Image inspection device

Abstract

This image inspection device comprises: an image acquisition unit that acquires a captured image obtained by imaging an inspection region using a camera; a camera information storage unit that stores camera distance information pertaining to the camera distance between the camera and the inspection region; a diagram information storage unit that stores diagram information having dimensional information for an irregular processed surface of the inspection region; a planar development processing unit that uses the camera distance information and the diagram information to convert the captured image of the irregular processed surface of the inspection region to a planar development captured image; an image scale correction unit that uses the camera distance information and the diagram information to calculate an image scale correction value for each irregular processed surface of the inspection region, and uses the image scale correction value to generate a scale-corrected image of the planar development captured image; and a defect extraction processing unit that extracts a defect from the scale-corrected image.

Description

Image inspection equipment

[0001] The present invention relates to an image inspection device.

[0002] Patent Document 1 discloses an inner circumference inspection device for a cylindrical container that includes an illumination unit, an imaging unit, and an image processing unit, and that takes images of the inside of the cylindrical container while rotating the cylindrical container.

[0003] JP 2012-26858 A

[0004] In Patent Document 1, the scale of the captured image differs between the front and back sides of the cylinder's side surface, which may make it impossible to extract defects of a specific size. One of the objects of the present invention is to provide an image inspection device that can prevent defects that appear at different scales from being overlooked and improve the accuracy of defect extraction.

[0005] An image inspection device in one embodiment of the present invention comprises an image acquisition unit that acquires an image of an inspection area captured by a camera, a camera information storage unit that stores camera distance information between the camera and the inspection area, a drawing information storage unit that stores drawing information having dimensional information of the unevenly processed surface of the inspection area, a planar development processing unit that uses the camera distance information and the drawing information to convert the image of the unevenly processed surface of the inspection area into a planar developed image, an image scale correction unit that uses the camera distance information and the drawing information to calculate an image scale correction value for each unevenly processed surface of the inspection area and generates a scale-corrected image of the planar developed image using the image scale correction value, and a defect extraction processing unit that extracts defects from the scale-corrected image.

[0006] According to one embodiment of the present invention, it is possible to prevent defects that appear at different scales from being overlooked, and improve the accuracy of defect extraction.

[0007] 1 is an overall schematic diagram of an image inspection device 1 of embodiment 1. FIG. 2 is a main flowchart showing the flow of image inspection in embodiment 1. FIG. 3 is a cross-sectional view of a cylindrical hole in embodiment 1. FIG. 4 is a time chart showing the flow of operation of a planar development processing unit and an image scale correction unit in embodiment 1. FIG. 5 is a time chart showing the flow of generating a virtual boundary line captured image in embodiment 1. FIG. 6 is a time chart showing the flow of acquiring a boundary line of a planar development captured image of a captured image developed rectangularly in the circumferential direction in embodiment 1. (a) and (b) are diagrams for explaining a method of determining whether occlusion has occurred in a captured image in embodiment 1. FIG. 7 is a flowchart showing the flow of determining whether occlusion has occurred and calculating an image scale value in embodiment 1. FIG. 8 is an overall schematic diagram of an image inspection device of embodiment 2. FIG. 9 is a main flowchart showing the flow of image inspection in embodiment 2.

[0008] First Embodiment FIG. 1 is a schematic diagram of an entire image inspection device 1 according to a first embodiment.

[0009] The image inspection device 1 of the first embodiment includes a hole lens camera (camera) 2 and a computer 3. The hole lens camera 2 captures an image of the cylindrical hole (inspection area) 17a, which has a textured surface, of the VTC cover 17, which is the object to be inspected. The computer 3 is, for example, a personal computer, and includes a CPU 4 and a memory 5. The CPU 4 includes a camera communication unit 6, an image acquisition unit 7, a plane development processing unit 8, an image scale correction unit 9, and a defect extraction processing unit 10. The camera communication unit 6 controls the image capture by the hole lens camera 2 and obtains a single image of the textured surface inside the cylindrical hole 17a in one shot, and the image acquisition unit 7 acquires the image. In this way, since a single image capturing the entire textured surface inside the cylindrical hole 17a in one shot is sufficient, the image capture time can be shortened. Details of the processing by the plane development processing unit 8, the image scale correction unit 9, and the defect extraction processing unit 10 will be described later. The memory 5 includes a captured image storage unit 11, a camera information storage unit 12, a drawing information storage unit 13, an image scale correction value storage unit 14, a scale-corrected image storage unit 15, and a defect extraction result storage unit 16. The captured image storage unit 11 stores the captured image of the inside of the cylindrical hole 17a acquired by the captured image acquisition unit 7, the camera information storage unit 12 stores the distance between the Hall lens camera 2 and the cylindrical hole 17a, the resolution of the Hall lens camera 2, etc., the drawing information storage unit 13 stores drawing information having dimensional information (design dimensions, shape data, etc.) of each unevenly processed surface inside the cylindrical hole 17a, the image scale correction value storage unit 14 stores the image scale correction value, the scale-corrected image storage unit 15 stores the scale-corrected image generated by the image scale correction unit 9, and the defect extraction result storage unit 16 stores the defect extraction results of the defect extraction processing unit 10.

[0010] FIG. 2 is a main flowchart showing the flow of image inspection in the first embodiment.

[0011] In step S1, the unevenly machined surface inside cylindrical hole 17a is converted into a planar developed image using the camera distance information stored in camera information storage unit 12, the drawing information stored in drawing information storage unit 13, and a single-shot image of the unevenly machined surface inside cylindrical hole 17a stored in captured image storage unit 11. In step S2, because the image is a single shot of the entire unevenly machined surface inside cylindrical hole 17a, the scale is not uniform between the front and back ends of cylindrical hole 17a (the image appears smaller toward the back). Therefore, an image scale correction value is calculated for each unevenly machined surface inside cylindrical hole 17a from the camera distance information, drawing information, and the planar developed image of the unevenly machined surface inside cylindrical hole 17a, and a scale-corrected image of the planar developed image of the unevenly machined surface inside cylindrical hole 17a with a uniform scale is generated. In step S3, defects are extracted from the scale-corrected image.

[0012] FIG. 3 is a cross-sectional view of the cylindrical hole portion in the first embodiment.

[0013] The uneven surface of the cylindrical hole portion 17a is composed of upper and lower chamfered surfaces 17a1, an upper cylindrical surface 17a2, a grooved surface 17a3, upper and lower grooved side surfaces 17a4, and a lower cylindrical surface 17a5.

[0014] FIG. 4 is a time chart showing the flow of processing by the planar development processing unit and the image scale correction unit in the first embodiment.

[0015] In the first block P1, the planar development processing unit 8 acquires a captured image captured in one shot and stored in the captured image storage unit 11. The grid pattern K (calibration image) of the captured image is projected using a laser or the like. In the second block P2, the planar development processing unit 8 converts the acquired captured image into a rectangular image in the circumferential direction to convert it into a planar development captured image, and corrects the brightness value for each uneven surface within the cylindrical hole 17a of the planar development captured image to process it into an image with uniform brightness. Although the circumferential lengths of each uneven surface are different, the differences are small, so planar development captured images with approximately the same length are used. In the third block P3, an excerpt of the planar development captured image is shown for explanation purposes. In the fourth block P4, the image scale correction unit 9 divides a portion of the planar development captured image into individual unevenly machined surfaces in the cylindrical direction. The image scale correction unit 9 calculates an image scale correction value for each unevenly machined surface based on the dimensional information of each unevenly machined surface, the resolution (pixels) of the Hall-lens camera 2, and the grid pattern K (calibration image) of the captured image so that the captured images of each unevenly machined surface are uniformly scaled. For example, the image scale correction values ​​for each unevenly machined surface are: chamfered surface 17a1: 0.8, upper cylindrical surface 17a2: 1.0, grooved surface 17a3: 1.2, grooved side surface 17a4: 0.8, and lower cylindrical surface 17a5: 1.4. In the fifth block P5, the image scale correction unit 9 generates uniformly scaled, divided, and scale-corrected images for each machined surface based on the image scale correction values ​​for each machined surface. In the sixth block P6, the image scale correction unit 9 combines the divided, scale-corrected images for each machined surface to generate a scale-corrected image. This prevents defects that appear at different scales from being overlooked, improving the accuracy of defect extraction. Also, because images with a uniform scale are obtained, defects can be extracted taking defect size into consideration. Furthermore, because images with uniform brightness are obtained, it is possible to prevent overlooking defects due to overexposure or darkness, improving the accuracy of defect extraction.

[0016] Figure 5 is a time chart showing the flow of generating a virtual boundary line image in embodiment 1, and Figure 6 is a time chart showing the flow of obtaining the boundary line of a planar expanded image of an image expanded rectangularly in the circumferential direction in embodiment 1.

[0017] In the first block Q1, a cylindrical hole and a camera coordinate system identical to those in the real environment are placed in a virtual space, such as a 3D CAD or CG software. The relative positional relationship between the cylindrical hole and the camera coordinate system in the real environment is measured using a calibration method, such as a chessboard. In the second block Q2, a virtual captured image is generated in which a virtual machining surface boundary is added to a virtual captured image of the target cylindrical hole as seen from the camera coordinate system, based on the internal camera parameters of the hole-lens camera (focal length, lens distortion, image pixel discretization value, etc.). That is, in 3D space, for example, a virtual captured image with the extracted virtual machining surface boundary added can be generated by displaying only the lines representing the virtual machining surface boundary and projecting them onto the virtual captured image. This allows the boundary positions between the concave and convex machining surfaces in the virtual captured image to be geometrically extracted. In the third block Q3, the virtual captured image with the virtual machining surface boundary added is rectangularly expanded in the circumferential direction. In the fourth block Q4, each machining surface boundary candidate is estimated using an edge extraction method for the rectangularly expanded planar captured image in the circumferential direction. This makes it possible to estimate the boundary lines between the uneven machined surfaces in the planar unfolded captured image. In the fifth block Q5, a candidate line 1 is identified as the machined surface boundary line of the planar unfolded captured image corresponding to the virtual machined surface boundary to be detected in the virtual captured image, based on the minimum distance between the boundary lines, etc. In the sixth block Q6, the processing of the fifth block Q5 is repeated to identify and obtain all machined surface boundary lines of the corresponding planar unfolded captured image. This makes it possible to identify the boundary lines between the uneven machined surfaces in all planar unfolded captured images.

[0018] 7A and 7B are diagrams illustrating a method for determining whether an occlusion has occurred in a captured image in embodiment 1. That is, it is determined whether an occluded portion that is not captured has occurred in the captured image.

[0019] 6A shows a method for determining whether occlusion has occurred on each machining surface of a captured image by calculating the intersection of the virtual image using the optical axis vector of each machining surface and the camera. It can be determined that the occluded portion is a portion of the groove machining surface 17a3 and the groove machining side surface 17a4. (b) shows a method for determining whether occlusion has occurred on each machining surface of a captured image when a boundary line of the captured image corresponding to the detection target virtual machining surface boundary line of the virtual image is acquired. In the fifth block Q5, a candidate line is identified as the machining surface boundary line of the planar developed captured image corresponding to the detection target virtual machining surface boundary line based on the minimum distance between the boundaries shown in FIG. 6 . If the machining surface boundary line of the planar developed captured image corresponding to the detection target virtual machining surface boundary line is not found within a predetermined threshold distance, it is determined that the machining surface is hidden due to occlusion and the machining surface boundary line is not visible. This allows for a uniform scale to be obtained even in a planar developed captured image that contains an area partially obscured by occlusion, enabling defect extraction taking defect size into account.

[0020] FIG. 8 is a flowchart showing the flow of determining whether an occlusion has occurred and calculating an image scale value in the first embodiment.

[0021] In step S11, a setting value (e.g., 1 mm = 5 pixels) is set to indicate how many pixels wide the actual dimensions (real dimensions) of the inside of the cylindrical hole should be in the captured image, based on the dimensional information in the drawing information and the camera resolution. In step S12, it is determined whether there is no occlusion on each machining surface in the captured image and whether the entire surface is captured in the captured image. If there is no occlusion on each machining surface in the captured image, proceed to step S13; if there is occlusion on each machining surface in the captured image, proceed to step S16. In step S13, the actual cylindrical width of each machining surface and the setting value are used to calculate how many pixels wide the surface will be in the image. In step S14, the pixel width in the captured image is measured. In step S15, an image scale correction value for each machining surface is calculated. (For example, if the actual cylindrical width of the machined surface, 2 mm, is displayed using 5 pixels, the 2 mm is desired to be enlarged to 10 pixels, so the image scale correction value is 2.0.) In step S16, the pixel width of the cylindrical grid width of each machined surface in the captured image is measured from a grid pattern (calibration image) projected using a laser or the like. In step S17, an image scale correction value is calculated from the actual grid size of each machined surface in the captured image and the set value / pixel width. (For example, if a grid with a width of 1 mm in the cylindrical direction is displayed using 2 pixels, the image scale correction value is desired to be corrected to 1 mm = 5 pixels, so the image scale correction value is 5 / 2 = 2.5.) This allows a captured image with a uniform scale to be obtained even for an image with an occluded area. The grid pattern K (calibration image) projected using a laser or the like makes it possible to calculate the scale correction value for the occluded area, and a captured image with a uniform scale is obtained, so that defects can be extracted taking defect size into consideration.

[0022] Next, the effects of the first embodiment will be described.

[0023] (1) The image inspection device 1 includes an image acquisition unit 7 that acquires an image captured by the Hall lens camera 2 of the inside of the cylindrical hole 17a, a camera information storage unit 12 that stores camera distance information between the Hall lens camera 2 and the cylindrical hole 17a, a drawing information storage unit 13 that stores drawing information containing dimensional information of the unevenly machined surfaces (17a1-17a5) inside the cylindrical hole 17a, a planar development processing unit 8 that uses the camera distance information and drawing information to convert the image of the unevenly machined surfaces (17a1-17a5) inside the cylindrical hole 17a into a planar development image, an image scale correction unit 9 that uses the camera distance information and drawing information to calculate an image scale correction value for each of the unevenly machined surfaces (17a1-17a5) inside the cylindrical hole 17a and generate a scale-corrected image of the planar development image, and a defect extraction processing unit 10 that extracts defects from the scale-corrected image. This prevents defects that appear at different scales from being overlooked and improves the accuracy of defect extraction.

[0024] (2) The image scale correction unit 9 generates a scale-corrected image in which the uneven processed surface (17a1-17a5) in the cylindrical hole 17a is displayed at a uniform scale. Therefore, a captured image with a uniform scale can be obtained, making it possible to extract defects taking into account the defect size.

[0025] (3) The image is taken in one shot of the uneven surface inside the cylindrical hole 17a, thereby reducing the image taking time.

[0026] (4) The image scale correction unit 9 determines that occlusion has occurred in the captured image of the unevenly machined surface (17a1-17a5) inside the cylindrical hole 17a based on the camera distance information and drawing information, and generates a scale-corrected image displayed at a uniform scale for each of the unevenly machined surfaces (17a1-17a5) inside the cylindrical hole 17a, taking the occlusion into consideration. Therefore, even in a captured image that has an area that is partially obscured due to occlusion, a captured image with a uniform scale can be obtained, making it possible to extract defects taking the defect size into consideration.

[0027] (5) The image scale correction unit 9 calculates an image scale correction value for each of the unevenly processed surfaces (17a1-17a5) in the cylindrical hole 17a based on the camera distance information, drawing information, and the grid pattern K (calibration image) projected by a laser or the like, and generates a scale-corrected image of the planar development captured image. This makes it possible to calculate the scale correction value for an area that is partially invisible due to occlusion, and obtain a captured image with a uniform scale, making it possible to extract defects taking into account the defect size.

[0028] (6) The plane development processing unit 8 extracts the positions of the virtual machined surface boundary lines between the uneven machined surfaces (17a1-17a5) in the virtual captured image of the cylindrical hole 17a, which are geometrically determined based on the camera distance information and the drawing information. Therefore, it is possible to geometrically extract the boundary line positions between the uneven machined surfaces (17a1-17a5) in the virtual captured image.

[0029] (7) The plane development processing unit 8 uses edge extraction processing on the captured image of the unevenly machined surfaces (17a1-17a5) in the cylindrical hole 17a to estimate boundary line candidates between the unevenly machined surfaces (17a1-17a5) in the cylindrical hole 17a in the captured image. Therefore, it is possible to estimate the boundary lines between the unevenly machined surfaces (17a1-17a5) in the captured image.

[0030] (8) The planar development processing unit 8 calculates the distance between the extracted virtual machining surface boundary line between the geometrically determined uneven machining surfaces (17a1-17a5) in the virtual captured image of the cylindrical hole portion 17a based on the camera distance information and drawing information, and the position of the boundary line group between the uneven machining surfaces (17a1-17a5) in the cylindrical hole portion 17a estimated from the captured image, and the distance between the extracted virtual machining surface boundary line between the geometrically determined uneven machining surfaces (17a1-17a5) and the multiple boundary lines between the uneven machining surfaces (17a1-17a5) estimated from the captured image, and identifies which of the multiple boundary lines between the uneven machining surfaces (17a1-17a5) estimated from the captured image corresponds to the extracted virtual machining surface boundary line between the geometrically determined uneven machining surfaces (17a1-17a5). Thus, the boundary line between the uneven machining surfaces (17a1-17a5) in all captured images can be identified.

[0031] (9) The image scale correction unit 9 divides the planar developed image into individual unevenly processed surfaces (17a1-17a5) within the cylindrical hole 17a, generates scale-corrected images of the planar developed images divided by the image scale correction value for each unevenly processed surface (17a1-17a5) within the cylindrical hole 17a, and then combines the scale-corrected images of the divided planar developed images. This prevents defects on the boundary lines of the unevenly processed surfaces (17a1-17a5) within the cylindrical hole 17a from being displayed as small defects in the image, thereby improving the accuracy of defect extraction.

[0032] (10) The plane development processing unit 8 converts the captured image of the uneven surface (17a1-17a5) inside the cylindrical hole 17a into a plane development captured image by rectangular development. This makes it possible to correct distortion in the captured image that is curved in the circumferential direction of the cylindrical hole 17a, thereby obtaining a captured image with a uniform scale in the circumferential direction of the cylindrical hole 17a, and improving the accuracy of defect extraction.

[0033] (11) The image scale correction unit 9 corrects the brightness value for each unevenly processed surface (17a1-17a5) inside the cylindrical hole 17a in the planar developed captured image, processes it into a captured image with uniform brightness, and calculates an image scale correction value for each unevenly processed surface (17a1-17a5) inside the cylindrical hole 17a based on the camera distance information and drawing information, to generate a scale-corrected image of the planar developed captured image. Therefore, since a captured image with uniform brightness is obtained, it is possible to prevent overexposure and defects from being hidden due to darkness, and to improve the accuracy of defect extraction.

[0034] Second Embodiment FIG. 9 is a schematic diagram of an entire image inspection apparatus according to a second embodiment, and FIG. 10 is a main flowchart showing the flow of image inspection in the second embodiment.

[0035] In the first embodiment, the image scale correction unit 9 calculates an image scale correction value for each processed surface and generates a scale-corrected image, and the defect extraction processing unit 10 extracts defects from the scale-corrected image. In the second embodiment, however, the image scale correction unit 9 calculates an image scale correction value for each processed surface (step S2a in FIG. 10), and the defect extraction processing unit 10 extracts defects from the captured image based on the image scale correction value (step S3a in FIG. 10).

[0036] The other configurations are the same as those of the first embodiment, so the same components are denoted by the same reference numerals and the description thereof will be omitted. Therefore, the second embodiment provides the same effects as the first embodiment.

[0037] [Other Embodiments] Although the embodiments for carrying out the present invention have been described above, the specific configuration of the present invention is not limited to the configuration of the embodiments, and the present invention also includes design changes and the like within the scope of the gist of the invention. The object to be inspected is not limited to a VTC cover, and it may be anything that has a narrow cylindrical or polygonal hole as an inspection area and that performs image inspection of the inside of this hole.

[0038] The present invention is not limited to the above-described embodiments and includes various modifications. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and the present invention is not necessarily limited to those including all of the described configurations. Furthermore, it is possible to replace part of the configuration of one embodiment with the configuration of another embodiment, or to add the configuration of another embodiment to the configuration of one embodiment. Furthermore, it is possible to add, delete, or replace part of the configuration of each embodiment with other configurations.

[0039] This application claims priority to Japanese Patent Application No. 2024-097854, filed June 18, 2024. The entire disclosure of Japanese Patent Application No. 2024-097854, filed June 18, 2024, including the specification, claims, drawings, and abstract, is incorporated herein by reference in its entirety.

[0040] 1 Image inspection device, 2 Hole lens camera (camera), 3 Computer, 4 CPU, 5 Memory, 7 Captured image acquisition unit, 8 Plane development processing unit, 9 Image scale correction unit, 10 Defect extraction processing unit, 12 Camera information storage unit, 13 Drawing information storage unit, 17 VTC cover (inspection object), 17a Cylindrical hole portion (inspection area)

Claims

1. An image inspection device for inspecting defects in an inspection area of ​​an object to be inspected, comprising: an image acquisition unit that acquires an image of the inspection area captured by a camera; a camera information storage unit that stores camera distance information between the camera and the inspection area; a drawing information storage unit that stores drawing information having dimensional information of the unevenly finished surface of the inspection area; a plane development processing unit that uses the camera distance information and the drawing information to convert the image of the unevenly finished surface of the inspection area into a planar developed image; an image scale correction unit that uses the camera distance information and the drawing information to calculate an image scale correction value for each unevenly finished surface of the inspection area, and generates a scale-corrected image of the planar developed image using the image scale correction value; and a defect extraction processing unit that extracts defects from the scale-corrected image.

2. An image inspection device according to claim 1, wherein the image scale correction unit generates scale-corrected images displayed at a uniform scale for each uneven surface of the inspection area.

3. An image inspection device according to claim 2, wherein the image scale correction unit determines whether or not occlusion has occurred in the captured image of the unevenly processed surface of the inspection area based on the camera distance information and the drawing information, and if it is determined that occlusion has occurred, generates a scale-corrected image displayed at a uniform scale for each unevenly processed surface of the inspection area, taking the occlusion into consideration.

4. An image inspection device according to claim 3, wherein the image scale correction unit calculates an image scale correction value for each uneven surface of the inspection area based on the camera distance information, the drawing information, and a calibration image, and generates a scale-corrected image of the planar development captured image using the image scale correction value.

5. An image inspection device according to claim 1, wherein the plane development processing unit extracts the positions of a group of virtual machined surface boundary lines between concave and convex machined surfaces in a geometrically determined virtual captured image of the inspection area based on the camera distance information and the drawing information.

6. An image inspection device according to claim 5, wherein the plane development processing unit uses edge extraction processing on the captured image of the unevenly processed surface of the inspection area, and estimates candidate boundary lines between the unevenly processed surfaces of the inspection area in the captured image.

7. An image inspection device according to claim 6, wherein the plane development processing unit calculates the distance between the extracted virtual machining surface boundary line and the plurality of estimated boundary lines based on the positions of the extracted group of virtual machining surface boundary lines and the positions of the estimated boundary line candidates, and identifies which of the plurality of estimated boundary lines the extracted virtual machining surface boundary line corresponds to.

8. An image inspection device according to claim 1, wherein the image scale correction unit divides the planar unfolded captured image into sections for each uneven surface of the inspection area, generates scale-corrected images of the divided planar unfolded captured images for each uneven surface of the inspection area using the image scale correction value, and then combines the scale-corrected images of the divided planar unfolded captured images.

9. An image inspection device according to claim 1, wherein the plane development processing unit converts the captured image of the unevenly processed surface of the inspection area into a plane development captured image by rectangular development.

10. An image inspection device as claimed in claim 1, wherein the image scale correction unit corrects the brightness value for each uneven surface of the planar unfolded captured image of the inspection area, processes it into an image with uniform brightness, calculates an image scale correction value for each uneven surface of the inspection area based on the camera distance information and the drawing information, and generates a scale-corrected image of the planar unfolded captured image using the image scale correction value.

11. An image inspection device for inspecting defects in an inspection area of ​​an object to be inspected, comprising: an image acquisition unit that acquires an image of the inspection area captured by a camera; a camera information storage unit that stores camera distance information between the camera and the inspection area; a drawing information storage unit that stores drawing information having dimensional information of the unevenly finished surface of the inspection area; a plane development processing unit that uses the camera distance information and the drawing information to convert the image of the unevenly finished surface of the inspection area into a plane development image; an image scale correction unit that uses the camera distance information and the drawing information to calculate an image scale correction value for each of the unevenly finished surfaces of the inspection area; and a defect extraction processing unit that extracts defects from the image of the unevenly finished surface of the inspection area based on the image scale correction value for each of the unevenly finished surfaces of the inspection area.

Images