Within the object W, the transmittance of radiation R differs between the defects D and the sound areas, and accordingly, the intensity of radiation R incident on each pixel of the detection surface of the radiation detector 20 also changes. This difference in radiation R intensity appears as a change in density in the captured image P. Generally, defects D tend to have a lower material density compared to sound areas, so the transmittance of defects D is relatively high. Therefore, the density of the area corresponding to defects D in the captured image P is higher than that of the sound areas. In this way, the area corresponding to defects D in the captured image P has different density information than the sound areas, making it possible to visualize defects D.

[0041] In the technology relating to this disclosure, a defect D present in the object W being inspected refers to a discontinuous portion of the object W that lacks structural continuity with the sound portion, or a portion that is incompletely processed, due to manufacturing defects, processing defects, deterioration over time, etc. Furthermore, a distinction is made between the defect D present in the object W being inspected and the portion of the defect D present in the captured image P, and the image of the portion corresponding to the defect D in the captured image P is called a defect image DI (see Figure 4). A defect image DI refers to a characteristic image in the captured image P that corresponds to a region where the density is discontinuous or where the shape is incomplete. Examples of defects D include, for example, welding defects such as blowholes, cracks, poor penetration, and poor fusion in a welded joint. The concept of defect D includes "flaws" in the JIS standard. Each defect image DI corresponding to these defects D has a form according to the characteristics of each defect D.

[0042] In the radiographic testing method specified by JIS standards, defects D in the object under inspection W are detected from the defect image DI in the captured image P, and the type of the detected defect D (hereinafter also referred to as the defect type) is determined. Then, in the radiographic test, points are assigned according to the determined defect type, or the defects D are classified based on the determined defect type, taking into account the thickness of the base material and other factors, and points are assigned according to the classification result of the defects D. Based on the results of such radiographic testing, the final pass or fail of the inspection of the object under inspection W is determined by comprehensively evaluating the detected defects D. Therefore, the defect type is information for investigating the properties of the detected defects D and is of important importance as information necessary for inspection.

[0043] Figures 3 and 4 conceptually illustrate the radiographic testing method for a welded object W. The object W shown in Figures 3 and 4 is formed by butting the edges of two flat metal plates, base materials 42 and 44, with their planes parallel, and welding the joined edges together. When welding is performed, the edges of each base material 42 and 44 are melted and joined, and a welded area called a bead 40 is formed at the boundary between each base material 42 and 44. The bead 40 is the raised portion created by melting. Radiographic testing is performed to investigate processing defects or deterioration over time in the bead 40, and the presence, location, and type of defects in the bead 40 are examined. In such radiographic testing, the bead 40 is the main part, so the bead 40 is the main subject of the imaging.

[0044] For example, in arc welding, where the joint portion of the base materials 42 and 44 is heated by an arc discharge, the width and height of the bead 40 are often greater on the heating source side, as shown in Figure 4. In radiographic testing of the object W including the bead 40, the purpose is to investigate processing defects or deterioration over time in the bead 40, and therefore, an image P with the bead 40 as the main subject is acquired. If the side of the object W with a greater height of the bead 40 is considered the front side of the object W, then, as shown in Figures 3 and 4, the radiation source 14 is positioned facing the front side of the object W, and the radiation R is irradiated toward the front side of the object W.

[0045] Basically, the inspection object W, the radiation source 14, and the radiation detector 20 are positioned such that the bead 40 is captured in the center of the captured image P. The bead 40 extends linearly along the boundary between each base material 42 and 44. The radiation source 14 is positioned such that the central axis A of the radiation beam is located substantially at the center in the width direction of the bead 40, which is orthogonal to the extending direction of the bead 40. Furthermore, the irradiation direction of the radiation R is set such that the radiation R is irradiated directly toward the front of the bead 40.

[0046] Here, the irradiation direction of the radiation R refers to the direction in which the radiation R emitted from the radiation source 14 irradiates the inspection object W. This can also be rephrased as the incident direction or incident angle of the radiation R with respect to the inspection object W. The irradiation direction is determined by the relative positional relationship including the relative position or posture between the central axis A of the radiation beam of the radiation R emitted from the radiation source 14 and the inspection object W. That is, the irradiation direction is a concept that includes any of the following cases: the direction changes when the radiation source 14 is displaced while the inspection object W is fixed, the direction changes when the inspection object W is displaced while the radiation source 14 is fixed, and the direction changes when both the inspection object W and the radiation source 14 are displaced.

[0047] Since the bead 40 is located at the boundary between the two base materials 42 and 44, the irradiation direction in which the radiation R is irradiated directly toward the front of the bead 40 (hereinafter simply referred to as the front direction of the bead 40) is defined based on the included angle α between the two base materials 42 and 44. The included angle α between the two base materials 42 and 44 is the inter-base-material angle formed by the respective surfaces of the two base materials 42 and 44. The front direction of the bead 40 is the direction in which the central axis A of the radiation beam of the radiation R follows the bisector that bisects the included angle α. Here, the surface refers to the surface on the side where the bead 40 is formed in each of the two base materials 42 and 44 to be welded, and is also referred to as the welding surface. Here, the welding surface is an example of the "welding surface" according to the technology of the present disclosure, and the "included angle α" is an example of the included angle formed by the welding surfaces of two base materials.

[0048] Specifically, the inspection object W shown in FIGS. 3 and 4 is joined in an attitude where the surfaces of the two base materials 42 and 44 are parallel, so the included angle α is 180 degrees. Such a joining method is also called butt jointing. In butt jointing, the included angle α between the two base materials 42 and 44 is 180 degrees, and the bisector extends in the 90-degree direction, which is half of the included angle α. Here, the direction in which this bisector extends is referred to as the reference direction K. When the irradiation direction of the radiation R is set to the front direction of the bead 40, the radiation source 14 is aligned to a position facing the bead 40 in a plane orthogonal to the reference direction K, and the attitude is adjusted and positioned such that the central axis A of the radiation flux extends along the reference direction K. Thereby, the irradiation direction of the radiation R becomes the front direction of the bead 40. As described above, the front direction of the bead 40 is defined with the reference direction K as a reference.

[0049] In accordance with the radiation source 14 positioned as described above, the radiation detector 20 is disposed in an attitude facing the radiation source 14 with the inspection object W interposed therebetween. In FIG. 4, a captured image P is drawn at the position of the radiation detector 20, and this is for conceptually showing the relationship between a defect D inside the inspection object W and a defect image DI in the captured image P corresponding to the defect D.

[0050] However, since the defect image DI in the captured image P is a two-dimensional projected image, only the captured image P obtained by irradiating the radiation R from the front direction of the bead 40 as shown in FIGS. 3 and 4 cannot obtain information in the depth direction of the defect D corresponding to the defect image DI, so the type of the defect D may not be determined in some cases.

[0051] For example, in a radiographic test of an object W containing a bead 40, typical defects D that may be detected include "poor penetration" and "cracking." JIS 3104-1995 defines "poor penetration" and "cracking" as different types of defects, one being "Type 2" and the other "Type 3." The impact of defect D on the weld strength and the repair methods for defect D differ depending on whether the defect is "Type 2" or "Type 3." Therefore, it is necessary to clearly determine which type of defect D it is. Furthermore, "Type 2" defects include elongated slag inclusions, pipes, poor penetration, poor fusion, and similar defects D. On the other hand, "Type 3" defects include cracking and similar defects D.

[0052] The defect type is a distinction based on the form of the defect D itself. For example, "poor penetration" refers to a situation where the actual penetration is insufficient compared to the design penetration, and is a defect D in which a gap remains in a part of the joint surface between the two base materials being welded. "Poor penetration" is a gap that remains in a part of the thickness direction of the base materials 42 and 44 between the joint surfaces, and its shape may be linear. For example, if the extension direction of such a linear defect D coincides with the extension direction of the bead 40, and the image is taken in the irradiation direction shown in Figures 3 and 4, the captured image P will also appear as a linear shadow, and will be captured as a linear defect image DI that is darker (i.e., has a higher density) than the surrounding area.

[0053] On the other hand, "cracks" are caused by stress and other factors, but as shown in Figure 3, they extend in the depth direction of the bead 40 (similar to the thickness direction of the base material 42, 44). When such a "crack" defect D is photographed in the irradiation direction shown in Figures 3 and 4, it appears as a linear defect image DI in the captured image P.

[0054] Thus, both "poor fusion" and "cracking" defects D can appear as linear defect images DI in the captured image P. Therefore, it can be difficult to distinguish whether the linear defect image DI in a single captured image P is a "poor fusion" defect D or a "cracking" defect D. This is because the captured image P is a two-dimensional projection image of radiation R transmitted through the object W being inspected, and defects D with three-dimensional shapes present within the object W are captured in the captured image P as two-dimensional defect images DI, where depth information is lost. Furthermore, distinguishing between the two becomes even more difficult when the size of the defect image DI is small, the area in which the defect image DI exists is narrow, or when there is little difference in the characteristics of multiple defect images DI being distinguished.

[0055] Figure 5 shows an example of a defect type, although it partially overlaps with the example of defect D described above. Figure 5 shows incomplete penetration (IP), lack of fusion (LF), blowhole (BH), pipe (P), slag inclusion (SL), crack (C), and tungsten inclusion (T). Other defect types include porosity (PO), and crack (C) is sometimes distinguished into longitudinal crack (LC) and transverse crack (TC). These defect types are quoted from the Japan Waterworks Steel Pipe Association website (https: / / www.wsp.gr.jp / qanda / taikei-d-3.html), and specific examples (images, etc.) showing the form of each defect type are described on that website.

[0056] Similar to "poor penetration" and "cracking," "poor fusion" and "slag inclusion" are also typical defects D detected in radiation testing of welds, as shown in bead 40 as an example, and it is difficult to distinguish between the defect images DI corresponding to these defects D.

[0057] "Slag inclusion" refers to a nonmetallic substance, mainly composed of oxides, sulfides, or flux, which is formed when various elements added to the welding material and shielding gas react. Under normal circumstances, it floats to the surface of the molten metal, but if it becomes incorporated into the weld metal, it results in defect D.

[0058] "Fusion failure" refers to a state where, for example, each base material 42 and 44 is in contact with the bead 40, and there are no gaps as seen in poor penetration, but the two are not melted together and integrated. Fusion failure can also occur between multiple segments within the bead 40. Figure 6 shows an example of "fusion failure" occurring between multiple segments within the bead 40. In Figure 6, multiple adjacent segments, indicated by the arrows in Figure 6, are in contact but not integrated.

[0059] The defects D for "fusing failure" and "slag inclusion" are less dense than the surrounding area within the object W being inspected. Therefore, as shown in Figure 4, in the image P taken with the radiation R irradiated in the direction of the front of the bead 40, the defect images DI corresponding to "fusing failure" and "slag inclusion" appear as high-density black images, and compared to "poor penetration" or "cracks," they appear as linear shapes with greater thickness (i.e., line width) or as elliptical shadows.

[0060] Because the shapes of "fusing failure" and "slag inclusion" are diverse, there are many patterns, and it is extremely difficult to distinguish between the two solely from the image P taken with the radiation R irradiated in the direction of the bead 40. Even expert inspectors often cannot make a determination.

[0061] Non-destructive testing includes various methods besides radiographic testing, such as ultrasonic testing (UT). By using multiple such tests in combination, it is possible to determine the type of defect D that is difficult to distinguish from the captured image P alone.

[0062] For example, in order to capture the surface characteristics of defect D and distinguish between different types of defects, it is effective to use ultrasonic testing (UT) in conjunction with the ultrasonic testing method. Since ultrasonic testing is a method that transmits pulses to a test object and evaluates the amplitude of the reflected wave, a strong signal is displayed when the test object has a surface perpendicular to the direction of ultrasonic propagation, indicating that the reflected signal from defect D is present.

[0063] "Poor penetration" is linear, while "cracks" have surface characteristics, making them distinguishable using ultrasonic testing.

[0064] However, it is not efficient to perform ultrasonic testing separately from radiographic testing using specialized equipment in order to identify defect type D.

[0065] Furthermore, by obtaining a three-dimensional image of the object W under inspection using tomographic imaging methods such as CT scans or tomosynthesis, depth information of the defect D (such as depth and position information in the depth direction) can be obtained, which is the most effective method for determining the type of defect D. For example, "fusing failure" occurs along the boundary of the bead 40 and has planar characteristics, while "slag inclusion" has strong volumetric characteristics, such as a deformed spherical shape. Three-dimensional images allow for the distinction between these two types of defects.

[0066] However, CT scans typically take more than an hour per subject and require specialized equipment and viewers. Furthermore, CT scanners are large and cumbersome devices, and their applicability is limited when performing 100% inspection on multiple objects W. Therefore, they are not typically used for non-destructive testing of welded parts such as weld beads 40.

[0067] Furthermore, among existing technologies, the parallax method using double exposure and the GUCHI (Geometric Unravel for Crack Height Image) method are known as methods for quantitatively determining the depth of defect D.

[0068] The parallax method involves shifting the position of the radiation source and irradiating the same test specimen from different positions, taking two images. The depth of the defect D is then calculated geometrically based on the amount of displacement of the defect image in the two resulting images. However, while the parallax method can geometrically determine the depth of the defect D, it cannot determine the type of defect D.

[0069] On the other hand, the GUCHI method, when the existence of defect D is known, is used for sizing purposes by placing a reference copper wire or similar (called a reference line) on the object surface, and acquiring two images by changing the position of the radiation source, similar to the parallax method, and then geometrically determining the depth of defect D from the two images. However, the GUCHI method requires the existence of defect D to be known and the physical placement of a reference line, so it cannot be used when the presence and location of defect D are unknown. Furthermore, it cannot determine the type of defect D.

[0070] In the parallax method and the GUCHI method, two images are acquired by shifting the position of the radiation source primarily through translation. Since the direction of irradiation does not change, it is not always possible to obtain information that can identify the type of defect D.

[0071] Therefore, in the radiation imaging system 10 according to this embodiment, the control unit 22 is capable of acquiring not only an image P taken with the radiation R irradiation direction facing the front of the bead 40, as shown in Figure 7(A), but also an image P taken with the radiation R irradiation direction facing oblique to the bead 40, as shown in Figure 7(B). Figure 7 is shown as an enlarged view centered on the bead 40 shown in Figures 3 and 4 to make the irradiation direction easier to understand. Here, for convenience, as shown in Figure 7(A), imaging with the radiation R irradiation direction facing the front of the bead 40 is also called front imaging, and as shown in Figure 7(B), imaging with the radiation R irradiation direction facing oblique to the bead 40 is also called inclined imaging.

[0072] In the front view shown in Figure 7(A), the direction of radiation R is along the reference direction K, which is the angle bisector of the central axis A of the beam. In contrast, in the inclined view shown in Figure 7(B), the direction of radiation R is in the direction in which the central axis A of the beam is inclined with respect to the reference direction K. Figures 8 and 9 show the state of Figure 7(B) more clearly. Figure 8 is a perspective view showing how the central axis A of the beam is inclined with respect to the reference direction K of the two base materials 42 and 44, while Figure 9 is a side view. In Figure 8, the symbol E indicates the extension direction of the bead 40. In inclined photography, radiation R is irradiated from an oblique angle onto the object W including the bead 40. By observing the defect image DI in the image P obtained by inclined photography as shown in Figure 7(B), it becomes possible to obtain, for example, depth information of the defect D, thus enabling the determination of defect types that are difficult to identify using only the image P obtained by front photography.

[0073] The image P obtained by the inclined imaging shown in Figure 7(B) is an image P in which the irradiation direction is a first direction inclined with respect to the reference direction K by an angle smaller than half the angle α between the two welded base materials 42 and 44, and is an example of the "first image" of this disclosure. More specifically, the first direction is the direction in which the central axis A of the beam of radiation R is inclined with respect to the reference direction K in a plane including the reference direction K. As described above, in the case of a butt joint shown in Figure 7(B), the angle α between the two base materials 42 and 44 is 180 degrees, and the first direction is a direction inclined with respect to the reference direction K by an angle θ smaller than 90 degrees, which is half of 180 degrees. In the example of Figure 7(B), the range of angle θ is greater than 0 degrees and less than 90 degrees, and preferably 20 degrees or more and 60 degrees or less.

[0074] Furthermore, the captured image P obtained by frontal photography shown in Figure 7(A) is, as described above, a captured image P in which the illumination direction is along the reference direction K corresponding to the angle bisector of the angle α. The captured image P obtained by frontal photography is an example of the "second captured image" of this disclosure, and the frontal direction is an example of the "second direction different from the first direction".

[0075] The determination of defect types will be explained with reference to Figure 7. For example, consider the case where a defect D in the bead 40 is a "crack" that extends in the depth direction. When a frontal view is taken of the bead 40 containing such a defect D, as shown in the upper part of Figure 7(A), an image P is obtained as shown in the lower part of Figure 7(A). In the image P of the lower part of Figure 7(A), the defect image DI loses the depth information of the defect D, resulting in an elongated linear image extending in the direction of the bead 40's extension. When an inclined view is taken of the same bead 40, as shown in the upper part of Figure 7(B), an image P is obtained as shown in the lower part of Figure 7(B). In the image P of the lower part of Figure 7(B), the depth information of the defect D extending in the depth direction of the bead 40 is captured. Specifically, in the frontal image P shown in Figure 7(A), the defect image DI is a long, narrow, linear shape. However, in the tilted image P shown in Figure 7(B), the linear defect image DI appears relatively wider due to the inclusion of shadows in the depth direction. In reality, the depth of the crack varies from place to place, so even a wide, linear defect image DI is not uniform in width; for example, the line may gradually become thinner in one direction.

[0076] If defect D is not a "crack" but a "poor penetration," the three-dimensional shape is likely to be a cylindrical gap with roughly the same thickness in the depth and width directions of the bead 40. Therefore, in both the front view in Figure 7(A) and the inclined view in Figure 7(B), a linear defect image DI with little change in thickness is captured in both captured images P. Thus, by referring to two captured images P taken in two different irradiation directions—a first direction which is oblique to the bead 40 and a second direction which is the front view of the bead 40—it is sometimes possible to determine the type of defect D in the object W being inspected that corresponds to the defect image DI captured in those images.

[0077] Furthermore, the image P obtained by the inclined photography in Figure 7(B) (an example of the first image) can also be rephrased as follows: The first image is an image obtained by irradiating the bead 40 with radiation R from a first direction that is perpendicular to the direction in which the bead 40 (an example of a weld) formed by welding the base materials 42 and 44 extends (for example, E shown in Figure 8, etc.), and where the angle between the irradiation direction from which radiation R is irradiated and the weld surface on which the bead 40 of one of the base materials is located is an angle smaller than half the angle between the two welded base materials 42 and 44 (for example, the clamping angle α) (for example, the angle θ).

[0078] In the examples shown in Figures 3 to 9, the clamping angle α between the two base materials 42 and 44 is 180 degrees, and the surfaces of the base materials 42 and 44 are on the same plane, but as shown in Figures 10 and 11, there are also inspection objects W where the surfaces of the two base materials 42 and 44 sandwiching the bead 40 are not on the same plane. This method of joining two base materials 42 and 44 is also called fillet welding. In the inspection object W shown in Figures 10 and 11, as an example, the orientation of the surfaces of the two base materials 42 and 44 differs by 90 degrees, and the overall shape is L-shaped. In this case, the clamping angle α is naturally 90 degrees. The reference direction K is 45 degrees, which is half the angle of the clamping angle α.

[0079] In the inspection object W shown in Figures 10 and 11, the irradiation direction of radiation R in frontal imaging is such that the central axis A of the beam is aligned with the reference direction K, resulting in a 45-degree direction. The 45-degree direction is an example of the "second direction" in this disclosure. In contrast, the irradiation direction of radiation R in oblique imaging is such that it is oblique to the reference direction K by an angle θ smaller than 45 degrees, which is half of the angle α, and this direction is an example of the "first direction" in this disclosure. In the inspection object W shown in Figures 10 and 11, as in the examples in Figures 3 to 9, the type of defect D can sometimes be determined by referring to the image P obtained by such oblique imaging.

[0080] In this way, the control unit 22, which is an example of a processor, acquires two images P: an image P taken by tilting (an example of a first image) and an image P taken by frontal photography (an example of a second image), and determines the type of defect D based on the two acquired images P. This improves the accuracy of determining the type of defect D.

[0081] The control unit 22 may also determine the type of defect D by acquiring only the image P captured by tilted photography. In some cases, the accuracy of determining the type of defect D can be improved even with only the image P captured by tilted photography. Of course, as shown in the above embodiment, by using the image P captured by frontal photography in addition to the image P captured by tilted photography, it becomes possible to capture the defect image DI in three dimensions, and the control unit 22 can further improve the accuracy of determining the type of defect D.

[0082] Furthermore, in the above embodiment, the control unit 22 acquires two images, an image P taken by tilt photography and an image P taken by front photography, to determine the type of defect D. However, instead of or in addition to the image P taken by front photography, two or more images P taken by tilt photography (an example of the first image) may be acquired. An image taken by tilt photography P is an image in which the irradiation direction of radiation R is the first direction tilted by an angle θ smaller than half of the angle α with respect to the reference direction K corresponding to the bisector of the angle α. As shown in Figures 3 to 7, if the angle α is 180 degrees, one of the two images P taken by tilt photography may be an image P with an angle θ of 20 degrees, and the other may be an image P with an angle θ of 40 degrees. Even in this case, it is possible to obtain depth information of the defect D.

[0083] Furthermore, as shown in Figure 12, the control unit 22 may acquire two captured images P, each with two illumination directions that are symmetrical with respect to the reference direction K corresponding to the angle bisector of the angle α. One illumination direction is one in which the central axis A of the wire bundle is inclined counterclockwise with respect to the reference direction K by an angle θ smaller than half the angle α, and is an example of the "first direction" in this disclosure. In contrast, the other illumination direction is one in which the central axis A of the wire bundle is inclined clockwise with respect to the reference direction K by an angle θ smaller than half the angle α. This illumination direction is an example of the "second direction symmetrical to the first direction" with respect to the angle bisector of the angle α. That is, the two illumination directions are symmetrical with respect to the reference direction K as the axis of symmetry. The control unit 22 may determine the type of defect D based on the two captured images P taken in these two illumination directions. Furthermore, since the direction of illumination is often confirmed visually, the concept of "symmetry" includes not only strict symmetry but also approximate symmetry, and an error of about plus or minus 10 degrees is acceptable.

[0084] In this embodiment, the control unit 22 performs processing to display on the display unit 32 an image P captured with the first direction as the illumination direction (an example of a first image) and an image P captured with the second direction as the illumination direction (an example of a second image). As a result, the control unit 22 can present the two acquired images P to the user.

[0085] Furthermore, the control unit 22 may perform a process to identify the location of defects D contained in the bead 40 (an example of a weld) based on the acquired captured image P (an example of a first captured image). For example, the control unit 22 identifies the location of defects D by performing image analysis on the captured image P. For the image analysis, rule-based image analysis methods such as pattern matching may be used, or machine learning analysis methods such as semantic segmentation may be used. The control unit 22 displays the location of the defect D corresponding to the identified defect D in an identifiable manner by adding annotations to the location of the defect image DI corresponding to the defect D. The location of the defect D can be identified by the user visually checking the defect image DI in the captured image P and adding annotations to the location of the defect image DI corresponding to the defect D through user operation, or it can be done by the control unit 22.

[0086] Furthermore, the control unit 22 may perform a process to identify at least one of the type and category of the defect D based on the acquired captured image P (an example of a first captured image).

[0087] Specifically, the applicable type is identified from the following four types (Type 1 to Type 4). Here, Types 1 to Type 4 are examples of defect classifications in JIS standards and are also called types of defects. The "type" of defect in this disclosure is a concept that includes the sub-classification of such defect types. Therefore, identifying the "type" of defect includes identifying sub-classifications such as "failure to fuse" or "crack," as well as identifying sub-classifications such as Type 1 or Type 2. Type 1 (Circular): "Blowhole," "Porosity," "Low-density foreign matter" Type 2 (Linear): "Failure to fuse," "Slag inclusion," "Pipe," "Failure to penetrate," "Undercut" Type 3 (Crack): "Crack (Longitudinal crack, Transverse crack)" Type 4 (High-density defect): "Tungsten inclusion," "Spatter," "High-density foreign matter"

[0088] In determining the type of defect D, if the control unit 22 cannot identify a single type due to, for example, insufficient information to identify the type, it may present multiple candidate types in order of likelihood. Alternatively, the control unit 22 may present multiple candidate types in order of least likelihood. The multiple types are presented to the user by being displayed on the display unit 32. This allows the user, for example, to narrow down the candidates from the presented list of types to one while referring to the defect image DI of the captured image P.

[0089] Furthermore, the control unit 22 may perform a process to estimate the volume of the identified defect D when determining the severity of the defect D. The volume estimation process is performed, for example, by estimating the three-dimensional size of the defect image DI based on the defect image DI contained in the frontal image P (an example of a second image) and the tilted image P (an example of a first image).

[0090] When estimating the volume of defect D, the control unit 22 may determine from the volume estimation result whether or not the volume exceeds a predetermined judgment criterion that has been entered in advance. Alternatively, the control unit 22 may determine the severity of defect D from the volume estimation result. The severity is assigned, for example, by an evaluation value such as a score or ranking based on the volume estimation result. This makes it possible to inform the user of the severity of defect D.

[0091] The control unit 22 may estimate at least one of the type, severity, depth, and volume of a defect D if a defect image DI corresponding to a defect D in the bead 40 (an example of a welded part) is included in the image P (an example of a first image) taken by tilt photography, using the lengths of one or more line segments that cross the defect image DI.

[0092] For example, as shown in Figure 13, the control unit 22 may use the lengths of two or more line segments that cross the defect image DI and determine the type of defect D based on the lengths of these line segments. In the example in Figure 13, two values ​​(y2 - y1) and (y4 - y3) are used as length information.

[0093] Here, we will explain a specific method by which the control unit 22 identifies the type of defect D based on information about the lengths of two or more line segments.

[0094] The control unit 22 derives the lengths of two or more parallel line segments that cross the defect image DI, and compares the information of the lengths of the two or more derived line segments.

[0095] The control unit 22 then determines, based on the comparison results, that the variation in the lengths of two or more line segments crossing the defect image DI (for example, the difference in absolute values, the rate of variation relative to the average, etc.) is less than a predetermined range, that the defect D corresponding to the defect image DI is a defect D with uniform depth, and determines the type of the defect D to be, for example, poor penetration. For example, the defect image DI included in the upper captured image P of Figure 14 (an example of a first captured image) is an example where the difference in the lengths l and m of the two line segments crossing the defect image DI is less than a certain range.

[0096] On the other hand, if the variation in the lengths of two or more line segments crossing the defect image DI exceeds a predetermined range, the control unit 22 determines that the defect D corresponding to the defect image DI is a defect D that lacks uniformity in the depth direction, and determines that the type of defect D is, for example, a crack. For example, the defect image DI included in the lower captured image P of Figure 14 (an example of a first captured image) is an example where the difference in the lengths l and m of two line segments crossing the defect image DI exceeds a certain range.

[0097] Furthermore, the control unit 22 may use the length of one or more line segments that cross such defect image DI to estimate the depth of the defect D. For example, Figure 15 shows an example in which two images P are obtained from a single bead 40 by performing frontal and oblique imaging, with different irradiation directions of radiation R. The defect image DI corresponding to a single defect D is included in both the image P obtained by oblique imaging (an example of the first image) and the image P obtained by frontal imaging (an example of the second image). The control unit 22 can, for example, determine the width h2 from the length of the line segment that crosses the defect image DI in the image P obtained by frontal imaging, and on the other hand, determine the width h1 from the length of the line segment that crosses the defect image DI in the image P obtained by oblique imaging. Then, the control unit 22 can estimate the length n in the depth direction of the defect D from the geometric relationship between the widths h2 and h1 and the respective irradiation directions of frontal and oblique imaging. For example, if the angle of inclination of the illumination direction for inclined imaging relative to the illumination direction for frontal imaging is known, the control unit 22 may be able to estimate the length n in the depth direction of the defect D using trigonometric functions based on the widths h1 and h2 and the inclination angle.

[0098] The control unit 22 can estimate the volume of the defect D by multiplying the estimated length n in the depth direction by the lengths of line segments representing the size of the defect image DI, such as the widths h1 and h2 of the defect image DI, and summing these results over the entire region where the defect image DI occurs.

[0099] Furthermore, as shown in Figures 16 and 17, if multiple defect images DI corresponding to multiple defects D contained in the bead 40 are included in each of two captured images P (first captured image and second captured image) with different irradiation directions, the control unit 22 may identify the types of the multiple defects D based on the positional relationship of the multiple defect images DI in each captured image P.

[0100] As shown in Figure 17, consider the case where multiple defects D are present at different locations within the bead 40. The control unit 22 acquires two images P of this bead 40, one taken from the front and one taken at an angle. The image P shown in Figure 16 is an example of the first image obtained by the angled photography.

[0101] The two defects D shown in Figure 17 are located at different positions in the depth direction z, but their positions in the plane perpendicular to the depth direction z overlap. Therefore, although not shown in the illustration, in the image P obtained by frontal imaging (an example of a second image), the positions of the two defect images DI overlap. On the other hand, in the image P obtained by tilt imaging, shown as an example in Figure 16, the two defect images DI corresponding to the two defects D are captured at different positions with a gap between them, according to the difference in their positions in the depth direction z. Thus, the positional relationship of the two defect images DI corresponding to the two defects D differs between the image P obtained by frontal imaging and the image P obtained by tilt imaging. Here, the positional relationship of the two defect images DI in the image P obtained by tilt imaging is an example of the "first positional relationship" of this disclosure, and the positional relationship of the two defect images DI in the image P obtained by frontal imaging is an example of the "second positional relationship" of this disclosure.

[0102] The control unit 22 can determine which of the two defects D is located deeper based on two positional relationships (first positional relationship and second positional relationship) of the two defect images DI contained in the two captured images P with different irradiation directions. The control unit 22 can also determine the shape of a defect D by comparing the defect image DI contained in each of the two captured images P for a single defect D. For example, the control unit 22 can determine that one defect is linear and the other is point-shaped based on the two defect images DI contained in the captured image P shown in Figure 16, which was obtained by tilt imaging. In this way, the control unit 22 can estimate the location and shape of the defect D from the two positional relationships described above, and in some cases can estimate the type of each defect D from these estimation results.

[0103] Furthermore, the control unit 22 may receive prior information regarding the welding method and use the received prior information to identify the type of defect D. For example, if there is prior information that it is a submerged welding, it can be determined that penetration defects will not occur because submerged welding is a welding method that uses a large current. As a result, the control unit 22 can exclude penetration defects from the options when identifying the type of defect D.

[0104] The control unit 22 displays the determination result of the type of defect D identified as described above on the display unit 32. When displaying the determination result on the display unit 32, it may be visually displayed on the screen of the display unit 32 in correspondence with the location of defect D. Alternatively, the information may be output as a report. Furthermore, the basis for the determination of the type of defect D may also be output.

[0105] Next, we will describe the specific processing performed by the control unit 22 of the control device 18 in the radiation imaging system 10 configured as described above. Figure 18 is a flowchart showing an example of the processing flow performed by the control unit 22 of the control device 18 in the radiation imaging system 10 according to this embodiment. The processing in Figure 18 is started when the user instructs the execution of the program 30 stored in the ROM 26.

[0106] In step 100, the CPU 24 acquires an image P (an example of a second image) captured by irradiating with radiation R from a reference direction K from the radiation detector 20 and proceeds to step 102.

[0107] In step 102, the CPU 24 acquires an image P (an example of a first image) from the radiation detector 20, which is an image taken by irradiating with radiation R from an oblique direction (an example of a first direction) tilted with respect to the reference direction K, and proceeds to step 104.

[0108] In step 104, the CPU 24 identifies the location of the defect D from the two acquired captured images P and proceeds to step 106.

[0109] In step 106, the CPU 24 determines whether or not defect D exists. If the determination is negative, the process proceeds to step 108; if it is positive, the process proceeds to step 110.

[0110] In step 108, the CPU 24 displays on the display unit 32 that there are no defects D, and the series of processes ends.

[0111] Meanwhile, in step 110, the CPU 24 performs defect type identification processing and proceeds to step 112. In the defect type identification processing, the location and type of defect D are identified using the two captured images P acquired in steps 100 and 102, and the type of defect D is identified. Alternatively, the location or type of defect D may be identified.

[0112] In step 112, the CPU 24 displays the result of identifying defect D on the display unit 32 and terminates the series of processes. Alternatively, during the defect type identification process, the volume of defect D may be estimated, and a score or rank representing the severity of defect D may be assigned and further displayed on the display unit 32.

[0113] Thus, in this embodiment, the control unit 22 acquires an image P (an example of a first image) obtained by irradiating the bead 40 of the object to be inspected W with radiation R from an oblique direction. This makes it possible to improve the accuracy of determining the type of defect D.

[0114] Furthermore, the control unit 22 can further improve the accuracy of determining the type of defect D by acquiring two or more captured images P with different radiation irradiation directions.

[0115] In the process shown in Figure 18 of the above embodiment, the control unit 22 is shown to acquire two images P. However, by omitting step 100, only the images P obtained by inclined imaging with radiation R irradiated from an oblique direction may be acquired to identify at least one of the location and type of defect D. Alternatively, the control unit 22 may acquire multiple images P obtained by inclined imaging with different irradiation directions (an example of the first images) in addition to the images obtained by frontal imaging with radiation R irradiated from a reference direction K (an example of the second images), to identify at least one of the location and type of defect D.

[0116] The following further notes are disclosed regarding the embodiments described above. (Note 1) An image processing device comprising a processor, wherein the processor acquires a first image obtained by irradiating the welded portion of two welded base materials with radiation from a first direction, the first direction being a direction inclined with respect to the angle bisector of the angle formed by the welded surfaces of the two base materials by an angle smaller than half of the angle formed by the angle. (Note 2) The image processing device according to Note 1, wherein the processor further acquires a second image obtained by irradiating the welded portion with radiation from a second direction different from the first direction. (Note 3) The image processing device according to Note 2, wherein the processor acquires an image as the second image, in which the second direction is a direction along the angle bisector of the angle. (Note 4) The image processing device according to Note 2, wherein the processor acquires an image as the second image, in which the second direction is a direction symmetric to the first direction with respect to the angle bisector of the angle. (Note 5) The image processing apparatus according to any one of Notes 1 to 4, wherein the processor identifies at least one of the location and type of defect in the weld based on the first captured image. (Note 6) The image processing apparatus according to Note 5, wherein if the type cannot be identified as a single type, the processor presents a plurality of types in order of likelihood or likelihood. (Note 7) The image processing apparatus according to any one of Notes 2 to 4, wherein the processor performs processing for displaying the first captured image and the second captured image on a display unit. (Note 8) The image processing apparatus according to any one of Notes 2 to 7, wherein the processor estimates the volume of the defect in the weld using the first captured image and the second captured image, and determines the severity of the defect using the estimated volume. (Note 9) The image processing apparatus according to any one of Notes 5 to 8, wherein the processor estimates at least one of the type, severity, depth, and volume of the defect using the length of one or more line segments that cross the defect image when a defect image corresponding to the defect in the weld is included in the first captured image.(Note 10) The image processing apparatus according to any one of Notes 5 to 9, wherein the processor further acquires a second image obtained by irradiating the weld from a second direction different from the first direction, and when a plurality of defect images corresponding to a plurality of defects contained in the weld are included in the first image and the second image, the processor identifies at least one of the location and type of the defect based on a first positional relationship of the plurality of defect images in the first image and a second positional relationship of the plurality of defect images in the second image. (Note 11) An image processing method in which a computer performs the process of acquiring a first image obtained by irradiating the weld of two welded base materials from a first direction, wherein the first direction is a direction inclined with respect to the bisector of the angle between the weld surfaces of the two base materials by an angle smaller than half of the angle between the weld surfaces of the two base materials. (Note 12) An image processing program that causes a computer to perform the process of acquiring a first image obtained by irradiating the welded joint of two welded base materials with radiation from a first direction, wherein the first direction is a direction inclined with respect to the bisector of the clamping angle by an angle smaller than half the clamping angle formed by the welded surfaces of the two base materials.

[0117] In the above embodiment, the processing performed by the control unit 22, which is an example of a processor, is performed on any computer. Furthermore, any computer may perform these processes using a processor as hardware, a program as software, or a combination thereof. In that case, the processor is configured to work in cooperation with the program to perform the various processes in this embodiment, and can function as each unit or means in this embodiment. Also, the execution order of the processes by the processor is not limited to the order described and may be changed as appropriate.

[0118] Any computer may be a general-purpose computer, a computer designed for a specific purpose, a workstation, or any other system capable of performing each process. A processor may consist of one or more pieces of hardware, and the type of hardware is not limited. For example, a processor may consist of programmable logic devices such as a CPU (Central Processing Unit), MPU (Micro Processing Unit), FPGA (Field Programmable Gate Array), dedicated circuits for performing specific processes such as an ASIC (Application Specific Integrated Circuit), a GPU (Graphic Processing Unit), or an NPU (Neural Processing Unit). Furthermore, the type of hardware may be a combination of different types of hardware. When multiple hardware components are configured to execute one or more processes of a processor, these components may reside in physically separate devices or in the same device. Furthermore, in any embodiment, the order of the processes performed by the processor is not limited to the order described above and may be changed as appropriate. The hardware components are composed of electrical circuits (circuits) and the like, which are combinations of circuit elements such as semiconductor elements.

[0119] Furthermore, the program may be firmware or software such as microcode. Alternatively, the program may be, for example, a set of program modules, each function of which may be implemented by a processor configured to perform its respective function. The program may be program code or multiple code segments stored on one or more non-temporary computer-readable media (e.g., storage media and other storage). The program may be divided and stored on multiple non-temporary computer-readable media located in physically separate devices. Program code or code segments may represent any combination of procedures, functions, subprograms, routines, subroutines, modules, software packages, classes, or instructions, data structures, or program statements. Program code or code segments may be connected to other code segments or hardware circuits by sending and receiving information, data, arguments, parameters, or memory contents.

[0120] The technology of this disclosure can be appropriately combined with the various embodiments and / or variations described above. Furthermore, it is understood that various configurations can be adopted without departing from the gist of the invention, and the invention is not limited to the embodiments described above. In addition, the technology of this disclosure extends to storage media for storing programs non-temporarily. The storage media are computer-readable non-temporarily storage media such as USB (Universal Serial Bus) memory, flexible disks, and CD-ROMs (Compact Disc Read Only Memory). Programs may also be provided online via a network such as the Internet. Furthermore, the technology of this disclosure extends to program products in addition to programs. Program products include all forms of products for providing programs. Like programs, program products may be stored and provided on computer-readable non-temporarily storage media, or they may be provided online.

[0121] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[0122] In this specification, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0123] The disclosures of Japanese Patent Application No. 2025-043986, filed on 18 March 2025, and Japanese Patent Application No. 2026-027490, filed on 24 February 2026, are incorporated herein by reference in their entirety. Furthermore, all documents, patent applications, and technical standards described herein are incorporated herein by reference to the same extent as if the incorporation of each individual document, patent application, and technical standard were specifically and individually noted.

Claims

1. An image processing device comprising a processor, wherein the processor acquires a first image obtained by irradiating the welded portion of two welded base materials with radiation from a first direction, the first direction being a direction inclined with respect to the bisector of the angle between the welded surfaces of the two base materials by an angle smaller than half of the angle between the welded surfaces.

2. The image processing apparatus according to claim 1, wherein the processor further acquires a second image obtained by irradiating the welded portion with radiation from a second direction different from the first direction.

3. The image processing apparatus according to claim 2, wherein the processor acquires an image as the second captured image, the second direction being the direction along the angle bisector of the angle between the two sides.

4. The image processing apparatus according to claim 2, wherein the processor acquires an image as the second captured image, in which the second direction is symmetric to the first direction with respect to the angle bisector.

5. The image processing apparatus according to claim 1, wherein the processor identifies at least one of the location and type of defects in the weld based on the first captured image.

6. The image processing apparatus according to claim 5, wherein, if the processor cannot identify one of the types, it presents a plurality of the types in order of likelihood or likelihood.

7. The image processing apparatus according to claim 2, wherein the processor performs processing for displaying the first captured image and the second captured image on a display unit.

8. The image processing apparatus according to claim 2, wherein the processor estimates the volume of a defect in the weld using the first captured image and the second captured image, and determines the severity of the defect using the estimated volume.

9. The image processing apparatus according to claim 5, wherein the processor estimates at least one of the type, severity, depth, and volume of a defect using the lengths of one or more line segments crossing the defect image when a defect image corresponding to a defect in the weld is included in the first captured image.

10. The image processing apparatus according to claim 5, wherein the processor further acquires a second image obtained by irradiating the weld from a second direction different from the first direction, and when a plurality of defect images corresponding to a plurality of defects contained in the weld are included in the first image and the second image, the processor identifies at least one of the location and type of the defect based on a first positional relationship of the plurality of defect images in the first image and a second positional relationship of the plurality of defect images in the second image.

11. An image processing method in which a computer performs the process of acquiring a first image obtained by irradiating the welded joint of two welded base materials with radiation from a first direction, wherein the first direction is a direction inclined with respect to the bisector of the clamping angle by an angle smaller than half the clamping angle formed by the welded surfaces of the two base materials.

12. An image processing program for causing a computer to perform the process of acquiring a first image obtained by irradiating the welded joint of two welded base materials with radiation from a first direction, wherein the first direction is a direction inclined with respect to the bisector of the angle between the welded surfaces of the two base materials by an angle smaller than half of the angle between the welded surfaces.

Images