JP2019138639A - Defect detection device and defect detection method - Google Patents

Abstract

To provide a defect detection device and a defect detection method which can have an improved accuracy of detection of defects such as a concave and a convex which have a relatively gently curving surface shape while preventing decrease in working efficiency of inspection.SOLUTION: A defect detection device 10 detects a defect on a surface of an object to be inspected 1. The defect detection device 10 comprises: an illumination device 4 which can illuminate the surface from one side of the surface and the other side opposite to the one side with the surface being in between; a photographing device 3 which images the surface illuminated by the illumination device 4; and an image processing section 11 which acquires a pair of images generated by imaging the surface illuminated from the one side and the surface illuminated from the other side, respectively. The image processing section 11 creates a defect detection image which is an image for detection of the defect on the basis of the pair of images.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a defect detection apparatus and a defect detection method for detecting defects on the surface of an object.

  The defect detection apparatus is an apparatus that images the surface of an object and extracts surface defects based on the captured image. This defect detection apparatus includes an illumination unit that illuminates the surface of an object, an imaging unit that captures an image of the surface illuminated by the illumination unit, and a process that generates an image for detecting defects based on an image captured by the imaging unit. And a section. In addition, such a defect detection apparatus illuminates the surface of an object from a plurality of directions, and represents a representative value for each pixel (for example, the maximum value of the gradation minus the minimum value) from a plurality of images obtained in the respective illumination directions. Value) and generating an inspection image based on the representative value is known (see Patent Document 1). Thereby, the damage | wound etc. of the surface of a target object are detectable.

JP 2017-156212 A

  By the way, in the inspection of the surface of an object, there is a desire to detect a dent or a bulge whose surface shape changes gently without detecting a groove-like scratch or the like whose surface shape changes rapidly. For example, in a terminal of an electrical product, a long and narrow scratch may be a good product, and a dent or bulge may be a defective product. In such a case, when an inspection is performed using the apparatus disclosed in Patent Document 1, there is a possibility that even a scratch that can be a good product is detected as a defective product. As a result, re-inspection, visual inspection, etc. were required, and the workability of the inspection was deteriorated.

  The present invention has been made in view of such circumstances, and its purpose is to reduce the workability of inspection while improving the detection accuracy for defects such as dents and bulges whose surface shape changes relatively slowly. It is an object of the present invention to provide a defect detection apparatus and a defect detection method that can suppress the above.

The defect detection apparatus of the present invention is proposed to achieve the above object, and is a defect detection apparatus for detecting defects on the surface of an object,
An illumination unit capable of illuminating the surface from one side of the surface and the other side opposite to the one side with the surface in between;
An imaging unit that images the surface illuminated by the illumination unit;
A processing unit for capturing a pair of images by capturing the surface illuminated from the one side and the surface illuminated from the other side,
The processing unit generates a defect detection image that is an image for detecting a defect based on the pair of images.

  According to this configuration, it is possible to improve detection accuracy for defects such as dents and bulges whose surface shape changes are relatively gradual, and to suppress detection of defects for scratches whose surface shape changes are relatively abrupt. . This eliminates the need for re-inspection and visual inspection after the inspection, improving the workability of the inspection.

Further, in the above configuration, the illumination unit can illuminate the surface from the one side and the other side in a first direction and a second direction intersecting the first direction,
The processing unit acquires the pair of images in the first direction and the second direction, respectively, and converts the pair of images in the first direction and the pair of images in the second direction. Based on this, it is desirable to generate the defect detection image.

  According to this configuration, it is possible to further improve the detection accuracy with respect to defects such as dents and bulges whose surface shape changes relatively slowly.

  Furthermore, in any one of the above-described configurations, it is preferable that the processing unit compares the pair of images for each pixel to obtain a representative value, and generates the defect detection image based on the representative value.

  In the above configuration, it is preferable that the processing unit compares the pair of images for each pixel and acquires a maximum value and a minimum value as the representative value.

  Furthermore, in any one of the above configurations, the processing unit compares the pair of images for each pixel to obtain a maximum value and a minimum value, and obtains a difference between the maximum value and the minimum value for each pixel. It is desirable to generate the defect detection image based on the difference.

  According to these configurations, it is possible to improve the detection accuracy for defects such as dents and bulges whose surface shape changes are relatively gradual, and to suppress detection of defects for scratches whose surface shape changes are relatively abrupt. it can.

In any one of the above configurations, the illumination unit can illuminate the surface from the one side and the other side in a plurality of directions.
The processing unit preferably acquires the pair of images in the plurality of directions, and generates a defect detection image that is an image for detecting a defect based on the pair of images in the plurality of directions.

  According to this configuration, it is possible to further improve the detection accuracy for a defect such as a dent or bulge whose surface shape changes relatively slowly.

And the defect detection method of the present invention is a defect detection method for detecting defects on the surface of an object,
A pair of images is obtained by imaging each of the surface illuminated from one side and the surface illuminated from the other side opposite to the one side across the surface,
A defect detection image that is an image for detecting a defect is generated based on the pair of images.

  According to this method, it is possible to improve detection accuracy for defects such as dents and bulges whose surface shape changes are relatively slow, and to suppress detection of defects for scratches whose surface shape changes are relatively abrupt. . This eliminates the need for re-inspection and visual inspection after the inspection, improving the workability of the inspection.

It is a schematic diagram explaining the structure of a defect detection apparatus. It is a flowchart explaining operation | movement of a defect detection apparatus. It is an example of the captured image of the surface of the to-be-inspected object illuminated with the 1st lighting fixture. It is an example of the captured image of the surface of the to-be-inspected object illuminated by the 2nd lighting fixture. It is an example of the captured image of the surface of the to-be-inspected object illuminated with the 3rd lighting fixture. It is an example of the captured image of the surface of the to-be-inspected object illuminated with the 4th lighting fixture. 5 is an example of an image obtained by combining the maximum values of FIG. 3 and FIG. 4. 5 is an example of an image obtained by combining the minimum values of FIG. 3 and FIG. 11 is an example of an image obtained by combining the maximum values of FIG. 9 and FIG. 11 is an example of an image obtained by combining the minimum values of FIG. 9 and FIG. 10. It is an example of the image obtained by the difference of FIG. 7 and FIG. It is an example of the image obtained by the difference of FIG. 8 and FIG. 13 is an example of an image obtained by combining the maximum values of FIG. 11 and FIG. It is an example of the image obtained by binarizing FIG. It is a flowchart explaining operation | movement of the conventional defect detection apparatus. 7 is an example of an image obtained by combining the maximum values of FIGS. 7 is an example of an image obtained by combining the minimum values of FIGS. It is an example of the image obtained by the difference of FIG. 16 and FIG. It is an example of the image obtained by binarizing FIG.

  DESCRIPTION OF EMBODIMENTS Hereinafter, embodiments for carrying out the present invention will be described with reference to the accompanying drawings. In the embodiments described below, various limitations are made as preferred specific examples of the present invention. However, the scope of the present invention is not limited to the following description unless otherwise specified. However, the present invention is not limited to these embodiments.

  FIG. 1 is a schematic diagram showing the configuration of the defect detection apparatus 10. The defect detection apparatus 10 in this embodiment includes a stage 2, an imaging apparatus 3 (corresponding to an imaging unit in the present invention), an illumination device 4 (corresponding to an illumination unit in the present invention), a computer 5, a display device 6, and an input device. 7 and the like, and detects defects on the surface of the inspection object 1 (one kind of object) such as a semiconductor wafer, a metal workpiece, a circuit board, and other electronic components by image processing. In particular, the defect detection apparatus 10 in the present embodiment detects defects such as dents and bulges whose surface shape changes relatively slowly as defective products, and detects defects such as scratches whose surface shape changes relatively rapidly as defective products. It is configured not to detect as. The defect detection method will be described later.

  The imaging apparatus 3 includes an optical system 8 including a lens or a microscope having a magnification sufficient to optically expand the inspection object 1 and detect a defect, and the inspection object 1 enlarged by the optical system 8. CCD camera 9 etc. which receive the image of this and convert it into an electric signal are provided. The imaging device 3 images the surface of the inspection object 1 on the stage 2 (hereinafter also referred to as imaging), converts it into image data, and outputs the image data to the computer 5. The illumination device 4 is a light source that illuminates the surface of the inspection object 1 placed on the stage 2 (specifically, the surface (that is, the upper surface) on the imaging device 3 side). The illuminating device 4 in this embodiment is comprised by four lighting fixtures 4a-4d which consist of a lamp | ramp, LED, etc., for example. Each lighting fixture 4a-4d is arrange | positioned with respect to the said stage 2 so that the stage 2 (or the surface of the to-be-inspected object 1) may be enclosed.

  In the following description, the lighting fixture arranged on one side (upper side in FIG. 1) in the y direction (corresponding to the first direction in the present invention) with respect to the stage 2 is the first lighting fixture 4a, the stage. 2 is a lighting device arranged on the other side in the y direction (on the opposite side to the one side with the surface of the inspected object 1 in between, the lower side in FIG. 1) as a second lighting device 4b and a stage. 3 is a lighting fixture arranged on one side (left side in FIG. 1) in the x direction (corresponding to the second direction in the present invention) intersecting (in the present embodiment, orthogonal) with respect to 2 in the y direction. The lighting fixture 4c is arranged on the other side in the x direction with respect to the stage 2 (on the opposite side to the one side with the surface of the inspection object 1 in between, the right side in FIG. 1). This is called a lighting fixture 4d. And the 1st lighting fixture 4a and the 2nd lighting fixture 4b which become a pair facing each other illuminate the surface of the to-be-inspected object 1 from the one side and the other side in ay direction. Moreover, the 3rd lighting fixture 4c and 4th lighting fixture 4d which become a pair facing each other illuminate the surface of the to-be-inspected object 1 from the one side and other side in a x direction.

  Each lighting fixture 4a-4d in this embodiment has a rectangular-shaped light emission surface, and illuminates the surface on the stage 2 (namely, surface of the to-be-inspected object 1) from the inclined low angle. For example, the angle (elevation angle) of the optical axis from each of the lighting fixtures 4a to 4d with respect to the surface of the inspection object 1 is set to 30 degrees or less. In addition, the light emitted from each lighting fixture 4a-4d is not restricted to visible light, An ultraviolet-ray, infrared rays, etc. may be sufficient. In addition, the light may be light having a single wavelength or light in which a plurality of wavelengths are mixed. In short, the light from each of the lighting fixtures 4a to 4d may be light of any wavelength as long as it can be reflected by the surface of the inspection object 1 and captured by the CCD camera 9. In FIG. 1, the first lighting fixture 4a and the second lighting fixture 4b that are paired are arranged in line symmetry (that is, the phases are shifted by 180 degrees), and similarly, the third lighting fixture 4a is paired. Although the lighting fixture 4c and the 4th lighting fixture 4d are arrange | positioned axisymmetrically (namely, a phase shift | offset | differs 180 degree | times), it is not restricted to these arrangement | positioning, The phase of each lighting fixture 4a-4d used as a pair is It may be arranged slightly deviated from 180 degrees. In short, the light emitting surface of the paired first lighting fixture 4a and the light emitting surface of the second lighting fixture 4b are opposed to each other across the surface of the inspected object 1, and the light emitting surface of the third lighting fixture 4c that is paired with each other. And the light emission surface of the 4th lighting fixture 4d should just oppose on both sides of the surface of the to-be-inspected object 1. FIG.

  The computer 5 is an image processing apparatus that controls the imaging apparatus 3 and performs a process (filtering process or the like) for detecting a defect of the inspection object 1 based on image data obtained from the imaging apparatus 3. The computer 5 in this embodiment includes an image processing unit 11 (corresponding to a processing unit in the present invention) and a storage unit 12. The storage unit 12 is a device that stores predetermined data composed of, for example, a hard disk drive. The storage unit 12 stores an operation system, various application programs, image processing data, and the like. The image processing unit 11 includes a CPU, a ROM, a RAM, and the like (not shown), and performs various types of processing in addition to image processing such as defect detection processing according to the operation system stored in the storage unit 12. In addition, the image processing unit 11 performs processing of inputting captured image data (hereinafter simply referred to as a captured image) of the inspection object 1 captured by the imaging device 3 and storing the captured image in the storage unit 12. Furthermore, the image processing unit 11 performs processing for detecting (extracting) a predetermined defect by performing image processing to be described later on the captured image. The display device 6 connected to the computer 5 is a display device such as a liquid crystal display. The input device 7 includes a keyboard and a mouse, for example, and outputs an operation signal to the computer 5 in response to a user operation.

  Next, the operation (that is, the defect detection method) of the defect detection apparatus 10 in the present embodiment will be described. FIG. 2 is a flowchart for explaining the flow of the defect detection process. 3 to 14 are examples of images of the surface of the inspection object 1 obtained in the course of the defect detection process. In this embodiment, the defect of the terminal provided in the electronic device which is the to-be-inspected object 1 is inspected. That is, an image of a region where the terminal of the electronic device is formed is imaged, and the presence or absence of a terminal defect is inspected by image processing. 3-14, four rectangular figures arranged in two rows are terminals.

  First, when the inspection object 1 is set on the stage 2, the lighting fixtures 4a to 4d are sequentially turned on, and the surface of the inspection subject 1 illuminated by the respective lighting fixtures 4a to 4d is photographed by the photographing device 3 ( Image). In other words, the surface of the inspection object 1 is illuminated from one side and the other side in the x direction, and from one side and the other side in the y direction, respectively, and the imaging device 3 captures images. FIG. 3 is a captured image of the surface of the inspection object 1 illuminated by the first lighting fixture 4a, in other words, a captured image of the surface of the inspection object 1 illuminated from one side in the y direction. FIG. 4 is a captured image of the surface of the inspection object 1 illuminated by the second lighting fixture 4b, in other words, a captured image of the surface of the inspection object 1 illuminated from the other side in the y direction. FIG. 5 is a captured image of the surface of the inspection object 1 illuminated by the third lighting fixture 4c, in other words, a captured image of the surface of the inspection object 1 illuminated from one side in the x direction. FIG. 6 is a captured image of the surface of the inspection object 1 illuminated by the fourth lighting fixture 4d, in other words, a captured image of the surface of the inspection object 1 illuminated from the other side in the x direction.

  As can be seen from the captured images shown in FIGS. 3 to 6, the upper left terminal has a bulge in which the change in surface shape is relatively gentle. In this embodiment, this bulge should be detected as a defect and determined to be defective. On the other hand, the lower right terminal has a scratch extending left and right. This flaw changes the surface shape relatively abruptly, and in this embodiment, it should be determined as a good product without being detected as a defect. Such a flaw is imaged by illumination from the direction intersecting the extending direction (in this embodiment, illumination from the first illumination device 4a or illumination from the second illumination device 4b). Brightly emphasized in the image (see FIGS. 3 and 4). Further, such a flaw is caused by illumination from a direction along the extending direction in the captured image (in this embodiment, illumination from the third illumination fixture 4c or illumination from the fourth illumination fixture 4d). ), The captured image becomes dark (see FIGS. 5 and 6).

  Here, the captured image of the surface of the inspection object 1 illuminated by the first illumination fixture 4a and the captured image of the surface of the inspection object 1 illuminated by the second illumination fixture 4b are used in the subsequent image processing. Processed as a pair of images in the y direction. Further, the captured image of the surface of the inspection object 1 illuminated by the third lighting fixture 4c and the captured image of the surface of the inspection subject 1 illuminated by the fourth lighting fixture 4d are x in the subsequent image processing. Processed as a pair of images in the direction. These images are converted into image data, respectively, and taken into the image processing unit 11 in the computer 5 (step S1). At this time, each captured image is taken into the image processing unit 11 as digital data of a predetermined gradation (for example, 4096 gradations (12 bits)) by an A / D converter (not shown) and temporarily stored in the storage unit 12.

  Next, maximum value synthesis and minimum value synthesis are performed from each pair of images (step S2). Specifically, in a pair of images, the tone values are compared for each corresponding pixel, and the maximum value is acquired as a representative value (maximum value synthesis). Similarly, in a pair of images, the gradation values are compared for each corresponding pixel, and the minimum value is acquired as a representative value (minimum value synthesis). FIG. 7 is an image obtained by combining the images of FIG. 3 and the image of FIG. 4 with the maximum value, that is, a maximum value combined image obtained from a pair of images in the y direction. FIG. 8 is an image obtained by synthesizing the minimum values of the image of FIG. 3 and the image of FIG. 4, that is, a minimum value synthesized image obtained from a pair of images in the y direction. FIG. 9 is an image obtained by combining the images of FIG. 5 and the image of FIG. 6 with the maximum value, that is, a maximum value combined image obtained from a pair of images in the x direction. FIG. 10 is an image obtained by synthesizing the minimum values of the image of FIG. 5 and the image of FIG. 6, that is, a minimum value synthesized image obtained from a pair of images in the x direction.

  Next, the difference between the maximum value and the minimum value obtained in step S2 is acquired for each pixel (step S3). Specifically, in the pair of images in the y direction, the difference between the maximum value and the minimum value obtained in step S2 is acquired. Similarly, in a pair of images in the x direction, the difference between the maximum value and the minimum value obtained in step S2 is acquired. FIG. 11 is an image obtained by the difference between the image in FIG. 7 and the image in FIG. 8, that is, a difference image between a maximum value synthesized image and a minimum value synthesized image obtained from a pair of images in the y direction. FIG. 12 is an image obtained by the difference between the image of FIG. 9 and the image of FIG. 10, that is, a difference image between the maximum value synthesized image and the minimum value synthesized image obtained from a pair of images in the x direction. Here, by performing such processing, as shown in the image of FIG. 11 and the image of FIG. 12, scratches and the like can hardly be recognized. That is, by acquiring the difference between the maximum value and the minimum value in a pair of images, it detects a bulge or the like in which the surface shape changes relatively slowly without detecting a scratch or the like in which the surface shape changes relatively rapidly. be able to. In short, a flaw or the like whose surface shape changes relatively abruptly appears in the same way in illumination from one side and the other side in one direction, and is thus not detected by performing the above processing. On the other hand, a bulge, a dent, or the like whose surface shape changes relatively gently appears differently in illumination from one side and the other side in one direction, and can be detected even if the above processing is performed.

  Next, the maximum value synthesis is performed using the difference between the maximum value and the minimum value obtained from the pair of images in the y direction and the difference between the maximum value and the minimum value obtained from the pair of images in the x direction. Perform (step S4). That is, in the difference image in the y direction and the difference image in the x direction, the gradation values are compared for each corresponding pixel, and the maximum value is acquired as a representative value. FIG. 13 is a maximum value synthesized image obtained by synthesizing the maximum value of the image of FIG. 11 and the image of FIG. Thereafter, binarization processing is performed (step S5). Specifically, it is determined whether or not the gradation value exceeds a predetermined threshold value for each pixel, and if it is below, it is converted to black, and if it exceeds, it is converted to white. FIG. 14 is a defect detection image obtained by binarizing the maximum value composite image of FIG. Thus, by binarizing, noise and the like can be removed, and the defect becomes clear. Finally, defect detection is performed using the defect detection image obtained in this way (step S6), and the defect detection process is terminated. In addition, as a method of determining a defect from a defect detection image, for example, determination is performed by comparing the area of a defective portion (that is, a white portion) with a predetermined value (determination criterion). Or the ratio of the area of the defective portion to the area of the terminal is determined by comparing with a predetermined value (determination criterion). Further, the defect detection image displayed on the display device 6 can be visually confirmed and judged.

  Next, as a comparison, a conventional defect detection method will be described. FIG. 15 is a flowchart for explaining the flow of conventional defect detection processing. Also in the conventional defect detection processing, first, each of the lighting fixtures 4a to 4d is sequentially turned on, and the surface of the inspection object 1 illuminated by each of the lighting fixtures 4a to 4d is photographed (captured). The captured images are converted into image data, respectively, and taken into the image processing unit 11 in the computer 5 (step S11). Next, maximum value synthesis and minimum value synthesis are performed from all images (step S12). FIG. 16 is an image obtained by combining the maximum values using the four images of FIGS. 3 to 6. FIG. 17 is an image obtained by combining the minimum values using the four images of FIGS. Thereafter, the difference between the maximum value and the minimum value obtained in step S12 is acquired for each pixel (step S13). That is, the difference between the image obtained by combining the maximum values and the image obtained by combining the minimum values is acquired for each pixel. FIG. 18 is an image obtained by the difference between the image of FIG. 16 and the image of FIG. The binarization process is performed on the image (step S14), the defect is detected (step S15), and the defect detection process is terminated. Note that FIG. 19 is an image obtained by binarizing the image of FIG.

  In the conventional defect detection process, as shown in the image of FIG. 19, not only the swelling at the upper left terminal, but also the flaw at the lower right terminal was detected. On the other hand, as shown in the image of FIG. 14, the present invention can detect only the bulge in the upper left terminal without detecting the scratch in the lower right terminal. In short, in the present invention, it is possible to improve detection accuracy for defects such as dents and bulges whose surface shape changes are relatively gradual, and to suppress detection of defects for scratches whose surface shape changes are relatively abrupt. it can. This eliminates the need for re-inspection and visual inspection after the inspection, improving the workability of the inspection. In the present embodiment, since the defect detection image is generated using both the pair of images in the y direction and the pair of images in the x direction, the surface shape changes relatively slowly, such as a dent or a bulge. The detection accuracy for the defect can be further improved.

  By the way, in above-mentioned embodiment, the illuminating device 4 is equipped with the four lighting fixtures 4a-4d, and performs four images (in other words, two pairs of images) by performing illumination using each lighting fixture 4a-4d. The defect detection image is generated based on these four images. However, the present invention is not limited to this. The lighting device may further include a plurality of lighting fixtures. In short, the illumination device is configured to be able to illuminate the surface of the object to be inspected from one side and the other side in a plurality of directions, and obtains a pair of images in the plurality of directions, respectively, based on the pair of images in the plurality of directions. Thus, a defect detection image that is an image for detecting a defect may be generated. For example, based on a pair of images in a plurality of directions, difference images can be obtained in the same manner as in steps S1 to S5 described above, and a defect detection image can be generated by combining the maximum values using all the difference images. . In this way, it is possible to further improve the detection accuracy for defects such as dents and bulges whose surface shape changes relatively slowly.

  Alternatively, the lighting device may include only one lighting fixture, and the lighting fixture may be configured to go around the stage. Even in this case, the illumination device can illuminate the surface of the inspection object from one side and the other side in a plurality of directions. Further, it is possible to adopt a configuration in which the lighting device includes only one lighting fixture and the stage rotates. Even in this case, by rotating the stage, the illumination device can illuminate the surface of the inspection object from one side and the other side in a plurality of directions. Note that when the stage is rotated, the captured image is also rotated in accordance with this rotation, and therefore image processing is performed to rotate the captured image in accordance with the rotation of the stage.

  The present invention can be widely applied to detection of defects in MEMS devices, semiconductor devices, other electronic components, etc. in addition to detection of defects in terminals such as circuit boards.

  DESCRIPTION OF SYMBOLS 1 ... Test object, 2 ... Stage, 3 ... Imaging device, 4 ... Illuminating device, 4a ... 1st lighting fixture, 4b ... 2nd lighting fixture, 4c ... 3rd lighting fixture, 4d ... 4th illumination Instruments 5 ... Computer 6 ... Display device 7 ... Input device 8 ... Optical system 9 ... CCD camera 10 ... Defect detection device 11 ... Image processing unit 12 ... Storage unit

Claims (7)

A defect detection device for detecting defects on the surface of an object,
An illumination unit capable of illuminating the surface from one side of the surface and the other side opposite to the one side with the surface in between;
An imaging unit that images the surface illuminated by the illumination unit;
A processing unit for capturing a pair of images by capturing the surface illuminated from the one side and the surface illuminated from the other side,
The processing unit generates a defect detection image, which is an image for detecting a defect, based on the pair of images. The illumination unit can illuminate the surface from the one side and the other side in a first direction and a second direction intersecting the first direction,
The processing unit acquires the pair of images in the first direction and the second direction, respectively, and converts the pair of images in the first direction and the pair of images in the second direction. The defect detection apparatus according to claim 1, wherein the defect detection image is generated based on the defect detection image.   The said process part compares the said pair of image for every pixel, acquires a representative value, and produces | generates the said defect detection image based on the said representative value, The Claim 1 and Claim 2 characterized by the above-mentioned. Defect detection device.   The defect detection apparatus according to claim 3, wherein the processing unit compares the pair of images for each pixel and acquires a maximum value and a minimum value as the representative value.   The processing unit compares the pair of images for each pixel to acquire a maximum value and a minimum value, acquires a difference between the maximum value and the minimum value for each pixel, and detects the defect based on the difference The defect detection apparatus according to claim 3, wherein an image is generated. The illumination unit can illuminate the surface from the one side and the other side in a plurality of directions,
The processing unit acquires the pair of images in the plurality of directions, and generates a defect detection image that is an image for detecting a defect based on the pair of images in the plurality of directions. The defect detection apparatus according to any one of claims 1 to 5. A defect detection method for detecting defects on the surface of an object,
A pair of images is obtained by imaging each of the surface illuminated from one side and the surface illuminated from the other side opposite to the one side across the surface,
A defect detection method, wherein a defect detection image that is an image for detecting a defect is generated based on the pair of images.