Inspection Method and Inspection Device

Abstract

The present invention provides an inspection method and an inspection apparatus. This inspection apparatus includes: image sensors (101 to 116) that photograph a sheet (S); a lighting device (2) that irradiates the sheet (S) with light; and an image processing device (7) that generates an image of an object (E) based on the output of the image sensors (101 to 116). The image processing device (7) generates input images (Pa to Pd) that commonly include an image of a specified range of the sheet (S) based on the output of the image sensors (101 to 116). The image processing device (7) extracts the image of the object (E) in the sheet (S) for each generated input image (Pa to Pd). The image processing device (7) correlates the images of the extracted objects E with each other based on the feature amounts of the objects (E) manifested in the images. The image processing device (7) synthesizes the images of the correlated objects (E) with each other.

Description

Technical Field

[0001] The present invention relates to an inspection apparatus and an inspection method for an object to be inspected using a plurality of image sensors, which improve the detection reproducibility and detection probability of an object in the object to be inspected. Background Art

[0002] In the fields of devices such as semiconductors, electronic devices, and secondary batteries, a defect detection apparatus is known that uses a photoelectric conversion type image sensor to detect an object (foreign matter, defect, etc.) in an object to be inspected.

[0003] In recent years, in these fields, due to the high-precision and miniaturization of products, the sizes of foreign matters and defects in the object to be inspected have become smaller. In addition, there are demands for production efficiency improvement and quality improvement, and along with this, high-speed processing of manufacturing processes and improvement of yield are required. For high-speed processing of manufacturing processes and improvement of yield, high resolution and high responsiveness of the image sensor are required.

[0004] However, in the production of a high-resolution and high-responsiveness image sensor, a large amount of development costs and development time are required. Therefore, in Patent Document 1, a high-speed detector is realized by arranging a plurality of image sensors side by side and processing them simultaneously.

[0005] Prior Art Documents

[0006] Patent Documents

[0007] Patent Document 1: Japanese Patent No. 5172162 Summary of the Invention

[0008] An inspection apparatus according to an embodiment of the present invention is an inspection apparatus that inspects an object to be inspected by detecting the size of an object in the object to be inspected, and includes: a plurality of image sensors that photograph the object to be inspected; an illumination device that irradiates light to the object to be inspected; and an image processing device that generates an image of the object based on outputs of the plurality of image sensors. The image processing device generates a plurality of input images that commonly include an image of a specified range of the object to be inspected based on the outputs of the plurality of image sensors, extracts an image of the object in the object to be inspected for each of the generated plurality of input images, establishes an association between the extracted images of the object based on a feature amount of the object appearing in the image, and synthesizes the associated images of the object with each other.

[0009] The inspection method according to an embodiment of the present invention is an inspection method for inspecting an object in an object to be inspected by detecting the size of the object in the object to be inspected using an inspection device. The inspection device includes: a plurality of image sensors for photographing the object to be inspected; an illumination device for irradiating light onto the object to be inspected; and an image processing device for generating an image of the object in the object to be inspected based on the outputs of the plurality of image sensors. The inspection method includes: a step of generating a plurality of input images that commonly include an image of a specified range of the object to be inspected based on the outputs of the plurality of image sensors; a step of extracting an image of the object in the object to be inspected for each of the generated plurality of input images; a step of establishing an association between the extracted images of the object based on the feature amounts of the object manifested in the images; and a step of synthesizing the images of the object for which the association has been established with each other. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 is a side view of the inspection device according to the present embodiment.

[0011] Figure 2 is a top view of the inspection device according to the present embodiment.

[0012] Figure 3 is a top view showing an example of the arrangement of the image sensors in the imaging device according to the present embodiment.

[0013] Figure 4 is a flowchart for explaining the process of the image generation method in the image processing device according to the present embodiment.

[0014] Figure 5A is a diagram for explaining the image generation method in the image processing device according to the present embodiment.

[0015] Figure 5B is a diagram for explaining the image generation method in the image processing device according to the present embodiment.

[0016] Figure 6A is a diagram for explaining the image generation method in the image processing device according to the present embodiment.

[0017] Figure 6B is a diagram for explaining the image generation method in the image processing device according to the present embodiment.

[0018] Figure 7 is a flowchart for explaining the process of the method for extracting an image of an object and the method for synthesizing images of an object in the image processing device according to the present embodiment.

[0019] Figure 8 This is a diagram for explaining the method of synthesizing images of an object in the image processing apparatus according to this embodiment.

[0020] Figure 9A This is a diagram showing the synthesized image of the object output by the inspection apparatus according to this embodiment.

[0021] Figure 9B This is a diagram showing the synthesized image of the object output by an existing inspection apparatus.

[0022] Figure 10 This is a diagram for explaining the relationship between the types of sensors and the types of moving images and the optical system in the inspection apparatus according to this embodiment.

[0023] -Symbol Explanation-

[0024] 1 Imaging device

[0025] 2 Lighting device

[0026] 101 - 116 Image sensors

[0027] 3 - 5 Rollers

[0028] 6 Rotary encoder

[0029] 7 Image processing device

[0030] 8 Conveyor belt

[0031] 9 - 10 Driving devices

[0032] Pa - Pd Input images

[0033] E Object. Detailed Embodiment

[0034] In Patent Document 1, in order to accurately detect an object, a plurality of images output from an image sensor are synthesized to generate a high-precision image. In Patent Document 1, based on the configuration of the image sensor, the positions of the plurality of images are respectively offset (corrected) and then the images are synthesized.

[0035] However, for example, when the direction of the light irradiated by the lighting device is not constant, or when the object to be inspected is three-dimensional, etc., the irradiation method of the light on the object to be inspected may not be constant. In this case, it is feared that the position of the object deviates greatly in the plurality of images output from the image sensor. Therefore, by respectively correcting the positions of the plurality of images based on the configuration of the image sensor, it may not be possible to correct the deviation of the position of the object, and the object cannot be detected.

[0036] In particular, when the object to be inspected is being conveyed and inspected, deviation in the position of the object is likely to occur.

[0037] An object of the present invention is to improve the detection reproducibility and detection probability of an object in an object to be inspected in an inspection apparatus and an inspection method using a plurality of image sensors.

[0038] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. The following description of the preferred embodiments is merely exemplary in nature and is in no way intended to limit the present invention, its applications, or its uses.

[0039] Figure 1 A side view showing the inspection apparatus according to the present embodiment, Figure 2 shows a top view of the inspection apparatus according to the present embodiment. As Figure 1 and Figure 2 shown, the present inspection apparatus includes: an imaging device 1, an illumination device 2, rollers 3 to 5, a rotary encoder 6, and an image processing device 7. A conveyor belt 8 is wound around the outer peripheries of the rollers 3 to 5.

[0040] The present inspection apparatus inspects a sheet S (object to be inspected) configured as a sheet. Specifically, the present inspection apparatus detects an object in the sheet S. Here, as the object, the present inspection apparatus detects defects and foreign matters in the sheet S. Such defects include, for example, not only missing parts or insufficient parts during the production of the sheet S to be inspected, such as short circuits and open circuits in the sheet S, but also damage to the sheet S (for example, scratches caused by contact between the sheet S and other members). When the detected object is larger than a specified size, the present inspection apparatus determines that the sheet S contains the object. The sheet S is conveyed in the direction of the arrow shown by the solid line in Figure 1 and Figure 2 .

[0041] The imaging device 1 includes a plurality of image sensors and photographs the sheet S conveyed by the conveyor belt 8. Here, the imaging device 1 is configured as a line sensor that photographs the sheet S between the rollers 4 and 5. The imaging device 1 sends the pixel signals output from the image sensors to the image processing device 7. In the following description, the scanning direction of the imaging device 1 is set as the X direction, the sub-scanning direction of the imaging device 1 is set as the Y direction, and the direction perpendicular to the X direction and the Y direction is set as the Z direction.

[0042] The illumination device 2 has a light source including, for example, an LED, a laser, a halogen light source, etc., and irradiates light on the scanning area (sheet S) of the imaging device 1 between the rollers 4 and 5. Specifically, the illumination device 2 is arranged such that the irradiation direction of the light is inclined at an angle of about 10° with respect to the conveyor belt 8. That is, the imaging device 1 and the illumination device 2 include a dark-field optical system.

[0043] The roller 3 rotates through the drive mechanism 9, drives the conveyor belt 8, and conveys the sheet S in the direction of the solid-line arrow in the drawing. The drive mechanism 9 is, for example, an actuator connected to the roller 3. The actuator is, for example, a rotary motor.

[0044] The roller 5 causes the conveyor belt 8 to meander in the X direction ( Figure 2 the direction of the dotted-line arrow in) at a prescribed speed through the drive mechanism 10. By causing the conveyor belt 8 to meander in the X direction, the alignment of the sheet S in the process after the inspection by this inspection device is performed. The drive mechanism 10 is, for example, an actuator connected to the roller 5 and causing the roller 5 to reciprocate in the X direction.

[0045] The rotary encoder 6 detects the rotational speed of the roller 4 and detects the movement amount of the sheet S conveyed by the conveyor belt 8. The rotary encoder 6 sends the detected movement amount of the sheet S to the image processing device 7.

[0046] The image processing device 7 is, for example, a computer, and based on the pixel signal received from the imaging device 1 and the movement amount of the sheet S detected by the rotary encoder 6, generates a plurality of captured images by the image generation method described later. The image processing device 7 extracts the image of the object E from the plurality of generated captured images by the object extraction method described later. The image processing device 7 synthesizes the extracted images of the object E by the image synthesis method described later to generate a synthesized image of the object E.

[0047] (Regarding the imaging device)

[0048] Figure 3 is a top view showing an example of the configuration of the image sensor in the imaging device according to the present embodiment. As Figure 3 shown, the imaging device 1 includes image sensors 101 to 116. The image sensors 101 to 116 each include a plurality of pixels.

[0049] Here, a case where the image processing apparatus 7 generates a composite image of an object based on four images will be described as an example. In the present embodiment, the number of divisions nx in the X direction of one pixel in the composite image is set to 2, and the number of divisions ny in the Y direction is set to 2. In addition, the resolution x in the X direction of each pixel constituting the image sensors 101 to 116 is set to 42.3 νm, and the resolution y in the Y direction is set to 42.3 μm. In addition, the image sensors 101 to 116 are 432 pixels (18273.6 μm) in the X direction and 1 pixel (42.3 μm) in the Y direction, respectively. In order to improve the detection accuracy of the object, it is preferable that the plurality of images to be synthesized are generated by a plurality of image sensors that are displaced in units smaller than one pixel. The number of divisions nx is an index of how much the plurality of image sensors deviate from a reference sensor (for example, the image sensor 101, etc.) in the X direction. For example, when the number of divisions nx is 2, it means that two image sensors deviate by 1 / 2 pixel (0.5 pixel) in the X direction. Similarly, the number of divisions ny is an index of how much the plurality of image sensors deviate from a reference sensor (for example, the image sensor 101, etc.) in the Y direction. For example, when the number of divisions ny is 2, it means that two image sensors deviate by 1 / 2 pixel (0.5 pixel) in the Y direction.

[0050] The image sensor 105 is arranged at a position of a distance y1 in the Y direction with respect to the image sensor 101 as a reference. The distance y1 is y / ny + a × y. a is a coefficient (an integer of 0 or more). The image sensor 105 is arranged with a distance that is a times the resolution y plus 0.5 pixel (y / ny) with respect to the image sensor 101 as a reference. Here, a = 94. Therefore, the distance y1 = 3997.35 μm. The image sensor 105 is arranged at the same position in the X direction with respect to the image sensor 101 as a reference.

[0051] The image sensor 109 is arranged at a position of a distance y2 in the Y direction and a distance x1 in the X direction with respect to the image sensor 101 as a reference. The distance y2 is 0 + b × y. That is, the image sensor 109 is arranged with a distance that is b times the resolution y with respect to the image sensor 101 as a reference. "0" means that the image sensor 109 deviates from the image sensor 101 in the Y direction by an integer multiple of 1 pixel, excluding the fractional part. B is a coefficient (an integer of 0 or more). Here, b = 189. Therefore, the distance y2 = 7994.7 μm. In addition, the distance x1 is x / nx + c × x. c is a coefficient (an integer of 0 or more). Here, c = 0. Therefore, the distance x1 = 21.15 μm.

[0052] The image sensor 113 is arranged at a position that is at a distance of y3 in the Y direction and x1 in the X direction with respect to the image sensor 101. The distance y3 is y / ny + d×y. d is a coefficient (an integer of 0 or more). Here, d = 283. Therefore, the distance y3 = 11992.05 μm.

[0053] With the above structure, the image sensor 105 is arranged to be deviated by 94.5 pixels in the Y direction with respect to the image sensor 101. Ignoring the integer part, the image sensor 105 is arranged to be deviated by 0.5 pixels in the Y direction with respect to the image sensor 101. The image sensor 109 is arranged to be deviated by 0.5 pixels in the X direction with respect to the image sensor 101. The image sensor 113 is arranged to be deviated by 0.5 pixels in the X direction and 283.5 pixels in the Y direction with respect to the image sensor 101. Ignoring the integer part, the image sensor 113 is arranged to be deviated by 0.5 pixels in the X direction and 0.5 pixels in the Y direction with respect to the image sensor 101.

[0054] Similarly, the image sensors 106, 107, and 108 are respectively arranged at positions that are at a distance of y1 in the Y direction with respect to the image sensors 102, 103, and 104. The image sensors 110, 111, and 112 are respectively arranged at positions that are at a distance of y2 in the Y direction and x1 in the X direction with respect to the image sensors 102, 103, and 104. The image sensors 114, 115, and 116 are respectively arranged at positions that are at a distance of y3 in the Y direction and x1 in the X direction with respect to the image sensors 102, 103, and 104.

[0055] In addition, a part of the image sensor 102 overlaps with the image sensor 101 in the X direction. Here, the image sensor 102 overlaps with the image sensor 101 by 423 μm (10 pixels) in the X direction. In addition, in the Y direction, it is arranged at a distance of y4 (21 pixels in this example) with respect to the image sensor 101. Therefore, the image sensor 102 is arranged at a distance of 17850.6 μm (422 pixels) in the X direction and -888.3 μm (21 pixels) in the Y direction with respect to the image sensor 101.

[0056] The image sensor 103 is configured such that, in the X direction, a part thereof overlaps with the image sensor 102. Here, in the X direction, the image sensor 103 overlaps with the image sensor 102 by 423 μm (10 pixels). In addition, in the Y direction, it is arranged at a position of a distance y4 (21 pixels) with respect to the image sensor 102. Therefore, the image sensor 103 is arranged at a distance of 17850.6 μm (422 pixels) in the X direction and a distance of 888.3 μm (21 pixels) in the Y direction with respect to the image sensor 102 as a reference. In addition, in the Y direction, it is arranged at a position of a distance y4 (21 pixels) with respect to the image sensor 102. The image sensor 104 is configured such that, in the X direction, a part thereof overlaps with the image sensor 103. Here, in the X direction, the image sensor 104 overlaps with the image sensor 103 by 423 μm (10 pixels). In addition, in the Y direction, it is arranged at a position of a distance y4 (21 pixels) with respect to the image sensor 103. Therefore, the image sensor 104 is arranged at a distance of 17850.6 μm (422 pixels) in the X direction and a distance of -888.3 μm (21 pixels) in the Y direction with respect to the image sensor 103 as a reference.

[0057] Similarly, the image sensors 106, 110, and 114 are respectively arranged at a distance of 422 pixels in the X direction and a distance of -21 pixels in the Y direction with respect to the image sensors 105, 109, and 113 as references. The image sensors 107, 111, and 115 are respectively arranged at a distance of 422 pixels in the X direction and a distance of 21 pixels in the Y direction with respect to the image sensors 106, 110, and 114 as references. The image sensors 108, 112, and 116 are respectively arranged at a distance of 422 pixels in the X direction and a distance of -21 pixels in the Y direction with respect to the image sensors 107, 111, and 115 as references.

[0058] (Regarding the method for generating the input image)

[0059] Refer to Figures 1 to 6B , and the method for generating an image in the image processing apparatus will be described. Figure 4 is a flowchart for explaining the process of the image generation method in the image processing apparatus, Figure 5A , Figure 5B , Figure 6A and Figure 6B are diagrams for explaining the image generation method in the image processing apparatus. Specifically, Figure 5A is a top view showing the position of the mark on the conveyor belt, Figure 5B is the input image generated based on the signal from the imaging device 1, Figure 6A is the input image shifted in the Y direction, Figure 6B is the input image shifted in the X direction.

[0060] In step S1, the image processing device 7 generates an intake image of the sheet S based on the pixel signals received from the imaging device 1 (pixel signals output from the image sensors 101 to 116) and the amount of movement of the sheet S detected by the rotary encoder 6.

[0061] As Figure 5A shown, on the conveyor belt 8, marks L1 and L2 extending in the X direction are formed. The image processing device 7 receives the pixel signals output from the image sensors 101 to 116 and generates intake images before processing (images P1 to P16). In addition, the images P1 to P16 are images generated based on the respective pixel signals of the image sensors 101 to 116.

[0062] In step S2, the image processing device 7 offsets (corrects) the position in the Y direction of the intake image with reference to the marks L1 and L2. As Figure 5B and Figure 6A shown, the image processing device 7 corrects the position in the Y direction of the images P1 to P16 so that the positions in the Y direction of the marks L1 and L2 coincide.

[0063] In step S3, based on the arrangement of the image sensors 101 to 116, the position in the X direction of the intake image is offset (corrected). As Figure 6A and Figure 6B shown, based on the X - direction arrangement (X - direction overlap) of the image sensors 101 to 116, the position in the X direction of the images P1 to P16 is corrected (offset). For example, the image sensors 101 and 102 overlap each other by 10 pixels in the X direction, so the image processing device 7 overlaps the images P1 and P2 by 10 pixels in the X direction. At this time, the overlapping portion of the images P1 and P2 is set to the brightness level of either one, or to the brightness level obtained by averaging both.

[0064] Here, the image processing device 7 overlaps the images P1 to P4, P5 to P8, P9 to P12, and P13 to P16 by 10 pixels in the X direction respectively.

[0065] In step S4, in the images P1 to P4, P5 to P8, P9 to P12, and P13 to P16 that are overlapped in the X direction, the image processing device 7 sets a specified area with reference to the coordinate origins Oa, Ob, Oc, and Od respectively ( Figure 6BThe region surrounded by the dashed-dotted line is extracted (generated) as the input images Pa, Pb, Pc, and Pd. The coordinate origin Ob is a point that is offset by 1 / ny pixels (here, 0.5 pixels in the Y direction) from the coordinate origin Oa in the Y direction. The coordinate origin Oc is a point that is offset by 1 / nx pixels (here, 0.5 pixels in the X direction) from the coordinate origin Oa in the X direction. The coordinate origin Od is a point that is offset by 1 / nx pixels in the X direction and 1 / ny pixels (here, 0.5 pixels in the Y direction and 0.5 pixels in the X direction) from the coordinate origin Oa in the Y direction.

[0066] Through the above method, the image processing device 7 generates four input images Pa to Pd based on the pixel signals of the image sensors 101 to 116.

[0067] (Method for extracting an image of an object and method for synthesizing an image of an object)

[0068] Refer to Figures 1 to 9B to describe the method for extracting an image of an object and the method for synthesizing an image of an object in the image processing device. Figure 7 is a flowchart for describing the process of the method for extracting an image of an object and the method for synthesizing an image of an object in the image processing device, Figure 8 is a diagram for describing the method for synthesizing an image of an object in the image processing device.

[0069] Next, the image processing device 7 extracts and synthesizes the image of the object E based on the input images Pa to Pd.

[0070] In step S5, the image processing device 7 performs a filtering process on the images Pa to Pd as preprocessing. For example, the image processing device 7 performs a median filtering process, a smoothing filtering process, a comparison difference filtering process, etc. on the images Pa to Pd, which are filtering processes that emphasize the object E (defects and foreign matters) or filtering processes that remove the influence of the substrate.

[0071] In step S6, the image processing device 7 performs a binarization process on the images Pa to Pd.

[0072] In step S7, the image processing device 7 extracts the image of the object E in images Pa to Pd based on the first feature quantity of the object E. For example, as the first feature quantity of the object E, the area of the object E, the maximum length, the aspect ratio, the vertical width, the horizontal width, the binarization threshold (brightness threshold), the brightness level (maximum value, minimum value, average value, etc.), the Feret diameter (maximum value, minimum value, etc.), the length of the major axis (maximum value, minimum value, etc.), etc. are used. In addition, the image processing device 7 may use the image data of the sheet S that does not contain the object as the first feature quantity, and extract the image of the object E by comparing this image data with images Pa to Pd. In addition, when the lighting device 2 can irradiate light in a certain direction, the image processing device 7 may use the light component in the specified direction as the first feature quantity, and extract the object E when the light component in the specified direction exceeds the specified value.

[0073] When the image processing device 7 includes the object E in a specified number of pixels or more, the image of the object E is extracted. Here, when the image processing device 7 includes the object E in 2 pixels or more, the image of the object E is extracted.

[0074] In step S8, the image processing device 7 associates (establishes an association) the extracted images of the object E based on the second feature quantity (feature quantity) of the object E. As the second feature quantity of the object E, the area of the object E, the aspect ratio, the center of gravity (area center of gravity, brightness value of the center of gravity, etc.), the center or center of gravity of the rectangle circumscribing the image of the object E, the brightness level (maximum value, minimum value, average value, etc.), the Feret diameter (maximum value, minimum value, etc.), the length of the major axis (maximum value, minimum value, etc.), etc. are used. Here, the image processing device 7 associates the images of the object E based on the center of gravity (area center of gravity) of the image of the object E.

[0075] As Figure 8 shown, the image processing device 7 extracts objects E1 to E4 from image Pa, extracts objects E5 to E8 from image Pb, extracts objects E9 to E12 from image Pc, and extracts objects E13 to E15 from image Pd. In this case, the image processing device 7 associates the images of the objects E1 to E15 based on the respective centers of gravity G1 to G15 of the objects E1 to E15. Specifically, when the images Pa to Pd are overlapped, the image processing device 7 associates the images of the object E whose centers of gravity are in the corresponding positions.

[0076] Here, when the center-of-gravity position of the object E is within a specified range, the image processing device 7 correlates the images of the object E. The specified range is preferably from 5 pixels (0.1 mm) to 50 pixels (1.0 mm), and here, it is set to 16 pixels (0.32 mm). However, it is not limited thereto, and the specified range can also be set larger according to the distance setting between the image sensors 101 to 116.

[0077] First, in Figure 8 the image processing device 7 correlates the image of the object E in the image Pa with the images of the object E in the images Pb, Pc, and Pd. Specifically, since the centers of gravity G5, G9, and G13 are within the specified range of the center of gravity G1, the image processing device 7 correlates the images of the objects E1, E5, E9, and E13. Since the centers of gravity G6 and G10 are within the specified range of the center of gravity G2, the image processing device 7 correlates the images of the objects E2, E6, and E10. Since the center of gravity G7 is within the specified range of the center of gravity G3, the image processing device 7 correlates the images of the objects E3 and E7.

[0078] Next, the image processing device 7 correlates the image of the object in the image Pb with the images of the objects in the images Pc and Pd. Since the centers of gravity G11 and G15 are within the specified range of the center of gravity G8, the image processing device 7 correlates the images of the objects E8, E11, and E15.

[0079] Next, the image processing device 7 correlates the image of the object in the image Pc with the image of the object in the image Pd. In Figure 8 since, in the images Pc and Pd, there is no uncorrelated object E among the images of the object E corresponding to the centers of gravity, the image processing device 7 does not correlate the images of the object E.

[0080] In step S9, the image processing device 7 synthesizes the images of the object E based on the third feature quantity. As the third feature quantity of the object E, there are the area, aspect ratio, center of gravity of the image of the object E, and the center of gravity of the rectangle circumscribing the image of the object E, etc. Here, the image processing device 7 synthesizes the images of the object E based on the center of gravity of the image of the object E.

[0081] Specifically, the image processing device 7 doubles the image size of the extracted object E in the X direction and the Y direction respectively. And the image processing device 7 overlaps the correlated images of the object E with the center of gravity (the third feature quantity) of the image of the object E as the center, adds the luminance levels of each pixel, and performs averaging.

[0082] For example, in Figure 8In this case, since the images of the objects E1, E5, E9, and E13 are associated, the image processing device 7 generates an image of the object E based on four images.

[0083] In step S10, the image processing device 7 outputs the generated image of the object E (for example, displayed on a display (not shown)). At this time, the image processing device 7 outputs the synthesized image of the object E and also outputs the image of the object E that is not associated. In Figure 8 this case, as the synthesized image of the object E, the image processing device 7 outputs the synthesized images of the objects E1, E5, E9, and E13, the synthesized images of the objects E2, E6, and E10, and the synthesized images of the objects E8, E11, and E15. As the images of the object E that are not associated, it outputs the image of the object E4, the image of the object E12, and the image of the object E14. The image processing device 7 performs image synthesis processing on the images of a plurality of associated objects to generate a synthesized image and outputs the synthesized image. On the other hand, the image processing device 7 skips the image synthesis processing on the image of a single, unassociated object and directly outputs the image. For example, the image of the object E4 is not synthesized with the images of the corresponding regions of the images Pb to Pd (the base image where the object does not exist) and is directly output. The image quality of the image of the object E4 is higher than the image quality of the synthesized image of the image of the object E4 and the base image. By directly outputting the image of the object E4, the detection accuracy of the object E4 can be improved.

[0084] Figure 9A And Figure 9B are diagrams for comparing the synthesized images of the objects output by the inspection device according to the present embodiment and the existing inspection device. Specifically, Figure 9A is the synthesized image of the object output by the inspection device according to the present embodiment, Figure 9B is the synthesized image of the object output by the existing inspection device.

[0085] In the existing inspection device, the image processing device 7 generates the input images Pa to Pd by the above-described method of generating input images. The image processing device 7 corrects the positions in the X direction and Y direction of the input images Pa to Pd based on the configurations of the image sensors 101 to 116, synthesizes the input images Pa to Pd, and extracts the image of the object E.

[0086] As Figure 9A And Figure 9BAs shown, compared with the existing inspection device, the image of the object E output by the inspection device according to this embodiment is clearer. In the existing inspection device, offset (correction) based on the configuration of the image sensor is performed, but the irradiation method of light on the object to be inspected is not considered. Therefore, the positions of the images of the object E in each captured image to be synthesized deviate, and the synthesized image of the object E becomes unclear. In contrast, in this embodiment, after extracting the image of the object E for each captured image, the extracted images of the object E are synthesized with the center of gravity of the object E as the center. Thereby, the deviation of the position of the synthesized image of the object E can be suppressed, and thus the synthesized image of the object E can be made clear.

[0087] With the above structure, this inspection device inspects the sheet S by detecting the size of the object E in the sheet S. This inspection device includes: image sensors 101 to 116 that capture images of the sheet S, an illumination device 2 that irradiates light on the sheet S, and an image processing device 7 that generates an image of the object E based on the outputs of the image sensors 101 to 116. The image processing device 7 generates captured images Pa to Pd that commonly include an image of a specified range of the sheet S based on the outputs from the image sensors 101 to 116. The image processing device 7 extracts the image of the object E in the sheet S for each generated captured image Pa to Pd. The image processing device 7 correlates (establishes an association) the extracted images of the object E with each other based on the center of gravity (feature amount) of the object E manifested in the image. Then, the image processing device 7 synthesizes the associated images of the object E with each other.

[0088] According to one aspect of the present disclosure, as described below, in an inspection device and an inspection method using a plurality of image sensors, the size of an object in an object to be inspected can be accurately detected, and thus the detection reproducibility and detection probability of the object in the object to be inspected can be improved.

[0089] When synthesizing the captured images Pa to Pd based on the configuration of the image sensors 101 to 116, it is feared that due to the irradiation method of light on the sheet S, the position of the object E in each captured image Pa to Pd deviates greatly. Therefore, it is feared that by correcting the captured images Pa to Pd based on the positions of the image sensors, the deviation of the position of the object E cannot be corrected and the object E cannot be detected. In contrast, in this inspection device, the image of the object E is extracted for each captured image Pa to Pd, and the extracted images of the object E are associated and synthesized with each other based on the feature amount. Thereby, the deviation of the position of the object E in each captured image Pa to Pd can be suppressed, and thus the synthesis of the image of the object E can be accurately performed, and the size of the object in the object to be inspected can be accurately detected. Therefore, the detection reproducibility and detection probability of the object (foreign matter or defect) in the object to be inspected (sheet S) can be improved.

[0090] In addition, the image processing device 7 does not synthesize and output the image of the unassociated (not established association) object E with the images of other objects E. Thereby, the detection reproducibility and detection probability of the object in the subject (sheet S) can be improved.

[0091] In addition, the image sensors 101 to 116 each set the resolution in the X direction to x and the resolution in the Y direction to y. In this case, the image sensors 101 to 104, 105 to 108, 109 to 112, and 113 to 116 are respectively arranged and configured in the X direction at every (1 / nx) + mx pixels. The image sensors 101, 105, 109, and 113 are arranged and configured in the Y direction at every (1 / ny) + py pixels. The image sensors 102, 106, 110, and 114 are arranged and configured in the Y direction at every (1 / ny) + py pixels. The image sensors 103, 107, 111, and 115 are arranged and configured in the Y direction at every (1 / ny) + py pixels. The image sensors 104, 108, 112, and 116 are arranged and configured in the Y direction at every (1 / ny) + py pixels. Here, nx is the number of divisions of pixels in the X direction, ny is the number of divisions of pixels in the Y direction, and m and p are integer coefficients. Thereby, one pixel is divided into the number of divisions nx in the X direction and the number of divisions ny in the Y direction, so that the resolution of the imaging device in the X direction and the Y direction can be improved.

[0092] (Other embodiments)

[0093] As described above, as an example of the technology disclosed in the present application, the embodiments have been described. However, the technology in the present disclosure is not limited thereto, and can also be applied to embodiments that have been appropriately changed, replaced, added, omitted, etc.

[0094] In addition, in the above embodiment, the imaging device 1 and the illumination device 2 are constituted by a dark field optical system, and may also be constituted by a bright field optical system. In addition, the imaging device 1 is configured as a line sensor, but may also be configured as an area sensor. In addition, the image processing device 7 may generate a moving image based on the pixel signals output from the image sensors 101 to 116, or may generate a still image.

[0095] Figure 10 It is a diagram for explaining the types of sensors, the types of moving images, and the relationship of the optical systems in the inspection device according to the present embodiment. Representing that the imaging error is above the set resolution as A, the imaging error being less than the set resolution to more than one-tenth of the set resolution as B, and the imaging error being less than one-tenth of the set resolution as C. As Figure 10As shown, when the imaging device 1 and the illumination device 2 are configured by a dark-field optical system, the inspection device can generate the clearest composite image of the object E. This is because by illuminating at a low angle, the base of the object (ground level) is not irradiated by the illumination light (the brightness of the base at the position where there is no foreign object is at a low level), and the object is irradiated by the light, so the SN (signal-to-noise (brightness of foreign object / brightness of base)) ratio is increased. In addition, even if the illumination device 2 is configured by a bright-field optical system, the same effect as that of this inspection device can be obtained, and the same effect can be obtained even when generating a moving image from the image sensor.

[0096] In addition, the image sensor of the imaging device 1 is not limited to the above configuration. In addition, the number of image sensors of the imaging device 1 is not limited to the above number.

[0097] In addition, the number of pixels constituting each image sensor is not limited to the above number of pixels.

[0098] In addition, the number of input images generated by the image processing device 7 is not limited to 4. The image processing device 7 only needs to generate at least 2 input images.

[0099] In addition, the object to be inspected by this inspection device is not limited to being configured in a sheet shape.

[0100] Industrial Applicability

[0101] The inspection device of the present invention can be used for the inspection of foreign objects, defects, etc. contained in components used in semiconductors, electronic devices, secondary batteries, etc.

Claims

1. An inspection device for inspecting an object to be inspected by detecting the size of an object in the object to be inspected, the inspection device comprising: A plurality of image sensors for photographing the object to be inspected; An illumination device for irradiating light onto the object to be inspected; and An image processing device for generating an image of the object based on the outputs of the plurality of image sensors, The image processing device generates a plurality of input images that commonly include an image of a specified range of the object to be inspected based on the outputs of the plurality of image sensors, The image processing device extracts an image of the object in the object to be inspected for each of the generated plurality of input images, The image processing device establishes an association between the extracted images of the object based on the feature amount of the object manifested in the image, The image processing device synthesizes the associated images of the object with each other, When there is only one image of the object extracted, the image processing device also outputs the one image of the object without associating it with other images.

2. The inspection device according to claim 1, wherein The feature amount includes any one of the area, aspect ratio, center of gravity, center and center of gravity of a rectangle circumscribing the object, brightness level, Feret diameter, and length of the major axis of the object manifested in the image.

3. The inspection device according to claim 1 or 2, wherein The plurality of image sensors include: A first image sensor; and A second image sensor configured to be separated by a first distance in the scanning direction, which is the direction in which the first image sensor scans the object to be inspected, and to be separated by a second distance in the sub-scanning direction, which is a direction perpendicular to the scanning direction, from the first image sensor, Wherein, the first distance is represented by (x / nx)+m×x, the second distance is represented by (y / ny)+p×y, x represents the resolution in the scanning direction, y represents the resolution in the sub-scanning direction, nx is the number of pixel divisions in the scanning direction, ny is the number of pixel divisions in the sub-scanning direction, and m, p are integer coefficients.

4. The inspection device according to claim 1 or 2, wherein The inspection device comprises: A conveyor belt for conveying the object to be inspected; A first drive mechanism for moving the conveyor belt in a sub-scanning direction perpendicular to the scanning direction in which the plurality of image sensors scan the object to be inspected; And A second drive mechanism for reciprocally moving the conveyor belt in the scanning direction while the first drive mechanism moves the conveyor belt in the sub-scanning direction.

5. The inspection device according to claim 4, wherein When the object to be inspected is conveyed in the sub-scanning direction by the conveyor belt, the plurality of image sensors scan the object to be inspected.

6. The inspection device according to claim 1 or 2, wherein The image sensor and the illumination device are composed of a dark field optical system.

7. An inspection method for inspecting an object to be inspected by detecting the dimensions of an object in the object to be inspected using an inspection device, the inspection device comprising: a plurality of image sensors for photographing the object to be inspected; a lighting device for irradiating light onto the object to be inspected; and an image processing device for generating an image of the object in the object to be inspected based on the outputs of the plurality of image sensors, The inspection method includes: a step of generating a plurality of input images that commonly include an image of a specified range of the object to be inspected based on the outputs of the plurality of image sensors; a step of extracting an image of the object in the object to be inspected for each of the generated plurality of input images; a step of establishing an association between the images of the object extracted from each other based on the feature amounts of the object appearing in the images; and a step of synthesizing the images of the object whose association has been established with each other, The inspection method includes: in the case where there is only one image of the object extracted, also outputting the one image of the object without establishing an association with other images.

Images