JP3803677B2 - Defect classification apparatus and defect classification method - Google Patents

Description

  The present invention relates to a defect classification device and a defect classification method.

  In the manufacturing process of a semiconductor wafer, a multilayer wiring pattern is formed by repeating processes such as film formation, resist coating (coating), exposure, development, etching, and resist peeling a plurality of times. In such a manufacturing process, a defect may be caused by foreign matter mixed into the back surface of the coating material formed on the subject.

  Hereinafter, the defect caused by the contamination will be described with reference to FIG. When the coating material 800 is formed on the subject 801 by the resist coating process, if the foreign matter 802 is mixed into the surface of the subject 801, the coated surface of that portion is raised due to the foreign matter. This is a foreign matter mixing defect on the back surface of the coating material (hereinafter referred to as a back surface foreign material defect).

  When such a defect occurs, the surface of the subject rises, and in the subsequent exposure process, only that point cannot be focused, and an accurate pattern cannot be exposed. Unlike the defects that can be removed by cleaning the surface of the coating material, the backside foreign object defect needs to be reapplied after the resist is peeled off, and is also classified as another defect from the viewpoint of optimizing the manufacturing process by identifying the cause of the defect. There is a need to.

  Conventionally, such defect classification has been performed mainly by visual inspection. However, the visual inspection has a problem that the burden on the inspector is large and there are individual differences, and a unified result cannot be obtained. Therefore, it has been attempted to realize such surface inspection using an optical device.

  Japanese Patent Application Laid-Open No. 2003-168114 by the present inventor discloses a method for classifying a back surface foreign object defect or the like by a macro inspection that inspects the surface of a subject using a relatively low resolution image. This defect classification method will be described below with reference to FIG.

An inspection image 805 is obtained from the interference image (or diffraction image) of the subject surface ((A) in FIG. 18). Next, a non-defective image 806 is obtained from the interference image (or diffraction image) of the non-defective wafer as a reference ((B) in FIG. 18). Next, the difference image 807 is extracted by comparing the inspection image 805 and the non-defective image 806 or by comparing in units of sections having the same pattern in the interference image (or diffraction image) of the subject surface (FIG. 18). (C)). Next, a defect extraction image 809 is obtained by extracting from the difference image 807 based on pixels having a luminance value with a large difference from the non-defective part using a predetermined threshold. Next, the back surface foreign matter defects are classified based on the shape information (area, circularity, etc.).
JP 2003-168114 A

  However, in the method disclosed in Japanese Patent Application Laid-Open No. 2003-168114, the shape information used for classification of the back surface foreign object defect greatly depends on the threshold value when extracting the defect area. For example, when an appropriate threshold value is used, an original circular image (FIG. 19B) is obtained from the difference image as shown in FIG. 19A, but is extracted with a value lower than the appropriate threshold value. In the case of excessive extraction, the shape changes as shown in FIG. 19C by extracting surrounding pixels. On the other hand, at the time of under-extraction extracted with a value higher than the appropriate threshold value, the defective portion cannot be extracted sufficiently, so that the shape changes as shown in FIG. For this reason, the conventional method cannot perform classification with high robustness against a change in threshold value at the time of defect extraction. No defect classification apparatus or defect classification method that solves such problems has been proposed so far.

  The present invention has been made paying attention to such a problem, and the object of the present invention is to use the information of the luminance gradient, so that a threshold value at the time of defect extraction is unnecessary or robust with respect to the threshold value. An object of the present invention is to provide a defect classification apparatus and a defect classification method capable of performing high classification.

  In order to achieve the above object, a first invention is a defect classification apparatus, which illuminates a surface of a coating formed on a subject, and takes an image of the coating surface through an imaging optical system. Imaging means for performing, radiation direction component calculation means for calculating the radial direction component amount of the surrounding luminance gradient for each pixel in the captured image, and classification means for classifying the defect based on the calculated radiation direction component amount It has.

  According to a second aspect of the present invention, in the defect classification apparatus according to the first aspect, the illuminating unit illuminates illumination light with a limited wavelength, and the imaging unit includes reflected light from the subject. Then, an interference image with the reflected light from the coating surface is taken.

  According to a third aspect of the present invention, in the defect classification apparatus according to the first aspect, the illumination means illuminates the coating surface from a direction perpendicular to the coating surface, and the imaging means is perpendicular to the coating surface. The coating surface is imaged from the direction.

  According to a fourth aspect of the present invention, in the defect classification apparatus according to the first aspect, the illumination unit illuminates the coating surface from a direction horizontal to the coating surface, and the imaging unit is perpendicular to the coating surface. The coating surface is imaged from the direction.

  According to a fifth invention, in the defect classification device according to any one of the first to fourth inventions, a pixel used for calculation of the radial direction component amount is selected based on a positional relationship with a central pixel. A selection means is further provided.

  According to a sixth aspect of the present invention, in the defect classification device according to any one of the first to fourth aspects of the present invention, the selection of the pixel used for the calculation of the radial direction component amount based on the intensity of the luminance gradient of the pixel. Means are further provided.

  The seventh aspect of the present invention is the defect classification apparatus according to any one of the first to sixth aspects, further comprising a defect region extracting unit that extracts a defect region from the image captured by the image capturing unit. The radial direction component amount is calculated only for each pixel near the defective area.

  The eighth invention is a defect classification method, comprising: illuminating a surface of a coating formed on a subject; imaging an image of the coating surface through an imaging optical system; And calculating a radiation direction component amount of a surrounding luminance gradient for each pixel of the pixel, and classifying defects based on the calculated radiation direction component amount.

  According to a ninth invention, in the defect classification method according to the eighth invention, the illumination step illuminates illumination light with a limited wavelength, and the imaging step includes reflected light from the subject and the coating. An interference image due to reflected light from the surface is captured.

  The tenth invention is the defect classification method according to the eighth invention, wherein the illumination step illuminates the coating surface from a direction perpendicular to the coating surface, and the imaging step is perpendicular to the coating surface. The coating surface is imaged from the direction.

  The eleventh invention is the defect classification method according to the eighth invention, wherein the illumination step illuminates the coating surface from a direction horizontal to the coating surface, and the imaging step is perpendicular to the coating surface. The coating surface is imaged from the direction.

  According to a twelfth aspect of the present invention, in the defect classification method according to any one of the eighth to eleventh aspects, the pixel used for calculation of the radial direction component amount is selected based on a positional relationship with a central pixel. The method further includes a step.

  According to a thirteenth invention, in the defect classification method according to any one of the eighth to eleventh inventions, a step of selecting a pixel to be used for calculating the radial direction component amount based on the intensity of the luminance gradient of the pixel. Is further provided.

  The fourteenth invention is the defect classification method according to any one of the eighth to thirteenth inventions, further comprising a defect region extraction step for extracting a defect region from the captured image, and the extracted defect region. The radial direction component amount is calculated only for each neighboring pixel.

  According to the present invention, it is possible to perform classification that does not require a threshold value for defect extraction or that is highly robust to the threshold value.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

(First embodiment)
FIG. 1 is a diagram showing the configuration of the defect classification apparatus according to the first embodiment of the present invention. The defect classification apparatus according to the first embodiment includes an illuminator 101 that illuminates an object 100 such as a semiconductor wafer, a band-pass filter 102 that limits the wavelength of illumination light from the illuminator 101, and reflected light from the object 100. A CCD 103 for converting the formed image of the subject 100 into an electrical signal, an image input board 105 for capturing the electrical signal from the CCD camera 104 as an image, and an image Storage memory 106 for use in processing of each means described later, defect area extraction means 107 for extracting a defect area from an image, and emission of a surrounding luminance gradient for pixels in the vicinity of the extracted defect area Radiation direction component calculation means 108 for calculating the direction component amount and classification means 109 for classifying the back surface foreign object defect based on the calculated radiation direction component amount. .

  Here, the memory 106 is realized by the memory in the PC 110, and the defect area extraction means 107, the radial direction component calculation means 108, and the classification means 109 are realized by the CPU in the PC 110. Note that the subject 100 here is coated with a light-transmitting coating material (not shown in FIG. 1).

  Next, an image of a back surface foreign object defect in the first embodiment will be described. In the first embodiment, as shown in FIG. 2A, the surface of the subject 100 to which the coating material 121 having light transmittance is applied is illuminated with the illumination light 120 whose wavelength is limited. Therefore, what is imaged through the lens 103 is interference light 123 in which the light reflected from the surface of the coating material 121 and the light reflected from the surface of the subject 100 interfere with each other. Since the intensity of light due to interference depends on the film thickness 124 of the coating material 121, the brightness as shown in FIG. 2B and FIG. An image with a change of is obtained. That is, an image in which a luminance gradient is generated in the radial direction around a location where foreign matter is mixed is obtained. In FIG. 2B, the luminance decreases as the distance from the center increases. In FIG. 2C, the luminance increases as the distance from the center increases, and then the luminance decreases. That is, a ring-shaped (doughnut-shaped) brightness rises.

  Hereinafter, the operation of the defect classification apparatus according to the present embodiment will be described with reference to FIGS. In this defect classification apparatus, the wavelength of light from the illuminator 101 is limited by the band-pass filter 102 to irradiate the subject 100, and interference light caused by reflected light from the subject 100 and reflected light from the coating material 121 is applied to the lens 103. And imaged by the CCD camera 104 and converted into an electrical signal. This electric signal is taken into the calculation memory 106 through the image input board 105, and this becomes an examination image of the subject.

  FIG. 3A shows an example of the inspection image 130 acquired in this way, and a mura defect 131 (FIG. 3B) and a back surface foreign object defect 132 (FIG. 3C) are observed.

  Next, the defect area extraction unit 107 extracts the above-described defect area from the acquired inspection image 130. Two examples of extraction methods are shown below. In the first method, a threshold value that is a luminance range of a non-defective product level is set for an inspection image 130 as shown in FIG. 4A, and a pixel area having a luminance exceeding this threshold value is detected as a defect extraction image. It is extracted as 133 (FIG. 4B). Here, the threshold value indicating the brightness range of the non-defective product level may be preset in the PC 110 or may be determined adaptively based on the brightness histogram in the image (The University of Tokyo Press: Image Analysis). (See Handbook: Mikio Takagi, Yoshihisa Shimoda, Supervision: 502P, Binarization).

  In the second method, a non-defective wafer image 806 as shown in FIG. 18B (or an image 134 of a reference section that becomes a non-defective product as shown in FIG. The inspection image 130 (or the corresponding section in the inspection image) as shown in FIG. 5A is aligned, the luminance difference between the overlapping pixels is obtained, and the difference image 135 ((B) in FIG. 5) is created. Using this difference image 132, a defect area is extracted by threshold processing similar to the first method.

  As described above, two examples of the defect region extraction method have been described. After the binarization process of each of the above methods, the area of the extracted defect region is obtained (this process is generally used as particle analysis). A method may be considered), and a process of re-extracting an area whose area is within a predetermined range as a defective area having a possibility of a backside foreign object defect may be added.

  After the defective area is extracted in this way, the radial direction component calculation means 108 calculates the radial direction component amount of the surrounding luminance gradient for each pixel near the defective area. Hereinafter, this calculation procedure will be described.

1. The direction θ of the luminance gradient vector Vp ′ at an arbitrary target pixel P ′ in the inspection image 130 is obtained. Here, θ is calculated using the following equation (1) based on the luminance values I1 to I8 of the pixels 140-1 to 140-8 located around the target pixel 140 as shown in FIG. Since only the luminance gradient vector in the pixels around the defect area is used in the processing described later, it is limited to the pixels within the X pixels around the defect area (X is larger than the maximum diameter R_max of the back surface foreign object defect). By calculating, the amount of calculation can be reduced.

2. Next, the radial direction component amount Z _PR of the luminance gradient around the arbitrary pixel P in the vicinity of the defect region and within the radius R is calculated according to the following equation (2).

Here, as shown in FIG. 7, P ′ represents an arbitrary pixel existing around the target pixel P and within the radius R, and λ _{p ′} represents the luminance gradient vector V _{p ′} of P _′ and P, P This is the minimum angle (0 to π / 2 rad) formed by the direction vector V _{p_p '} determined from the position of'. N is the number of P ′.

3. The radial direction component amount Z _p of the luminance gradient around an arbitrary pixel P in the vicinity of the defect area is calculated according to the following equation (3).

  Here, R_min and R_max are set in advance as the minimum diameter and the maximum diameter of the back surface foreign object defect, respectively.

4). Next, the radial direction component amount Z in the target defect area is calculated according to the following equation (4).

  However, Y is set in advance.

  FIG. 8 is a diagram for explaining the difference in the radial direction component amount of the luminance gradient. In FIG. 8, the luminance gradient direction is obtained for each of the back surface foreign object defect 1, the back surface foreign object defect 2, and the unevenness defect. As can be seen from FIG. 8, in the backside foreign object defect 1, the luminance decreases as the distance from the center increases, and the luminance gradient indicates a radial direction toward the center. Further, in the backside foreign object defect 2, the luminance at a portion slightly away from the center is annularly increased, and the luminance gradient shows a radial direction from the center to the outside toward the annular luminance peak. In the case of a mura defect, the luminance is uniform over the entire area, and the luminance gradient has no directionality. Therefore, the required amount of radial direction component is high in an image in which a luminance gradient is generated in the radial direction around a predetermined pixel as in the case of backside foreign object defects 1 and 2, and other defects that have no luminance gradient (here In the case of non-uniformity defect), it becomes low.

  After the component amount in the radial direction is calculated, the classification unit 109 classifies the back surface foreign object defect. This is realized by setting a predetermined threshold for the calculated amount of radiation direction component, and classifying it as a backside foreign object defect if the amount of radiation direction component is larger than this threshold, and classifying it as others if it is smaller.

FIG. 9 is a flowchart showing a flow of defect classification processing according to the first embodiment of the present invention. First, an inspection image is acquired by the CCD camera 104 (step S1). Next, a defect area is extracted from the acquired inspection image by the defect area extraction means 107 (step S2). Next, the direction of the luminance gradient vector V _{p ′} in each pixel within the X pixels surrounding the extracted defect area (larger than the maximum diameter of the back surface foreign object defect) is calculated by the radial direction component calculating means 108 (step S3). Next, the radial direction component calculation means 108 calculates the radial direction component amount Z _PR of the luminance gradient around an arbitrary pixel P in the vicinity of the defect area using the equation (2) (step S4). Next, the radiation direction component calculation unit 108 calculates the expression (3) of the radiation direction component amount Z _P of the luminance gradient around an arbitrary pixel P in the vicinity of the defect area (step S5). Next, the radial direction component calculation means 108 calculates the radial direction component amount Z of the luminance gradient in the target defect area using the equation (4) (step S6). Next, the defect area is classified based on the radial direction component amount Z by the classification means 109 (step S7). Next, the classification result is output (step S8).

  As described above, according to the first embodiment of the present invention, it is possible to realize highly robust defect classification that is not influenced by the threshold value at the time of defect extraction. In this example, the defect region is extracted before calculating the radial direction component amount from the viewpoint of reducing the calculation time, and the radial direction component amount is calculated by focusing on pixels near the extracted region. If there is no such restriction, the amount of radial direction component is calculated for each pixel in the captured image, and if a pixel with a high value is defined as a backside foreign object defect, defect extraction processing is not required. In addition, classification with high accuracy becomes possible.

(Second Embodiment)
The second embodiment of the present invention will be described below. FIG. 10 shows the configuration of the defect classification apparatus according to the second embodiment of the present invention. The constituent elements denoted by reference numerals 200 to 210 in FIG. 10 are the same as those in the reference numerals 100 to 110 in FIG. 9, but in FIG. 10, it further includes selection means 211 for selecting the pixels used for calculating the radial direction component amount. Features.

In an object such as a semiconductor wafer, a pattern existing in a lower layer may appear in an inspection image. FIG. 11 shows an example of the back surface foreign object defect image when such a pattern is reflected. In many cases, a pattern used in a semiconductor or the like is substantially constituted by a horizontal line and a vertical line. When the pattern is reflected as shown in FIG. 11, when the radial direction component calculation means 108 (FIG. 1) of the first embodiment calculates the Z _PR using the equation (2), the value of Z _PR becomes the pattern. Will change. Therefore, in the defect classification apparatus according to the second embodiment, the selection unit 211 selects a pixel P ′ for calculating λ _{P ′} to be added when calculating Z _PR . As a selection method, for example, the following method can be used.

1) Among arbitrary pixels existing within the radius R around the target pixel P as the center, a pixel existing outside the horizontal and vertical directions α ° (preset) with respect to the pixel P is defined as P ′. FIG. 12 shows the range of P ′ selected by this method. Thereby, the influence by the horizontal and vertical patterns is reduced.

2) In the inspection image, the intensity f of the luminance gradient is calculated, and among the arbitrary pixels existing within the radius R around the target pixel P that is the center, a pixel having the intensity f equal to or less than a predetermined threshold is denoted as P ′. To do. For calculating the intensity f of the luminance gradient, there is a method (Sobel filter) in which the surrounding pixel luminance values described with reference to FIG.

  Since the luminance gradient due to the normal pattern is steeper than the luminance gradient due to the foreign matter on the back surface, the influence of the pattern appearing is reduced.

  As described above, according to the second embodiment of the present invention, it is possible to accurately classify the back surface foreign object defect even in the inspection image in which the pattern of the semiconductor or the like is reflected.

(Third embodiment)
The third embodiment of the present invention will be described below. FIG. 13 shows the configuration of the defect classification apparatus according to the third embodiment of the present invention. The third embodiment is the same as the first embodiment except for the configuration relating to illumination and imaging (301 to 304), and the components 305 to 310 correspond to the components 105 to 110 in FIG. In the third embodiment, the following configuration is used to enable illumination and imaging from vertically above the surface of the subject 300. That is, the half mirror 302 is newly added, and the half mirror 302, the lens 303, and the CCD camera 304 are sequentially arranged at a predetermined distance above the subject 300. Further, the illuminator 301 is disposed at a position vertically above the surface of the subject 300 and where light is guided to the half mirror 302. In the third embodiment, unlike the first embodiment, a subject 300 to which a light shielding coating material (not shown in FIG. 13) is applied is used.

  In the configuration described above, the light from the illuminator 301 is reflected by the half mirror 302 and illuminates the surface of the coating material applied to the subject 300 from vertically above. Reflected light from the surface of the coating material is imaged by a lens 303 through a half mirror 302 and converted into an electrical signal by a CCD camera 304. This electrical signal is taken into the arithmetic memory 306 by the PC 310 through the image input board 305. This is an inspection image on the surface of the coating material.

  Here, an image captured by the CCD camera 304 will be described with reference to FIGS. FIG. 14A is a diagram for explaining the angle of the coating material surface and the reflected light intensity.

  Here, an image captured by the CCD camera 304 will be described with reference to FIGS. FIG. 14A is a diagram for explaining the angle of the coating material surface and the reflected light intensity, and FIG. 14B is a diagram showing an example of the back surface foreign object defect image. As shown in FIG. 14A, a light shielding coating material 324 is applied on the subject 300. Moreover, the foreign material 322 is mixed in the coating material back surface.

  The illumination light 320 irradiated on the surface of the coating material 324 from vertically above is reflected vertically upward by the reflected light intensity pattern 321 corresponding to the angle of the surface of the coating material 324. That is, the reflected light is the largest when the surface of the coating material 324 is horizontal, and decreases as the surface of the coating material 324 becomes oblique. When the reflected light is large, the luminance of the image is high. Therefore, as shown in FIG. 14B, the image of the portion where the foreign material 322 is present has a luminance gradient in the radial direction around the portion where the foreign material is mixed. It becomes an image. Therefore, after the image is captured, it is possible to classify the defect by calculating the radial direction component amount of the luminance gradient with the same configuration as in the first embodiment.

  As described above, according to the third embodiment of the present invention, when imaging the surface of the light-shielding coating material 324, it is possible to classify the back surface foreign object defects with high accuracy.

(Fourth embodiment)
The fourth embodiment of the present invention will be described below. FIG. 15 shows the configuration of a defect classification apparatus according to the fourth embodiment of the present invention. The fourth embodiment is the same as the first embodiment except for the configurations related to illumination and imaging (401 to 404), and the components 405 to 410 correspond to the components 105 to 110 in FIG. In the fourth embodiment, the following configuration is used to enable illumination from a horizontal direction on the subject surface and imaging from the vertical direction. That is, a pair of illuminators 401 are arranged at the left and right positions of the subject 400 so that the surface of the subject 400 can be irradiated, and the lens 403 and the CCD camera 404 are sequentially arranged at a predetermined distance vertically above the subject 400. To do. Further, unlike the first embodiment, the fourth embodiment uses a subject 400 to which a light shielding coating material (not shown in FIG. 15) is applied.

  In the above-described configuration, the light from the pair of illuminators 401 illuminates the surface of the coating material applied to the subject 400 from the horizontal direction. The reflected light from the surface of the coating material is imaged by the lens 403 and converted into an electrical signal by the CCD camera 404. This electric signal is taken into the arithmetic memory 406 by the PC 410 through the image input board 405. This is an inspection image on the surface of the coating material.

  Here, an image captured by the CCD camera 404 will be described with reference to FIGS. FIG. 16A is a diagram for explaining the angle of the coating material surface and the reflected light intensity, and FIG. 16B is a diagram showing an example of the back surface foreign object defect image. As shown in FIG. 16A, a light-shielding coating material 424 is applied over the subject 423. In addition, foreign matter 422 is mixed on the back surface of the coating material 424.

  The light irradiated on the surface of the coating material 424 from the horizontal direction is reflected vertically upward by the reflected light intensity pattern 421 corresponding to the angle of the surface of the coating material 424. In other words, the reflected light is the smallest (almost none) when the surface of the coating material 424 is horizontal, and increases as the surface of the coating material 424 becomes oblique. When the reflected light is large, the brightness of the image is high. Therefore, as shown in FIG. 16B, the image of the portion where the foreign object 422 exists has a brightness gradient in the radial direction centering on the portion where the foreign object 422 is mixed. Will result in an image. Therefore, after the image is captured, it is possible to classify the defect by calculating the radial direction component amount of the luminance gradient with the same configuration as in the first embodiment.

  As described above, according to the fourth embodiment of the present invention, when imaging the surface of the light-shielding coating material 424, the foreign object defect on the back surface can be accurately detected without a complicated illumination-imaging system that requires a half mirror or the like. Can be classified.

It is a figure which shows the structure of the defect classification apparatus which concerns on 1st Embodiment of this invention. It is a figure for demonstrating the image of the back surface foreign material defect in 1st Embodiment. It is a figure which shows an example of the test | inspection image. It is a figure for demonstrating the 1st method of defect area extraction. It is a figure for demonstrating the 2nd method of defect area extraction. It is a figure for demonstrating the procedure which calculates | requires the direction of the brightness | luminance gradient vector in the arbitrary object pixels in a test | inspection image. It is a figure for demonstrating the procedure which calculates the radial direction component amount of the brightness | luminance gradient within the radius R around the arbitrary pixels of the defect area | region vicinity. It is a figure for demonstrating the difference in the radial direction component amount of a brightness | luminance gradient. It is a flowchart which shows the flow of the defect classification process which concerns on 1st Embodiment of this invention. It is a figure which shows the structure of the defect classification device which concerns on 2nd Embodiment of this invention. It is a figure which shows the example of a back surface foreign material defect image when the pattern which exists in a lower layer is reflected in the test | inspection image. It is a figure which shows the range of pixel P 'which exists other than the horizontal and vertical direction (alpha) degree with respect to the pixel P among the arbitrary pixels which exist in the radius R around the target pixel P used as the center. It is a figure which shows the structure of the defect classification device which concerns on 3rd Embodiment of this invention. It is a figure for demonstrating the image taken in by the structure of 3rd Embodiment of this invention. It is a figure which shows the structure of the defect classification device which concerns on 4th Embodiment of this invention. It is a figure for demonstrating the image taken in by the structure of 4th Embodiment of this invention. It is a figure for demonstrating the defect by a foreign material mixing. It is a figure for demonstrating the defect classification method by a prior art. It is a figure for demonstrating the problem of a prior art.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 100 ... Subject, 101 ... Illuminator, 102 ... Band pass filter, 103 ... Lens, 104 ... CCD camera, 105 ... Image input board, 106 ... Memory, 107 ... Defect area extraction means, 108 ... Radiation direction component calculation means, 109: Classification means, 110: PC.

Claims (14)

An illumination means for illuminating the surface of the coating formed on the subject;
Imaging means for imaging an image of the coating surface through an imaging optical system;
Radiation direction component calculating means for calculating the amount of radial direction component of the surrounding luminance gradient for each pixel in the captured image;
Classification means for classifying defects based on the calculated radial direction component amount;
A defect classification apparatus comprising: The illuminating means illuminates illumination light with a limited wavelength, and the imaging means images an interference image caused by reflected light from the subject and reflected light from the coating surface. The defect classification apparatus according to claim 1. The defect according to claim 1, wherein the illumination unit illuminates the coating surface from a direction perpendicular to the coating surface, and the imaging unit images the coating surface from a direction perpendicular to the coating surface. Classification device. The defect according to claim 1, wherein the illumination unit illuminates the coating surface from a direction horizontal to the coating surface, and the imaging unit images the coating surface from a direction perpendicular to the coating surface. Classification device. 5. The defect classification apparatus according to claim 1, further comprising a selection unit that selects a pixel used for calculation of the radial direction component amount based on a positional relationship with a central pixel. . 5. The defect classification apparatus according to claim 1, further comprising a selection unit that selects a pixel used for calculation of the radiation direction component amount based on a luminance gradient intensity of the pixel. The defect direction extracting means for extracting a defective area from the image picked up by the image pickup means is further included, and the radial direction component amount is calculated only for each pixel in the vicinity of the extracted defective area. The defect classification apparatus according to any one of 1 to 6. Illuminating the surface of the coating formed on the subject;
Capturing an image of the coating surface through imaging optics;
Calculating a radial direction component for calculating a radial direction component amount of a surrounding luminance gradient for each pixel in the captured image; and
Classifying defects based on the calculated radial direction component amount;
The defect classification method characterized by comprising. The illumination step illuminates illumination light with a limited wavelength, and the imaging step captures an interference image by reflected light from the subject and reflected light from the coating surface. 8. The defect classification method according to 8. 9. The defect according to claim 8, wherein the illumination step illuminates the coating surface from a direction perpendicular to the coating surface, and the imaging step images the coating surface from a direction perpendicular to the coating surface. Classification method. 9. The defect according to claim 8, wherein the illuminating step illuminates the coating surface from a direction horizontal to the coating surface, and the imaging step images the coating surface from a direction perpendicular to the coating surface. Classification method. The defect classification method according to any one of claims 8 to 11, further comprising a step of selecting a pixel used for calculation of the radial direction component amount based on a positional relationship with a central pixel. The defect classification method according to any one of claims 8 to 11, further comprising a step of selecting a pixel used for calculation of the radial direction component amount based on an intensity of a luminance gradient of the pixel. 14. A defect area extraction step for extracting a defect area from the captured image, and calculating the radial direction component amount only for each pixel in the vicinity of the extracted defect area. The defect classification method according to any one of the above.

Images