Surface inspection device and surface inspection method

Abstract

In a surface inspection device (10), a scar identification unit identifies, for each inspection image obtained on the basis of a captured image, a scar reflected in the inspection image; a second direction addition processing unit that adds pixel values of the scar determined by the scar determination unit in a direction corresponding to a second direction orthogonal to the first direction in the inspection images to the plurality of inspection images in a direction corresponding to the second direction; a first direction addition processing unit that adds pixel values of the scar determined by the scar determination unit in a direction corresponding to the first direction in the inspection images with respect to the plurality of inspection images in the direction corresponding to the first direction; the continuous scar determination unit determines whether an occasionally generated single scar or a continuously generated continuous scar has occurred on the surface of the object on the basis of the addition result of the second direction addition processing unit and the addition result of the first direction addition processing unit.

Description

Technical Field

[0001] The present disclosure relates to a surface inspection device and a surface inspection method, and more particularly, to a surface inspection device and a surface inspection method for inspecting the surface properties of an object using an image. Background Art

[0002] Conventionally, magnetic particle inspection is known (for example, refer to Japanese Patent Application Laid-Open Nos. 2011-174893 and 2021-025878). Summary of the Invention

[0003] Problems to be solved by the invention

[0004] In magnetic particle inspection, the object being inspected is magnetized and then coated with a magnetic powder liquid. In the magnetized object, magnetic flux leaks from open flaws, and the magnetic powder in the applied magnetic powder liquid adheres to the flaws. Therefore, when ultraviolet light is irradiated on the object coated with the magnetic powder liquid, the attached magnetic powder is stimulated to emit light, and the luminescent portion appears as a flaw pattern. Inspection of the surface properties of the object (magnetic particle inspection) is performed, for example, by visually observing these flaw patterns in a darkroom. The flaws that can be identified during magnetic particle inspection vary in shape and pattern, and some may even resemble harmless, suspicious patterns caused by, for example, accumulation of magnetic powder liquid.

[0005] Inspectors must distinguish between flaws and suspected flaws, identifying them as harmful or harmless. However, this distinction is extremely difficult and relies heavily on the inspector's skills. Furthermore, magnetic particle inspections require long hours in a darkroom, further complicating the inspection process. Furthermore, complex object shapes further complicate inspection. Consequently, during magnetic particle inspections conducted by human inspectors, there are cases where surface flaws are overlooked and the object leaks, leading to significant complaints. Therefore, there is a need to automate magnetic particle inspections.

[0006] Magnetic particle inspection is performed by projecting uniform ultraviolet light onto the surface of an object coated with a magnetic powder liquid and visually observing or photographing the patterns produced by the excitation luminescence from the magnetic powder attached to the flaws with a camera. Various methods have been proposed for photographing objects. For example, methods for photographing objects with complex shapes including irregularities include photographing the object while it is rotating and connecting the captured images circumferentially, and methods for obtaining one-dimensional or two-dimensional images of the object while it is rotating or being transported.

[0007] For example, Japanese Patent Application Laid-Open No. 2011-174893 proposes a method for varying the diameter of the attached magnetic powder according to the size of each flaw by mixing magnetic powders of varying excitation emission colors for each diameter. This method allows the color of the excitation emission to vary depending on the size of the flaw, making it possible to detect the flaw size (i.e., flaw depth, etc.) using a color camera.

[0008] As examples of scratches on the surface of an object, there are single scratches that occur sporadically on the object, and continuous scratches that occur continuously on multiple objects due to abnormalities in the equipment. If no countermeasures are taken, continuous scratches may continue to occur, so in order to suppress the continuous production of objects with continuous scratches, early detection of continuous scratches is also required. Here, for example, as examples of continuous scratches on steel plates, there are examples of scratches that occur when foreign matter adheres to the rolling rollers and the shape of the foreign matter is transferred to the product during rolling. Such continuous scratches occur continuously in the order of manufacturing the steel plates (manufacturing order), so it is easy to grasp the continuity during inspection. Therefore, it is relatively easy to detect continuous scratches on steel plates at an early stage.

[0009] However, for one-piece molded products, such as forged products, the subsequent processes are performed in a different order from the manufacturing sequence (molding sequence). This order differs from the manufacturing sequence because, for example, a last-in, first-out (LIFO) process is used in various processes, such as during line output and loading onto transport carts. Single flaws are inherently discontinuous, so the order of transport is not an issue. However, for continuous flaws, the subsequent processes are performed in a different order from the molding sequence, thus losing information about the continuity of the continuous flaw and making it difficult to detect it early.

[0010] Therefore, an object of the present disclosure is to provide a surface inspection device and a surface inspection method that can detect continuous flaws at an early stage.

[0011] Means for solving problems

[0012] The first method disclosed in the present invention is a surface inspection device for inspecting the properties of the surface of an object, the surface inspection device comprising: a magnetization unit for magnetizing the object; a magnetic powder attachment unit for applying magnetic powder liquid to the magnetized object to attach magnetic powder to the object; an illumination unit for irradiating ultraviolet light to the object to which the magnetic powder is attached; an imaging unit for photographing a plurality of positions along a first direction of the object to which the magnetic powder that emits light by ultraviolet light is attached, thereby generating a plurality of camera images; and an inspection unit for inspecting the properties of the surface of the object based on the camera images, the inspection unit comprising: a flaw determination unit for determining the flaws reflected in the inspection image for each inspection image obtained based on the camera image; a second direction An addition processing unit, for adding, for a plurality of the inspection images, pixel values ​​of the flaw determined by the flaw determination unit in a direction corresponding to a second direction orthogonal to the first direction in the inspection image, in the direction corresponding to the second direction; a first direction addition processing unit, for adding, for a plurality of the inspection images, pixel values ​​of the flaw determined by the flaw determination unit in a direction corresponding to the first direction in the inspection image, in the direction corresponding to the first direction; and a continuous flaw determination unit, for determining, based on the addition result of the second direction addition processing unit and the addition result of the first direction addition processing unit, whether a single flaw generated sporadically or a continuous flaw generated continuously has occurred on the surface of the object.

[0013] Regarding the second embodiment of the present disclosure, in the surface inspection device of the first embodiment, the second direction addition processing unit adds the size of the flaw determined by the flaw determination unit in the direction corresponding to the second direction in the inspection image after setting it to a prescribed size, i.e., the flaw width length, and the first direction addition processing unit adds the size of the flaw determined by the flaw determination unit in the direction corresponding to the first direction in the inspection image after setting it to a prescribed size, i.e., the flaw circumference.

[0014] Regarding the third embodiment of the present disclosure, in the surface inspection device of the first embodiment or the second embodiment, there is a continuous scar position output unit. When the continuous scar determination unit determines that a continuous scar has occurred, the continuous scar position output unit determines the position of the continuous scar and outputs the position of the continuous scar based on the position of the scar determined by the scar determination unit, the addition result of the second direction addition processing unit, and the addition result of the first direction addition processing unit.

[0015] Regarding the fourth embodiment of the present disclosure, in a surface inspection device of any one of the first to third embodiments, there is a rotating portion that rotates the object around a rotation axis parallel to the second direction, and while the object is rotated by the rotating portion, the camera portion photographs multiple positions of the object along the first direction, thereby generating the multiple camera images.

[0016] The fifth mode of the present disclosure is a surface inspection method for inspecting the properties of the surface of an object, the surface inspection method uses a surface inspection device, and the surface inspection device comprises: a magnetization unit for magnetizing the object; a magnetic powder attachment unit for applying magnetic powder liquid to the magnetized object to attach magnetic powder to the object; an illumination unit for irradiating ultraviolet light to the object to which the magnetic powder is attached; an imaging unit for capturing multiple positions along a first direction of the object to which magnetic powder that emits light excited by ultraviolet light is attached, thereby generating multiple camera images; and an inspection unit for inspecting the properties of the surface of the object based on the camera images. The surface inspection method uses the inspection unit and has the following steps: a flaw determination step for determining the inspection image for each inspection image obtained based on the camera image. a second direction addition processing step, for a plurality of the inspection images, adding the pixel values ​​of the direction of the flaw determined by the flaw determination unit corresponding to the second direction orthogonal to the first direction in the inspection image in the direction corresponding to the second direction; a first direction addition processing step, for a plurality of the inspection images, adding the pixel values ​​of the direction of the flaw determined by the flaw determination unit corresponding to the first direction in the inspection image in the direction corresponding to the first direction; and a continuous flaw determination step, based on the addition result in the second direction addition processing step and the addition result in the first direction addition processing step, determining whether a single flaw generated sporadically or a continuous flaw generated continuously has occurred on the surface of the object.

[0017] According to the present disclosure, a surface inspection device and a surface inspection method capable of early detection of continuous flaws are provided. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a diagram showing an example of a schematic configuration of a surface inspection apparatus and an object to be inspected for surface properties according to an embodiment of the present disclosure.

[0019] Figure 2 This is a block diagram showing an example of the functional structure of the processing device.

[0020] Figure 3 This is a diagram showing an example of generating an inspection image from a plurality of captured images.

[0021] Figure 4 FIG. 1 is a diagram showing an example of inspection images obtained for a plurality of objects.

[0022] Figure 5 FIG. 1 is a diagram showing an example of inspection images obtained for a plurality of objects.

[0023] Figure 6 This is a diagram showing an example of a plurality of inspection images.

[0024] Figure 7 FIG. 1 is a diagram showing an example of a plurality of inspection images and a first flaw image.

[0025] Figure 8 FIG. 1 is a diagram showing an example of a plurality of inspection images and a second flaw image.

[0026] Figure 9 FIG. 1 is a diagram showing an example of a plurality of inspection images and a first flaw image.

[0027] Figure 10 FIG. 1 is a diagram showing an example of a plurality of inspection images and a second flaw image.

[0028] Figure 11 FIG. 1 is a diagram showing an example of a plurality of inspection images and a first flaw image.

[0029] Figure 12 FIG. 1 is a diagram showing an example of a plurality of inspection images and a second flaw image.

[0030] Figure 13 This is a diagram showing an example of the positions of continuous flaws identified based on a plurality of inspection images.

[0031] Figure 14 This is a block diagram showing an example of the hardware configuration of the processing device.

[0032] Figure 15 This is a flowchart showing an example of the flow of the surface inspection process.

[0033] Figure 16 It is a diagram showing a modified example of a case where the inspection image is corrected.

[0034] Figure 17 It is a diagram showing a modified example of a case where the inspection image is corrected. DETAILED DESCRIPTION

[0035] Hereinafter, one embodiment of the present disclosure will be described with reference to the drawings.

[0036] (Explanation of magnetic particle inspection)

[0037] In one embodiment of the present disclosure, a surface inspection device for performing magnetic particle inspection on the surface properties of an object to be inspected is described. Magnetic particle inspection is performed to inspect fine flaws (hereinafter referred to as "magnetic particle flaws") that are difficult to visually detect among surface defects of the object.

[0038] (Description of the Structure of Surface Inspection Device 10)

[0039] Figure 1 FIG. 1 shows an example of a schematic structure of a surface inspection device 10 according to an embodiment of the present disclosure and an object 80 as an inspection target for surface properties. The object 80 is a ferromagnetic object. The object 80 is, for example, an integrally formed product including a forged product. In the following description, Figure 1 As shown, the object 80 is shaped like a quadrangular prism and the surface inspection of the outer peripheral surface of the object 80 is performed while rotating the object 80. However, the object 80 may be in any shape and does not need to be rotated.

[0040] The surface inspection device 10 is an inspection device for inspecting the properties of the surface of an object 80. The surface inspection device 10 includes a magnetizing device 12, a magnetic powder adhering device 14, a rotating device 16, an illuminating device 18, an imaging device 20, a marking device 22, and a processing device 24.

[0041] The magnetizing device 12 is a device for magnetizing the target object 80, which is a ferromagnetic body. The magnetizing device 12 is an example of a magnetizing unit of the present disclosure.

[0042] The magnetic powder adhering device 14 is a device that applies magnetic powder liquid to the object 80 magnetized by the magnetizing device 12 to adhere magnetic powder to the object 80. The magnetic powder adhering device 14 is an example of a magnetic powder adhering unit of the present disclosure.

[0043] The rotating device 16 is a mechanism that rotates the object 80 around a rotation axis parallel to a second direction described later. More specifically, the rotating device 16 is a mechanism that rotates the object 80 around a rotation axis parallel to a direction perpendicular to the optical axis relative to the optical axis of the camera device 20 described later, so as to observe the outer peripheral surface of the inspection object, which is the object 80. The rotating device 16 is capable of rotating the object 80 at a desired speed by having a speed reducer and a motor. In this embodiment, the rotating device 16 rotates the object 80 around an axis of the quadrangular prism-shaped object 80 that is perpendicular to the optical axis of the camera device 20. In addition, the rotating device 16 is an example of a rotating unit disclosed herein.

[0044] The lighting device 18 irradiates ultraviolet rays onto the object 80 (more accurately, the space containing the object 80) to which the magnetic powder is attached. The lighting device 18 can be, for example, a ring-shaped lighting device that emits ultraviolet rays. In addition, the lighting device 18 is arranged on the same axis as the camera device 20 described later, and is configured to irradiate ultraviolet rays onto a range (irradiation range) including the imaging range 20A of the camera device 20. In addition, the lighting device 18 can also be configured so that the ultraviolet rays from the lighting device 18 are uniformly irradiated to the irradiation range by providing a diffusion film or the like in front of the lighting device 18. In addition, the lighting device 18 is an example of the lighting unit disclosed in the present invention.

[0045] The imaging device 20 captures multiple images of an object 80, to which ultraviolet-excited magnetic powder is attached, at multiple locations along a first direction, described later. The imaging device 20 is a camera capable of capturing the surface of the object 80 (for example, the outer peripheral surface around the rotation axis) as a monochrome or color image. For example, the imaging device 20 may be a two-dimensional camera comprising a two-dimensional array of imaging elements such as CCDs (Charge Coupled Devices) or CMOSs ​​(Complementary Metal Oxide Semiconductors).

[0046] The imaging device 20 is configured so that its optical axis is included in a plane containing the rotation axis of the object 80. The imaging device 20 captures a range that at least includes the outer peripheral surface of the object 80 as viewed from the imaging device 20, as an imaging range 20A, so that the outer peripheral surface is included in the field of view of the imaging device 20. Furthermore, as described above, the imaging device 20 is configured so that its optical axis is coaxial with the lighting device 18. By coaxially arranging the imaging device 20 and the lighting device 18, the imaging sensitivity can be brought to a state close to maximum. In the captured image, along with the surface of the object 80 included in the imaging range, magnetic powder attached to the surface of the object 80 and emitting light due to ultraviolet light emitted from the lighting device 18 is also reflected.

[0047] When generating a plurality of camera images, the camera device 20 generates each camera image by changing the relative positional relationship with an object 80 along a certain direction that can be set arbitrarily. Hereinafter, the direction in which such a relative positional relationship changes is referred to as the first direction. Therefore, for example, when a camera image is generated in a state where the object 80 is rotated by the rotating device 16, the first direction becomes the direction (rotational direction, circumferential direction) in which the rotation speed of the object 80 is directed on the optical axis of the camera device 20, and each position of the surface of the object 80 reflected in each camera image is arranged along the optical axis of the camera device 20 in appearance along the first direction. In addition, the direction orthogonal to the first direction is referred to as the second direction. Therefore, for example, when a camera image is generated in a state where the object 80 is rotated by the rotating device 16, the second direction becomes a direction parallel to the rotation axis when the object 80 is rotated. The camera device 20 is an example of the camera unit disclosed herein.

[0048] The marking device 22 is a device that marks the surface of the object 80. For example, the marking device 22 may be a robotic arm with a drawing tool at its tip. In this case, the marking device 22 draws a designated mark, text, or symbol at a designated location on the surface of the object 80, such as a location where a problematic surface characteristic is detected. This makes it easy to identify locations on the surface of the product object 80 that may be causing quality problems.

[0049] The processing device 24 is a device that performs various controls related to the surface inspection device 10 and various calculations related to the inspection of the object 80, and is composed of a computer having a hardware configuration described later. Figure 2 1 shows an example of the functional configuration of the processing device 24. The processing device 24 includes a control unit 26 and an inspection unit 28 as functional configurations.

[0050] The control unit 26 controls various devices included in the surface inspection apparatus 10 to perform their functions. More specifically, the control unit 26 controls the magnetizing device 12, the magnetic powder adhering device 14, the rotating device 16, the lighting device 18, the imaging device 20, and the marking device 22.

[0051] Specifically, at the start of the inspection, the control unit 26 controls the magnetizing device 12 and the magnetic powder adhering device 14 to operate, magnetizing the object 80. A magnetic powder liquid is then applied to the magnetized object 80, causing the magnetic powder to adhere to the surface of the object 80. Furthermore, at the start of the inspection, the control unit 26 turns on the lighting device 18 to initiate ultraviolet irradiation. Furthermore, at the start of the inspection, the control unit 26 operates the rotating device 16 to rotate the object 80. The lighting device 18 remains on until the inspection is complete. The control unit 26 synchronizes the timing of various controls by obtaining the rotation angle and rotation period of the object 80 from, for example, a rotary encoder provided on the rotating device 16.

[0052] Furthermore, the control unit 26 causes the imaging device 20 to capture images of the surface of the object 80 in accordance with a predetermined period of rotation of the object 80, thereby generating a plurality of captured images covering a plurality of points along the circumferential direction (first direction) over the entire circumference of the outer surface of the object 80. The control unit 26 then obtains the captured images generated by the imaging device 20 from the imaging device 20. Once the captured images covering the entire circumference of the object 80 have been generated, the control unit 26 controls the rotating device 16 to stop the rotation of the object 80.

[0053] The inspection unit 28 inspects the surface of the object 80 based on the captured images obtained by the control unit 26. The inspection unit 28 includes an image concatenation unit 30, a flaw identification unit 32, a second-direction summing unit 34, a first-direction summing unit 36, a continuous flaw determination unit 38, and an output unit 44 as functional units for performing the inspection. During the inspection by the inspection unit 28, each functional unit performs data processing on the multiple captured images obtained by the control unit 26, thereby obtaining inspection results regarding the surface properties of the object 80. The inspection results will be described later.

[0054] Figure 3The following illustrates an example of generating an inspection image 52 from multiple captured images 50. The image concatenation unit 30 extracts a central image region 50A, which forms a portion of each captured image 50, from each captured image 50 obtained by the imaging device 20. The central image region 50A is located at the center of the captured image 50 in the longitudinal direction (a direction corresponding to a first direction. For example, when the captured image is obtained by rotating the object 80, this direction corresponds to the rotational direction of the object 80 on the optical axis of the imaging device 20) and extends from one end to the other end of the captured image 50 in the transverse direction (a direction corresponding to a second direction. For example, when the captured image is obtained by rotating the object 80, this direction corresponds to the direction of the rotation axis of the object 80). The central image region 50A is an area on the captured image 50 corresponding to an imaging region 20A1 within the imaging range 20A of the imaging device 20, located on the optical axis of the imaging device 20 and extending from one end to the other end in the direction of the rotational axis of the object 80. Furthermore, the control unit 26 causes the imaging device 20 to capture images of the surface of the rotating object 80 at predetermined intervals to obtain a central image region 50A covering the entire circumference of the object 80. The image linking unit 30 then links the central image regions 50A covering the entire circumference of the object 80 for each object 80 in the order in which they were captured, thereby generating an inspection image 52.

[0055] Figure 4 and Figure 5 An example of the inspection image 52 obtained for a plurality of objects 80 is shown. More specifically, Figure 4 and Figure 5 An example of an inspection image 52 obtained for n (n=1, 2, 3, ...) objects 80 having the same shape is shown. Figure 4 In the example shown, each object 80 is rotated from a predetermined posture. Figure 5 In the illustrated example, the plurality of objects 80 include objects 80 rotated from a posture different from a predetermined posture (for example, n=2 objects 80 ).

[0056] As will be described later, during magnetic particle inspection, the position of the magnetic particle flaw is determined based on the coordinates of the inspection image 52, assuming that the object 80 is rotated from a predetermined position. Therefore, if the object 80 is rotated from a position different from the predetermined position, a misalignment occurs between the position of the magnetic particle flaw on the inspection image 52 and the position of the magnetic particle flaw on the surface of the object 80, making it impossible to accurately determine the position of the magnetic particle flaw.

[0057] Therefore, before the inspection, the posture (up and down, left and right, and rotation angle) of the object 80 when it is set is adjusted to a predetermined, prescribed posture, so that the object 80 can be rotated from the predetermined posture. For example, the rotation angle of the object 80 can be detected by a rotary encoder or the like provided in the rotation device 16, and the control unit 26 can control the rotation device 16 based on the detection result, thereby causing the object 80 to rotate from the predetermined posture. In addition, in the case where the object 80 is not a cylindrical, highly symmetrical shape, but rather has an asymmetric shape with an eccentric center of gravity, after the left and right sides of the object 80 are aligned and the object 80 is maintained in a rotatable state, the eccentric center of gravity of the object 80 will cause the object 80 to move to a configuration with its center of gravity downward, thereby enabling the object 80 to rotate from this changed posture, thereby reducing the labor required to adjust the posture. In this way, by rotating the object 80 from a predetermined posture, the positions of the object 80 corresponding to the starting end 52A and the terminal end 52B of the inspection image 52 can be made consistent (aligned) in multiple inspection images 52, so the position of the magnetic powder flaw can be determined based on the coordinates of each inspection image 52.

[0058] Figure 6 An example of n (n=9 as an example) inspection images 52 is shown in FIG. Figure 6 In the example shown, n inspection images 52 of n objects 80 are shown, and in each inspection image 52, a suspected pattern portion 54 is shown as an image. During magnetic particle inspection, magnetic powder liquid is retained due to the shape change portion or skin properties of the object 80, and the retained magnetic powder liquid sometimes shows the same indicative pattern as a flaw (i.e., suspected pattern). Such suspected pattern is not caused by harmful defects (leakage flux) such as flaws on the surface of the object 80, and is therefore harmless to the quality of the object 80. The suspected pattern portion 54 is a portion indicating suspected pattern. As an example, the suspected pattern portion 54 is generated at the corner connecting the side surfaces of the quadrangular prism object 80.

[0059] The suspected textured portion 54 occurs at a specific location on the object 80. Therefore, the location of the suspected textured portion 54 can be determined based on the design shape of the object 80. Furthermore, as described above, by aligning the locations of the object 80 corresponding to the start point 52A and the end point 52B of the inspection image 52 across multiple inspection images 52, the coordinates of the inspection image 52 corresponding to the location of the suspected textured portion 54 remain constant across each inspection image 52. Therefore, based on the location of the suspected textured portion 54 on the object 80, the suspected textured region 56, which is the image region where the suspected textured portion 54 occurs, can be set on the inspection image 52.

[0060] The flaw identification unit 32 identifies the flaws reflected in each inspection image 52 based on the camera image 50. More specifically, the flaw identification unit 32 first obtains data related to the shape of the object 80 and identifies the position of the suspected textured portion 54 on the object 80 based on the shape of the object 80. The data related to the shape of the object 80 may be pre-stored in the processing device 24 or provided to the processing device 24 externally. Then, based on the identified position of the suspected textured portion 54 on the object 80, the flaw identification unit 32 sets the suspected textured region 56, an image region in which the suspected textured portion 54 has occurred, on the inspection image 52.

[0061] Next, the flaw determination unit 32 sets the valid image area in the inspection image 52, excluding the suspected texture area 56, as the inspection area 58. The flaw determination unit 32 then performs predetermined image processing on the inspection area 58. Specifically, the flaw determination unit 32 performs predetermined preprocessing, such as shading and smoothing, on the inspection area 58. Furthermore, the flaw determination unit 32 binarizes the preprocessed inspection area 58 based on a predetermined threshold. The threshold can be determined experimentally in advance based on the type of the object 80, etc. The flaw determination unit 32 then removes noise components from the binarized inspection area 58 through predetermined processing. Thus, a portion of the magnetic powder flaws 60 is extracted from the inspection area 58.

[0062] exist Figure 6 In the example shown, magnetic powder scratches 60 appear as images in inspection images 52 for n = 1, 3, 5, 7, and 9. Magnetic powder scratches 60 are detrimental to the quality of object 80 and may be single scratches or continuous scratches. However, at this point, it is not determined whether magnetic powder scratches 60 are single scratches or continuous scratches. Early detection of continuous scratches is also required to prevent continued production of objects 80 with continuous scratches due to equipment abnormalities. Furthermore, magnetic powder scratches 60 are just one example of the scratches disclosed herein.

[0063] The flaw identification unit 32 identifies magnetic powder flaws 60 that appear in areas other than the suspected textured area 56 on the surface of the object 80 based on the inspection image 52. Specifically, the flaw identification unit 32 determines whether the magnetic powder flaws 60 appear as images in the inspection area 58 of the inspection image 52, and identifies the presence, position, size, and other information of the magnetic powder flaws 60 within the inspection image 52. The magnetic powder flaws 60 are identified based on, for example, the brightness value and coordinates of the magnetic powder flaws 60 appearing as images in the inspection image 52, the pitch obtained from the rotary encoder of the rotating device 16, and other information such as the size of the object 80.

[0064] The control unit 26 specifies the magnetic powder flaw 60 shown in the inspection image 52 via the flaw specifying unit 32 , and stores information (such as the presence, position, and size) related to the specified magnetic powder flaw 60 in the processing device 24 .

[0065] Furthermore, the control unit 26 controls the marking device 22 to mark the position of the magnetic powder flaw 60. Specifically, the control unit 26 notifies the marking device 22 of the position of the magnetic powder flaw 60 stored in the processing device 24 and controls the marking device 22 to mark the position of the magnetic powder flaw 60.

[0066] The second direction addition processing unit 34 adds, for the plurality of inspection images 52 , pixel values ​​of the flaws identified by the flaw identification unit 32 in the direction corresponding to the second direction in the inspection images 52 . Figure 7 An example of a plurality of inspection images 52 and a first flaw image 62 is shown. Figure 7 , shows an example of n (n = 8, for example) inspection images 52 and a first flaw image 62 generated based on the inspection images 52. When the flaw identification unit 32 identifies magnetic powder flaws 60 on the surface of the object 80, the second-direction addition processing unit 34 arranges the n inspection images 52 in a direction corresponding to the second direction, adds the brightness values ​​of the flaws reflected in each of the inspection images 52 in the direction corresponding to the second direction, and generates a first flaw image 62 representing the result of the addition. The following description uses as an example a case where the second direction is the axial direction of the rotation axis when the object 80 is rotated, and the direction corresponding to the second direction is the lateral direction in the inspection image 52 (or the captured image 50). In other words, the first direction is the circumferential direction when the object 80 is rotated, and the direction corresponding to the first direction is the longitudinal direction in the inspection image 52 (or the captured image 50).

[0067] exist Figure 7 In the example shown, a single flaw 66, a magnetic powder flaw 60, is displayed as an image in the inspection images 52 with n = 3, 5, and 6, as a flaw identified by the flaw identification unit 32. The second-direction addition processing unit 34 adds the pixel values ​​corresponding to the single flaw 66 displayed as an image in the inspection image 52 with n = 3, the pixel values ​​corresponding to the single flaw 66 displayed as an image in the inspection image 52 with n = 5, and the pixel values ​​corresponding to the single flaw 66 displayed as an image in the inspection image 52 with n = 6, in a direction corresponding to the second direction (axial direction), thereby generating a first flaw image 62. Since the first flaw image 62 is not processed in the direction corresponding to the first direction and is only added in the direction corresponding to the second direction, the first flaw image 62 includes information regarding the position of the magnetic powder flaw 60 in the first direction (circumferential position).

[0068] When adding pixel values ​​corresponding to the second direction in the inspection image 52, the second-direction addition processing unit 34 may convert the size of the magnetic powder flaw 60 in the inspection image 52 (i.e., the width dimension of the pixel corresponding to the magnetic powder flaw 60) to a predetermined size (hereinafter referred to as the flaw width length) regardless of the size of the magnetic powder flaw 60, and perform the addition based on this. Specifically, for example, even if a magnetic powder flaw 60 with a length of 30 pixels in the axial direction of the object 80 and a magnetic powder flaw 60 with a length of 2 pixels are depicted as images in the inspection image 52, the length of the first flaw image 62 is uniformly set to 5 pixels. Alternatively, even if a magnetic powder flaw 60 with a width of 8 pixels in the circumferential direction of the object 80 and a magnetic powder flaw 60 with a width of 2 pixels are depicted as images in the inspection image 52, the width of the first flaw image 62 is converted to 6 pixels. Furthermore, the position of the size-converted first flaw image 62 is set based on the center of gravity of the magnetic powder flaw 60. This allows the summed value to depend not on the widthwise size of the magnetic powder flaw 60 but on the number of magnetic powder flaws 60 present at the corresponding position.

[0069] Figure 8 An example of a plurality of inspection images 52 and a second flaw image 64 is shown. Figure 8 , an example of n (n = 8, for example) inspection images 52 and a second flaw image 64 generated based on the inspection images 52 is shown. Similar to the first flaw image 62, the first-direction addition processing unit 36 ​​arranges the n inspection images 52 in a direction corresponding to the first direction, adds the brightness values ​​of the flaws reflected in each of the inspection images 52 in the direction corresponding to the first direction, and generates a second flaw image 64 representing the added result.

[0070] Figure 8 The example shown is with Figure 7In the same example as shown, a single flaw 66 is identified as a flaw by the flaw identification unit 32 in the inspection images 52 with n = 3, 5, and 6. The first-direction addition processing unit 36 ​​adds the pixel values ​​corresponding to the single flaw 66 that appears as a portrait in the inspection image 52 with n = 3, the pixel values ​​corresponding to the single flaw 66 that appears as a portrait in the inspection image 52 with n = 5, and the pixel values ​​corresponding to the single flaw 66 that appears as a portrait in the inspection image 52 with n = 6, in a direction corresponding to the first direction (circumferential direction), thereby generating a second flaw image 64. Since the second flaw image 64 does not process the direction corresponding to the second direction but only adds the pixel values ​​in the direction corresponding to the first direction, the second flaw image 64 includes information regarding the position of the magnetic powder flaw 60 in the second direction (axial position).

[0071] When adding pixel values ​​corresponding to the first direction in the inspection image 52, the first-direction addition processing unit 36 ​​may convert the size of the magnetic powder flaw 60 in the inspection image 52 (i.e., the circumferential dimension of the pixel corresponding to the magnetic powder flaw 60) to a predetermined size (hereinafter referred to as the flaw circumference) regardless of the size of the magnetic powder flaw 60, and perform the addition based on this. Specifically, for example, even if a 6-pixel width and a 2-pixel width of a magnetic powder flaw 60 in the circumferential direction of the object 80 are depicted as images in the inspection image 52, the width of the second flaw image 64 is uniformly 3 pixels. Furthermore, for example, even if a 60-pixel length and a 10-pixel length of a magnetic powder flaw 60 in the axial direction of the object 80 are depicted as images in the inspection image 52, the length of the second flaw image 64 is converted to 20 pixels. Furthermore, the position of the second flaw image 64 after size conversion is set based on the center of gravity of the magnetic powder flaw 60. This allows the summed value to depend not on the circumferential size of the magnetic powder flaw 60 but on the number of magnetic powder flaws 60 present at the corresponding position.

[0072] Figure 9 An example of a plurality of inspection images 52 and a first flaw image 62 is shown. Figure 9 Another example of n (n=8 as an example) inspection images 52 and first flaw images 62 generated based on the inspection images 52 is shown. Figure 7 、 Figure 8 In the example shown, continuous flaws 68 appear as images of magnetic powder flaws 60 in the inspection images 52 for n=6, 7, and 8. The continuous flaws 68 are formed continuously at the same position on the surface of the object 80 for n=6, 7, and 8.

[0073] exist Figure 9 In the example shown, in addition to the first flaw image 62 corresponding to the single flaw 66, the second-direction addition processing unit 34 also adds the pixel values ​​corresponding to the continuous flaw 68 that appears as a portrait in the inspection image 52 with n=6, the pixel values ​​corresponding to the continuous flaw 68 that appears as a portrait in the inspection image 52 with n=7, and the pixel values ​​corresponding to the continuous flaw 68 that appears as a portrait in the inspection image 52 with n=8 in the second direction to generate the first flaw image 62.

[0074] exist Figure 9 In the example, the pixel values ​​corresponding to the continuous flaws 68 shown as images in the inspection images 52 with n=6, 7, and 8 are added together, thereby representing the non-continuous flaws 68 in the histogram 70. Figure 7 Added values ​​for the example shown.

[0075] Here, the sum value (axial sum value) obtained by adding in the direction corresponding to the axial direction and the axial threshold value are described. The sum value obtained by adding in the direction corresponding to the axial direction corresponds to the total number of pixels in the axial direction of the magnetic powder flaw 60 after addition. Figure 9 , the sum value obtained by adding in the direction corresponding to the axial direction according to the length of the histogram 70 shown by the added first flaw image 62 is shown. The axial threshold is a threshold for determining whether a continuous flaw 68 has occurred. That is, when the sum value obtained by adding in the direction corresponding to the axial direction exceeds the axial threshold, there is a possibility that a continuous flaw 68 has occurred. Figure 9 In the example shown, the axial threshold is set to a value that is greater than or equal to 2 pixels and less than or equal to 3 pixels in the histogram in the first flaw image 62. The specific value of the axial threshold can be set arbitrarily.

[0076] As described above, when generating the first flaw image 62, regardless of the size of the magnetic powder flaws 60 (the length in the direction corresponding to the width direction of the magnetic powder flaws 60 in the inspection image 52), the size of the magnetic powder flaws 60 in the inspection image 52 can be set to a predetermined size. Thus, even if the lengths of the magnetic powder flaws 60 vary or noise is present in the inspection image 52, the summed value in the direction corresponding to the axial direction corresponding to the number of magnetic powder flaws 60 can be obtained without being affected.

[0077] Figure 10 An example of a plurality of inspection images 52 and a second flaw image 64 is shown. Figure 10 Use with Figure 9The same example as the example shown is another example of n (as an example, n=8) inspection images 52 and a second flaw image 64 generated based on the inspection image 52. In the inspection images 52 of n=6, 7, and 8, a continuous flaw 68 as a magnetic powder flaw 60 is shown as an image.

[0078] exist Figure 10 In the example shown, in addition to the second flaw image 64 corresponding to the single flaw 66, the first-direction addition processing unit 36 ​​also adds the pixel values ​​corresponding to the continuous flaw 68 that appears as a portrait in the inspection image 52 with n=6, the pixel values ​​corresponding to the continuous flaw 68 that appears as a portrait in the inspection image 52 with n=7, and the pixel values ​​corresponding to the continuous flaw 68 that appears as a portrait in the inspection image 52 with n=8 in the first direction to generate the second flaw image 64.

[0079] exist Figure 10 In the example, the pixel values ​​corresponding to the continuous flaws 68 shown as images in the inspection images 52 with n=6, 7, and 8 are added together, thereby representing the non-continuous flaws 68 in the histogram 72A. Figure 8 Added values ​​for the example shown.

[0080] In addition, the single flaws 66 shown as images in the inspection images 52 of n=3 and 5 have different shapes, but are formed at the same position in the axial direction of the object 80. Figure 10 , the pixel values ​​corresponding to the single flaw 66 represented as an image in the inspection images 52 with n=3 and 5 are added together, thereby showing a histogram 72B.

[0081] Here, the sum value (circumferential sum value) obtained by adding in the direction corresponding to the circumferential direction and the circumferential threshold value are explained. The sum value obtained by adding in the direction corresponding to the circumferential direction corresponds to the total number of pixels in the circumferential direction of the magnetic powder flaw 60 after addition. Figure 10 , the summed value obtained by adding in the direction corresponding to the circumferential direction according to the length of the histograms 72A and 72B shown in the added second flaw image 64 is shown. The circumferential threshold is a threshold for determining whether a continuous flaw 68 has occurred. That is, when the summed value obtained by adding in the direction corresponding to the circumferential direction exceeds the circumferential threshold, there is a possibility that a continuous flaw 68 has occurred. Figure 10 In the example shown, the circumferential threshold is set to a value of 2 pixels or more and less than 3 pixels in the histogram in the second flaw image 64. The specific value of the circumferential threshold can be set arbitrarily.

[0082] As described above, when generating the second flaw image 64, regardless of the size of the magnetic powder flaws 60 (the length of the magnetic powder flaws 60 in the inspection image 52 in the direction corresponding to the circumferential direction), the second flaw image 64 can be generated by setting the size of the magnetic powder flaws 60 in the inspection image 52 to a predetermined size. Thus, even if the widths of the magnetic powder flaws 60 vary or if noise is present in the inspection image 52, the summed value in the direction corresponding to the circumferential direction corresponding to the number of magnetic powder flaws 60 can be obtained without being affected.

[0083] Figure 11 An example of a plurality of inspection images 52 and a first flaw image 62 is shown. Figure 11 In FIG. 2 , another example 2 of n (n=8 as an example) inspection images 52 and a first flaw image 62 generated based on the inspection images 52 is shown. Figures 7 to 10 In the example shown, a single flaw 66 appears as a portrait in the inspection images 52 with n=3, 5, and 6, and a continuous flaw 68 appears as a portrait in the inspection images 52 with n=1, 5, and 8.

[0084] exist Figure 11 In the example shown, the second direction addition processing unit 34 adds the pixel values ​​corresponding to the single flaw 66 and the pixel values ​​corresponding to the continuous flaw 68 in the second direction, similarly to the above, to generate the first flaw image 62. Figure 11 In the example, the pixel values ​​corresponding to the single flaw 66 shown as an image in the inspection image 52 with n=5 and the pixel values ​​corresponding to the continuous flaw 68 shown as an image in the inspection images 52 with n=1, 5, and 8 are added together, thereby showing the non-uniformity of the image by a histogram 70. Figure 7 and Figure 9 Added values ​​for the example shown.

[0085] Figure 12 An example of a plurality of inspection images 52 and a second flaw image 64 is shown. Figure 12 In FIG. 2 , another example 2 of n (n=8 as an example) inspection images 52 and a second flaw image 64 generated based on the inspection images 52 is shown. Figure 11 In the example shown, in the inspection images 52 with n=3, 5, and 6, a single flaw 66 appears as a portrait, and in the inspection images 52 with n=1, 5, and 8, a continuous flaw 68 appears as a portrait.

[0086] exist Figure 12 In the example shown, the first direction addition processing unit 36 ​​adds the pixel values ​​corresponding to the single flaw 66 and the pixel values ​​corresponding to the continuous flaw 68 in the first direction, similarly to the above, to generate the second flaw image 64. Figure 12In the example, the pixel values ​​corresponding to the single flaw 66 shown as an image in the inspection image 52 with n=6 and the pixel values ​​corresponding to the continuous flaw 68 shown as an image in the inspection images 52 with n=1, 5, and 8 are added together, thereby showing the non-uniformity of the image by a histogram 72. Figure 8 and Figure 10 Added values ​​for the example shown.

[0087] The continuous flaw determination unit 38 determines whether a single flaw has occurred sporadically on the surface of the object 80, or whether a continuous flaw has occurred continuously, based on the summed values ​​of the second-direction summing unit 34 and the first-direction summing unit 36. More specifically, the continuous flaw determination unit 38 determines whether continuous flaws 68 have occurred continuously on multiple objects 80 based on the summed values ​​of the axial directions added by the second-direction summing unit 34 and the summed values ​​of the circumferential directions added by the first-direction summing unit 36. Specifically, the continuous flaw determination unit 38 compares the summed values ​​of the axial directions with an axial threshold value and the summed values ​​of the circumferential directions with a circumferential threshold value. If the summed values ​​of the axial directions are lower than the axial threshold value or the summed values ​​of the circumferential directions are lower than the circumferential threshold value, the continuous flaw determination unit 38 determines that no continuous flaws 68 have occurred. On the other hand, when the added value in the axial direction exceeds the axial threshold and the added value in the circumferential direction exceeds the circumferential threshold, the continuous flaw determination unit 38 determines that a continuous flaw 68 has occurred.

[0088] For example, in Figure 7 、 Figure 8 In the example shown, the sum of the values ​​corresponding to the circumferential directions is lower than the circumferential threshold, or the sum of the values ​​corresponding to the axial directions is lower than the axial threshold, so the continuous flaw determination unit 38 determines that no continuous flaw 68 has occurred. Figures 9 to 12 In the example shown, the added value in the axial direction exceeds the axial threshold value, and the added value in the circumferential direction exceeds the circumferential threshold value. Therefore, the continuous flaw determination unit 38 determines that a continuous flaw 68 has occurred.

[0089] In addition, for example, Figure 11 In the inspection images 52 with n=1 and 8, there are no continuous flaws 68, and in the inspection image 52 with n=5, there are four single flaws 66 in parallel. The axial sum value also exceeds the threshold. Figure 12By also checking the circumferential summed value in this manner, the circumferential summed value does not exceed the threshold value, and thus it can be determined that the magnetic powder flaw 60 is not a continuous flaw 68. In other words, in order to determine that the magnetic powder flaw 60 is a continuous flaw 68, it is necessary that the summed value in the axial direction exceeds the axial threshold value, and the summed value in the circumferential direction exceeds the circumferential threshold value.

[0090] However, if the flaw identification unit 32 determines that a magnetic powder flaw 60 exists on the surface of the object 80, and the continuous flaw identification unit 38 subsequently determines that a continuous flaw 68 has occurred, the magnetic powder flaw 60 on the surface of the object 80 is determined to be a continuous flaw 68. On the other hand, if the flaw identification unit 32 determines that a magnetic powder flaw 60 exists on the surface of the object 80, and the continuous flaw identification unit 38 subsequently determines that a continuous flaw 68 has not occurred, the magnetic powder flaw 60 on the surface of the object 80 is determined not to be a continuous flaw 68, that is, the magnetic powder flaw 60 is determined to be a single flaw 66. Therefore, the continuous flaw identification unit 38 can be understood as a determination unit that determines whether the magnetic powder flaw 60 is a single flaw 66 or a continuous flaw 68.

[0091] When the continuous flaw determination unit 38 determines that a continuous flaw 68 has occurred, the output unit 44 (continuous flaw position output unit) determines the position of the continuous flaw 68 based on the summed values ​​in the axial direction, the summed values ​​in the circumferential direction, the circumferential position of the magnetic powder flaw 60 shown in the first flaw image 62, and the axial position of the magnetic powder flaw 60 shown in the second flaw image 64. Specifically, the output unit 44 determines the circumferential position of the continuous flaw 68 based on the histogram 70 where the summed values ​​in the axial direction exceed the axial threshold value, and determines the axial position of the continuous flaw 68 based on the histogram 72 where the summed values ​​in the circumferential direction exceed the circumferential threshold value.

[0092] Figure 13 An example of the position of the continuous flaw 68 determined based on the plurality of inspection images 52 is shown in FIG. Figure 13 , an example of the position of the continuous flaw 68 specified based on n (n=8 as an example) inspection images 52 is shown. Figure 13 The example shown is with Figure 11 、 Figure 12 The example shown is the same as the example. Figure 13 As shown, the position where the circumferential position of the continuous flaw 68 overlaps with the axial position of the continuous flaw 68 is determined as the position of the continuous flaw 68. Then, the output unit 44 outputs the inspection result including the result that the continuous flaw determination unit 38 determines that the continuous flaw 68 has occurred and the position of the continuous flaw 68 to the outside.

[0093] The output unit 44 may also identify the object 80 determined by the continuous flaw determination unit 38 to have a continuous flaw 68, and include the identified object 80 and the position of the identified continuous flaw 68 in the inspection result. Furthermore, the output unit 44 may include warning information in the inspection result that warns of an abnormality indicating the occurrence of a continuous flaw 68.

[0094] Figure 14 This figure shows an example of the hardware configuration of the processing device 24. The processing device 24 is configured as a computer. The processing device 24 includes a CPU (Central Processing Unit) 90, a memory 92, a storage device 94, an input device 96, an output device 98, a storage medium reader 100, and a communication interface 102. These components are connected via a bus 104 so that they can communicate with each other.

[0095] The storage device 94 stores programs for performing surface inspection processing. The CPU 90 is a central processing unit that executes various programs and controls various components. Specifically, the CPU 90 reads programs from the storage device 94 and executes them using the memory 92 as a work area. The CPU 90 controls the aforementioned components and performs various calculations according to the programs stored in the storage device 94.

[0096] Memory 92 is composed of RAM (Random Access Memory) and serves as a work area for temporarily storing programs and data. Storage device 94 is composed of ROM (Read Only Memory), an HDD (Hard Disk Drive), or an SSD (Solid State Drive), and stores various programs including the operating system and various data.

[0097] The input device 96 is a device for performing various inputs, such as a keyboard and a mouse. The output device 98 is a device for outputting various information, such as a display and a printer. A touch panel display can be used as the output device 98 to function as the input device 96.

[0098] The storage medium reader 100 reads data stored in various storage media, such as CD (Compact Disc) ROMs, DVD (Digital Versatile Disc) ROMs, Blu-ray Discs, or USB (Universal Serial Bus) memories, and writes data to the storage media. The communication I / F 102 is an interface for communicating with other devices. For example, the communication I / F 102 uses a standard interface such as Ethernet (registered trademark), FDDI, or Wi-Fi (registered trademark).

[0099] The computer is composed of a CPU 90, a memory 92 and a storage device 94. The computer can be configured as a subcomputer that controls the actions of a part of the processing device 24, or as a main computer that controls the actions of the entire processing device 24. A part or all of the computer can use, for example, an integrated circuit such as LSI (Large Scale Integration) or an IC (Integrated Circuit) chipset. The integration of the computer is not limited to LSI, and a dedicated circuit or processor can also be used. Moreover, the processor mentioned here is a processor in a broad sense, which can be a general-purpose processor, or can include a dedicated processor such as GPU (Graphics Processing Unit), ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array). In addition, the actions of the processor can be completed not only by one processor, but also by the collaboration of multiple processors existing in physically separate locations.

[0100] (Explanation of surface inspection method)

[0101] Figure 15 This figure shows an example of a flow of surface inspection processing executed by the CPU 90 of the processing device 24. The CPU 90 reads a program for executing the surface inspection processing from the storage device 94, expands it in the memory 92, and executes it. This allows the CPU 90 to function as each functional unit of the processing device 24 and execute the surface inspection processing. This allows the surface inspection method to be executed in the surface inspection device 10. In the surface inspection method, the object 80 is first transported to the surface inspection device 10.

[0102] Next, in step S10, the control unit 26 controls the magnetizing device 12 and the magnetic powder adhering device 14 to operate. Consequently, the magnetizing device 12 performs a magnetizing step for magnetizing the object 80, and the magnetic powder adhering device 14 performs a magnetic powder adhering step for applying a magnetic powder liquid to the object 80 magnetized by the magnetizing device 12 to adhere magnetic powder.

[0103] Next, in step S12, the control unit 26 controls and turns on the lighting device 18. Thus, the lighting device 18 performs an illumination step of illuminating the space including the object 80 with ultraviolet rays.

[0104] Next, in step S14, the control unit 26 controls and operates the rotating device 16. Thus, the rotating device 16 performs a rotation step of rotating the object 80 about the central axis of the object 80 as the rotation axis.

[0105] Next, in step S16, the control unit 26 controls the imaging device 20 so as to capture an image of the surface of the object 80 in accordance with a predetermined period of rotation of the object 80. Thus, the imaging device 20 executes an imaging step of capturing an image of the surface of the object 80 to which the magnetic powder is attached, within a space illuminated with ultraviolet light by the lighting device 18.

[0106] Next, in step S18, the image linking unit 30 executes an image linking step. In this image linking step, the central image region 50A is extracted from each captured image 50, and the central image regions 50A of the entire circumference of the object 80 are linked in the order in which they were captured to generate an inspection image 52. In this manner, an inspection image 52 is acquired for each object 80.

[0107] Next, in step S20 , the flaw determination unit 32 executes an area setting step, in which the position of the suspected texture portion 54 on the surface of the object 80 is determined, and an image area where the suspected texture portion 54 is generated, namely, a suspected texture area 56 is set on the inspection image 52 .

[0108] Next, in step S22, the flaw identification unit 32 performs an image processing step. In this image processing step, the valid image area of ​​the inspection image 52, excluding the suspected texture area 56, is set as the inspection area 58. Prescribed image processing (e.g., prescribed preprocessing, binarization, and noise removal) is performed on the inspection area 58. As a result, the magnetic powder flaws 60 are partially extracted from the inspection area 58.

[0109] Next, in step S24, the flaw identification unit 32 executes a flaw identification step. In this flaw identification step, the flaw identification unit 32 determines whether magnetic powder flaws 60 are present on the surface of the object 80 by determining whether magnetic powder flaws 60 appear as images in the inspection area 58 of the inspection image 52. If the flaw identification unit 32 determines that magnetic powder flaws 60 are present on the surface of the object 80, the surface inspection process proceeds to step S26.

[0110] Then, in step S26 , the control unit 26 executes a storage step of identifying the position of the magnetic powder flaw 60 from the inspection image 52 and storing the identified position of the magnetic powder flaw 60 in the storage device 94 .

[0111] Next, in step S28 , the control unit 26 executes a marking step. In the marking step, the control unit 26 controls the marking device 22 to mark the position of the magnetic powder flaw 60 .

[0112] Next, in step S30, an image generation step is performed. In the image generation step, the second direction addition processing unit 34 generates a first flaw image 62 representing the circumferential position of the magnetic powder flaw 60 based on the inspection image 52, and the first direction addition processing unit 36 ​​generates a second flaw image 64 representing the axial position of the magnetic powder flaw 60.

[0113] Next, in step S32, the second-direction addition processing unit 34 performs a second-direction addition processing step, in which the pixel values ​​of the first flaw images 62 of the magnetic powder flaws 60 having the same circumferential position are added together to calculate a summed value in the axial direction. The first-direction addition processing unit 36 ​​performs a first-direction addition processing step, in which the second flaw images 64 of the magnetic powder flaws 60 having the same axial position are added together to calculate a summed value in the circumferential direction. Furthermore, when the first flaw images 62 of the magnetic powder flaws 60 having the same circumferential position are generated, the second-direction addition processing unit 34 performs an addition process by adding the pixel values ​​of the first flaw images 62. Similarly, when the second flaw images 64 of the magnetic powder flaws 60 having the same axial position are generated, the first-direction addition processing unit 36 ​​performs an addition process by adding the pixel values ​​of the second flaw images 64.

[0114] Next, in step S34, the continuous flaw determination unit 38 executes a continuous flaw determination step. In this continuous flaw determination step, based on the result of comparing the added value in the axial direction with the axial threshold value and the result of comparing the added value in the circumferential direction with the circumferential threshold value, it is determined whether a continuous flaw 68 has occurred. If the continuous flaw determination unit 38 determines that a continuous flaw 68 has occurred, the surface inspection process shifts to step S36.

[0115] Then, in step S36, the output unit 44 determines the circumferential and axial positions of the continuous flaw 68 and outputs the inspection results, including the determination that the continuous flaw 68 has occurred and the position of the continuous flaw 68, to the outside. After the inspection results are output to the outside by the output unit 44, the surface inspection process ends.

[0116] On the other hand, if the flaw identification unit 32 determines in step S24 that no magnetic powder flaws 60 are present on the surface of the object 80, the surface inspection process proceeds to step S38. In this case, since there are no magnetic powder flaws 60 on the object 80, the object 80 is conveyed from the surface inspection apparatus 10 to the next process. Furthermore, if the continuous flaw determination unit 38 determines in step S34 that no continuous flaws 68 have occurred, the surface inspection process proceeds to step S38. In this case, since there are no continuous flaws 68 on the object 80, the object 80 is conveyed from the surface inspection apparatus 10 to the next process.

[0117] Then, in step S38, the CPU 90 determines whether the surface inspection has been completed for all objects 80. If the surface inspection has not been completed for all objects 80, the surface inspection process shifts to step S10 to perform a surface inspection on the next object 80. On the other hand, if the surface inspection has been completed for all objects 80, the surface inspection process ends.

[0118] (Effects of Surface Inspection Device 10 and Surface Inspection Method)

[0119] As described above, in this embodiment, an inspection object 80 is magnetized, and a magnetic powder liquid is applied to the magnetized object 80 to adhere magnetic powder. Furthermore, a space containing the object 80 is illuminated with ultraviolet light, and within the ultraviolet-illuminated space, an image of the surface of the object 80, to which magnetic powder adheres, is captured. Based on an inspection image 52 obtained from the captured image of the surface of the object 80, magnetic powder flaws 60 on the surface are identified. Based on the inspection image 52, a first flaw image 62 is generated, indicating positions corresponding to the circumferential direction of the magnetic powder flaws 60, and a second flaw image 64 is generated, indicating positions corresponding to the axial direction of the magnetic powder flaws 60. Furthermore, the pixel values ​​of magnetic powder flaws 60 with the same circumferential position are summed in the direction corresponding to the axial direction to calculate a summed value in the axial direction. Similarly, the pixel values ​​of magnetic powder flaws 60 are summed in the direction corresponding to the circumferential direction to calculate a summed value in the circumferential direction. Furthermore, based on whether the summed values ​​in the axial direction exceed a preset axial threshold value and whether the summed values ​​in the circumferential direction exceed a preset circumferential threshold value, it is determined whether continuous flaws 68 have occurred continuously on a plurality of objects 80. Therefore, it is possible to determine that continuous flaws 68 have occurred at the point in time when the summed values ​​in the axial direction exceed the axial threshold value and the summed values ​​in the circumferential direction exceed the circumferential threshold value, thereby enabling early detection of continuous flaws 68.

[0120] In particular, as in the case of an integrally molded product, by performing each process after the molding process in an order different from the molding order, an excellent effect of being able to detect the continuous flaw 68 early can be achieved even when information related to the continuity of the continuous flaw 68 is lost.

[0121] Furthermore, the second-direction summing unit 34 can convert the size of the magnetic powder flaws 60 in the width direction to a predetermined size before summing, regardless of the size of the magnetic powder flaws 60 (particularly the length of the magnetic powder flaws 60 in the axial direction of the object 80). This allows for the determination of a summed value in the axial direction corresponding to the number of magnetic powder flaws 60, regardless of the size of the magnetic powder flaws 60 (particularly the length of the magnetic powder flaws 60 in the axial direction of the object 80). This allows for the determination of a summed value in the axial direction corresponding to the number of magnetic powder flaws 60, even when the lengths of the magnetic powder flaws 60 vary or when noise is present in the inspection image 52.

[0122] Similarly, the first-direction summing unit 36 ​​can convert the size of the magnetic powder flaws 60 in the circumferential direction to a predetermined size before summing, regardless of the size of the magnetic powder flaws 60 (particularly the width of the magnetic powder flaws 60 in the circumferential direction of the object 80). This allows for the determination of a summed value in the circumferential direction corresponding to the number of magnetic powder flaws 60, even when the widths of the magnetic powder flaws 60 vary or when noise is present in the inspection image 52.

[0123] Furthermore, when it is determined that a continuous flaw 68 has occurred, the position of the continuous flaw 68 is determined and output based on the summed values ​​in the axial direction, the summed values ​​in the circumferential direction, the circumferential position of the magnetic powder flaw 60 determined by the flaw identification unit 32, and the axial position of the magnetic powder flaw 60 determined by the flaw identification unit 32. Therefore, not only can the occurrence of the continuous flaw 68 be detected early, but the position of the continuous flaw 68 can also be determined early.

[0124] Furthermore, by determining the position of the pseudo-texture portion 54 on the surface of the object 80 based on the shape of the object 80, it is possible to determine whether magnetic powder flaws 60 are present in areas other than the pseudo-texture region 56 on the surface of the object 80. This prevents the pseudo-texture portion 54 from being mistakenly determined as the continuous pattern portion 68.

[0125] Furthermore, the object 80 is rotated from a predetermined posture. Therefore, the positions of the object 80 corresponding to the start end 52A and the end end 52B of the inspection image 52 can be aligned in the plurality of inspection images 52. Thus, the position of the magnetic powder flaw 60 can be determined based on the coordinates of each inspection image 52.

[0126] (Variation)

[0127] Figure 16 、 Figure 17 FIG. 5 shows a modified example of the case where the inspection image 52 is corrected. Figure 16 In the example shown, the object 80 is rotated from a predetermined posture. Figure 17 In the example shown, the object 80 is rotated from a posture different from a predetermined posture. A mark 82 is provided on the surface of the object 80. Figure 16 、 Figure 17 In the example shown, markers 82 are provided at the corners of the side surfaces of object 80. Markers 82 are formed, for example, from a material that emits light upon fluorescence excitation. Control unit 26 controls rotation device 16 to rotate object 80 one or more revolutions. Consequently, in inspection image 52 captured by imaging device 20 while object 80 is rotating, two markers 82 appear as images.

[0128] The inspection unit 28 includes a correction unit 46 as an additional functional unit. The correction unit 46 corrects the inspection image 52 so that the mark 82 is arranged at a predetermined position in the inspection image 52. Figure 16 、 Figure 17 In the example shown, the inspection image 52 is corrected so that the mark 82 appears as an image at the start end 52A and the end end 52B of the inspection image 52. The corrected inspection image 52 is used in the surface inspection process described above.

[0129] If configured in this manner, for example, even when the object 80 is rotated from a posture different from a predetermined posture, the positions of the object 80 corresponding to the starting end 52A and the terminal end 52B of the inspection image 52 can be made consistent in multiple inspection images 52, so that the position of the magnetic powder flaw 60 can be determined based on the coordinates of each inspection image 52.

[0130] In the above embodiment, a ring-shaped imaging device 20 is used, but a linear imaging device 20 may also be used. Furthermore, a pair of linear lighting devices 18 may be arranged horizontally above and below the imaging device 20. Furthermore, multiple lighting devices 18 may be arranged to illuminate the object 80 from multiple directions, depending on the size and shape of the object.

[0131] The lighting device 18 is arranged to illuminate the surface of the object 80 in a vertical direction, but is not limited to this depending on the shape of the object 80 or the form of the magnetic powder flaws 60. The shape of the lighting device 18 is not limited to a ring or a line.

[0132] Furthermore, the shape of the object 80 is not limited to a quadrangular prism. Depending on the shape of the object 80, the inspection image 52 may be acquired without rotating the object 80 or while moving the object 80 in parallel.

[0133] Furthermore, in the above embodiment, a two-dimensional camera having imaging elements arranged two-dimensionally is used as the imaging device 20 , but a one-dimensional camera having imaging elements arranged one-dimensionally may also be used.

[0134] Furthermore, in the above-described embodiment, the surface of the object 80 may be repaired based on the marked position. The repair may be performed by a repair device or manually.

[0135] In the above embodiment, whether or not continuous flaws 68 have occurred is determined based on a first determination result of whether or not the summed value in the axial direction exceeds an axial threshold value, and a second determination result of whether or not the summed value in the circumferential direction exceeds a circumferential threshold value. However, whether or not continuous flaws 68 have occurred may be determined based on either the first or second determination result.

[0136] As mentioned above, although an example of the present disclosure was described, the present disclosure is not limited to the above, and it is obvious that various modifications other than the above can be made and implemented without departing from the scope of the present disclosure.

[0137] In addition, the disclosure of Japanese Patent No. 2023-016416 is incorporated herein by reference in its entirety.

[0138] Description of Reference Numerals

[0139] 10 surface inspection device; 12 magnetization device; 14 magnetic powder adhesion device; 16 rotation device; 18 lighting device; 20 camera device; 20A camera range; 20A1 camera area; 22 marking device; 24 processing device; 26 control unit; 28 inspection unit; 30 image connection unit; 32 flaw determination unit; 34 second direction addition processing unit; 36 first direction addition processing unit; 38 continuous flaw determination unit; 44 output unit; 46 correction unit; 50 camera image; 50A central image area; 52 inspection image; 52A starting end; 52B end end; 54 suspected texture portion; 56 suspected texture area; 58 inspection area; 62 first flaw image; 64 second flaw image; 80 object; 82 marking.

Claims

1. A surface inspection device for inspecting the properties of the surface of an object. The surface inspection device comprises: a magnetizing unit for magnetizing the object; a magnetic powder attaching unit for applying magnetic powder liquid to the magnetized object to attach magnetic powder to the object; an illumination unit for irradiating ultraviolet light toward the object to which the magnetic powder is attached; an imaging unit that captures images of a plurality of positions along a first direction of the object to which the magnetic powder that emits light when excited by ultraviolet rays is attached, thereby generating a plurality of captured images; as well as an inspection unit that inspects properties of the surface of the object based on the captured image, The inspection unit includes: a flaw identifying unit that identifies, for each inspection image obtained based on the captured image, a flaw reflected in the inspection image; a second direction addition processing unit for adding, for the plurality of inspection images, pixel values ​​of the flaws identified by the flaw identification unit in a direction corresponding to a second direction orthogonal to the first direction in the inspection images; a first direction addition processing unit for adding, for the plurality of inspection images, pixel values ​​of the flaws identified by the flaw identification unit in the inspection images corresponding to the first direction, in the direction corresponding to the first direction; as well as The continuous flaw determination unit determines whether a single flaw that occurs sporadically or a continuous flaw that occurs continuously occurs on the surface of the object based on the addition result of the second-directional addition processing unit and the addition result of the first-directional addition processing unit.

2. The surface inspection device according to claim 1, The second direction addition processing unit sets the size of the flaw identified by the flaw identification unit in the inspection image corresponding to the second direction to a predetermined size, namely, the flaw width, and then adds the size. The first-direction addition processing unit sets the size of the flaw identified by the flaw identification unit in the inspection image corresponding to the first direction to a predetermined size, ie, a flaw perimeter, and then adds the size.

3. The surface inspection device according to claim 1 or 2, The surface inspection device has a continuous scar position output unit. When the continuous scar determination unit determines that a continuous scar has occurred, the continuous scar position output unit determines the position of the continuous scar based on the position of the scar determined by the scar determination unit, the addition result of the second direction addition processing unit, and the addition result of the first direction addition processing unit, and outputs the position of the continuous scar.

4. The surface inspection device according to any one of claims 1 to 3, The surface inspection device includes a rotating unit that rotates the object around a rotation axis parallel to the second direction. The plurality of captured images are generated by the imaging unit capturing images of a plurality of positions of the object along the first direction while the object is rotated by the rotating unit.

5. A surface inspection method for inspecting the properties of the surface of an object. The surface inspection method utilizes a surface inspection device having: a magnetizing unit for magnetizing the object; a magnetic powder attaching unit for applying magnetic powder liquid to the magnetized object to attach magnetic powder to the object; an illumination unit for irradiating ultraviolet light toward the object to which the magnetic powder is attached; an imaging unit that captures images of a plurality of positions along a first direction of the object to which the magnetic powder that emits light when excited by ultraviolet rays is attached, thereby generating a plurality of captured images; as well as an inspection unit that inspects properties of the surface of the object based on the captured image, The surface inspection method uses the inspection unit and includes the following steps: a flaw identifying step of identifying, for each inspection image obtained based on the camera image, a flaw reflected in the inspection image; a second direction addition processing step of adding, for a plurality of the inspection images, pixel values ​​of the flaws determined by the flaw determination unit in a direction corresponding to a second direction orthogonal to the first direction in the inspection images in the direction corresponding to the second direction; a first direction addition processing step of adding, for a plurality of the inspection images, pixel values ​​of the flaws determined by the flaw determination unit in a direction corresponding to the first direction in the inspection images in the direction corresponding to the first direction; as well as The continuous flaw determination step determines whether a single flaw that occurs sporadically or a continuous flaw that occurs continuously occurs on the surface of the object based on the addition result in the second direction addition processing step and the addition result in the first direction addition processing step.

Images