JP2009198290A - Flaw detection method and detector - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a flaw detection method capable of enhancing the flaw detection sensitivity of an inspection target. <P>SOLUTION: The flaw detection method includes the flaw emphasizing treatment process and the flaw detecting process. The flaw emphasizing treatment process includes the process ST20 for forming a smoothed image from a photographed image, the process ST21 for successively selecting an inspection target pixel, the process ST22 for dividing a plurality of comparing target pixels, which are arranged to the periphery of the noticing pixel of the smoothed image corresponding to the inspection target pixel into a plurality of comparing target pixel groups to set them, the process ST23 for determining brightness difference data being the difference between the respective comparing target pixels of the comparing target pixel groups and the respective brightness values of the inspection target pixels and calculating the minimum brightness difference becoming minimum in its value at intervals of the comparing target pixel groups, and the process ST24 for setting the minimum brightness difference becoming maximum in its value in the minimum brightness difference calculated at intervals of the comparing target pixel groups. The flaw detecting process discriminates a flaw from the feature quantity of a flaw candidate region constituted of the flaw candidate pixels extracted by comparing a flaw emphasizing value with a threshold value. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a defect detection method for accurately and automatically detecting a defect displayed at a low contrast on the captured image due to scratches or spots on the inspection object by processing an image obtained by imaging the inspection object. And an apparatus.

  As a defect detection method for detecting a defect of an inspection object, a method of performing defect detection by processing an image obtained by imaging the inspection object is known (for example, see Patent Document 1).

  Patent Document 1 is a method for inspecting a defect using a spatial filter, in which the density of a pixel group located at a predetermined position determined in advance with reference to the target pixel of the spatial filter is extracted, and the density of the extracted pixel group is If it is within a certain range, it is judged that there is a defect and smoothing processing is not performed, and a differential processing spatial filter that emphasizes edges is applied. Otherwise, it is assumed that there is no defect. In this method, a defect is detected from an image with a noise component by applying a smoothing process and not applying a differentiation process.

JP 2004-333223 A

  However, in the method described in Patent Document 1, since defect detection is determined within a certain density value range, when the density of the defect differs depending on the location, for example, when shading such as uneven illumination is on the image. Is difficult to apply.

  The present invention has been made in view of the above-described problems, and an object of the present invention is to provide a defect detection method and apparatus that can reduce the influence of shading and improve the sensitivity of defect detection of an inspection object. And

  According to the defect detection method of the present invention, a defect enhancement processing step for performing defect enhancement processing on a captured image obtained by imaging an inspection object, and a defect based on a defect enhancement value of each pixel obtained in the defect enhancement processing step. A defect detection step for detecting, wherein the defect enhancement processing step performs a smoothing process on the captured image to create a smoothed image, and an inspection target pixel in the captured image. A plurality of comparison target pixels that are separated from the target pixel of the smoothed image corresponding to the selected inspection target pixel by a predetermined distance are arranged around the target pixel, and the comparison target pixels are sequentially selected. Luminance difference data that is a difference between a comparison target pixel group setting step that is set separately for a plurality of comparison target pixel groups, a luminance value of each comparison target pixel included in the comparison target pixel group, and a luminance value of the inspection target pixel Seeking Among the luminance difference data, the minimum luminance difference calculation step for obtaining the minimum luminance difference for which the value is minimum for each comparison target pixel group, and the minimum luminance difference calculated for each comparison target pixel group has the maximum value. A defect enhancement value calculation step that uses the minimum luminance difference as a defect enhancement value of the inspection target pixel, and the defect detection step compares the defect enhancement value at the inspection target pixel with a predetermined threshold value to obtain a defect candidate A feature is that a pixel is extracted, and a defect is determined from a feature amount of a defect candidate region constituted by the defect candidate pixel.

In addition, what is necessary is just to set an inspection object pixel and a comparison object pixel in the imaging pixel unit of the CCD camera which imaged the to-be-inspected object, for example.
In the present invention, in the defect enhancement processing step, the target pixel corresponding to the inspection target pixel selected in the captured image is set as a smoothed image, and a plurality of luminance values of the inspection target pixel and the periphery of the target pixel are arranged. Brightness difference data that is a difference from the brightness value of the comparison target pixel of the smoothed image is obtained, and the minimum brightness difference that minimizes the value is selected from the brightness difference data as the defect enhancement value of the inspection target pixel. Therefore, it is possible to detect a defect including the inspection target pixel and not including the comparison target pixel, for example, a planar spot defect.
That is, when there is no defect in the inspection target pixel portion and there is no luminance difference with the surrounding pixels, the minimum luminance difference is a very small value. Further, even if the inspection target pixel has a defect, if the defect extends to any of the comparison target pixel portions, there is almost no luminance difference between the inspection target pixel and the comparison target pixel included in the defective portion. The minimum luminance difference is also a very small value.
On the other hand, if the inspection target pixel has a defect and the surrounding comparison target pixel has no defect, that is, if the spot defect is within the area surrounded by the comparison target pixel, the luminance of the inspection target pixel Since the value is different from the luminance value of any comparison target pixel, the minimum luminance difference is also a relatively large value. As a result, when there is a spot defect having a size that can be accommodated in the area surrounded by the comparison target pixels, the minimum luminance difference becomes a relatively large value, and the defect is emphasized.

In the present invention, since the plurality of comparison target pixels are divided into a plurality of comparison target pixel groups, a line defect can be detected in addition to a spot defect. That is, when there is a line defect that overlaps at least one of the plurality of comparison target pixels and the detection target pixel, the difference in luminance value of each pixel on the line defect is small, so the minimum luminance difference is also a small value. Cannot be detected.
On the other hand, when the plurality of comparison target pixels are divided into a plurality of comparison target pixel groups and the minimum luminance difference is calculated for each comparison target pixel group as in the present invention, the position of each comparison target pixel is different. Even if the line defect overlaps the comparison target pixel of one comparison target pixel group and the line defect cannot be detected, the comparison target pixel of the other comparison target pixel group does not overlap the line defect, and the line defect is not detected. It can be detected.
For this reason, if the maximum value of each minimum luminance difference calculated for a plurality of comparison target pixel groups is used as the defect enhancement value of the inspection target pixel, both the spot defect and the line defect defect are not affected by shading. Can be emphasized. In particular, even if a line defect cannot be detected by any of the comparison target pixel groups, it can be detected by other comparison target pixel groups, so that detection sensitivity unevenness due to the angle of the line defect does not occur and detection sensitivity is improved. Can do.
In the present invention, since both a spot defect and a line defect can be detected simultaneously, the defect detection time can be shortened as compared with the case where a spot defect detection filter and a line defect detection filter are separately prepared and detected.

Furthermore, since the captured image is used as the image for setting the inspection target point, for example, compared to the case where the inspection target point is set in the image subjected to the smoothing process or the like, the defect component is weakened by the smoothing process or the like. There is no.
In addition, since the luminance comparison pixel is set to a smoothed image, it is possible to enhance the defect component without being affected by the noise component.
That is, when the comparison target pixel is also set as a captured image, even if a defect exists in the inspection target pixel, noise exists in the surrounding comparison target pixel portion, and a luminance value (density value) comparable to that of the inspection target pixel. ), There is almost no difference in luminance between the inspection target pixel and the comparison target pixel, and the defect cannot be detected even if the inspection target pixel in the defective portion is selected.
On the other hand, in the present invention, since the comparison target pixel is set to a smoothed image, the influence of noise on the comparison target pixel can be reduced, the defect component can be surely emphasized, and detection failure of the defect component can be prevented. Can be prevented.

Furthermore, in the defect enhancement processing step, not only the defect component but also the noise component is simultaneously enhanced to become a defect candidate, but in the defect detection step, the noise component and the defect component can be separated from the feature amount of the defect candidate. It can be identified and detected.
The defect candidate area may be set as one defect candidate area by collecting defect candidate pixels adjacent to each other.
Further, as the feature amount of the defect candidate region, for example, if the area of the defect candidate region is equal to or larger than a preset threshold, it is determined as a defect component, and if it is less than the threshold, it is determined as a noise component. Further, in addition to the area, the determination may be made in consideration of the average luminance, the maximum luminance, and the like of the defect candidate region.
In this way, in the present invention, since it is determined whether or not the defect component is based on the feature amount such as the area of the defect candidate region, it is possible to extract only the defect component by separating the noise component.

  In the defect detection method of the present invention, in the comparison target pixel group setting step, 4 × n (n is an integer of 2 or more) comparison target pixels are selected as the plurality of comparison target pixels, and these comparison target pixels are selected. Is preferably selected for each of four comparison target pixels arranged at intervals of 90 degrees in the circumferential direction centering on the inspection target pixel, and each comparison target pixel group is set.

Note that the specific number of the 4 × n comparison target pixels may be set based on the distance between the inspection target pixel and the comparison target pixel, and is usually any of 8, 12, or 16 You only have to set it.
Further, the distance between the inspection target pixel and the comparison target pixel may be set based on the size of the spot defect to be detected. Usually, the distance between the pixels is set to about 4 to 40 pixels, for example, 7 pixels. That's fine.

Furthermore, it is preferable that the comparison target pixels are arranged on the circumference and at equal intervals around the inspection target pixel. That is, an angle formed by a line segment connecting each comparison target pixel and the inspection target pixel and a line segment connecting another comparison target pixel and the inspection target pixel adjacent to the comparison target pixel is the same in each comparison target pixel. It is preferable.
Therefore, for example, when eight comparison target pixels are provided, they may be arranged at intervals of 45 degrees in the circumferential direction centering on the inspection target pixel. In this case, a first comparison target pixel group is configured by four inspection target pixels selected every other circumferential direction around the inspection target pixel, that is, four inspection target pixels arranged at intervals of 90 degrees. What is necessary is just to comprise a 2nd comparison object pixel group by the other four test object pixels which are not contained in this 1st comparison object pixel group.
Similarly, when 12 comparison target pixels are provided, they may be arranged at intervals of 30 degrees in the circumferential direction centering on the inspection target pixel. In this case, the first to third comparison target pixel groups are respectively formed by four inspection target pixels selected every two in the circumferential direction centering on the inspection target pixel, that is, four inspection target pixels arranged at intervals of 90 degrees. What is necessary is just to comprise.
Similarly, when 16 comparison target pixels are provided, they may be arranged at intervals of 22.5 degrees in the circumferential direction centering on the inspection target pixel. In this case, each of the first to fourth comparison target pixel groups is composed of four inspection target pixels selected every third in the circumferential direction around the inspection target pixel, that is, four inspection target pixels arranged at intervals of 90 degrees. What is necessary is just to comprise.

  Since each comparison target pixel group is composed of four comparison target pixels arranged at intervals of 90 degrees in the circumferential direction centering on the inspection target pixel, the presence or absence of a spot defect is confirmed with a minimum number of comparison target pixels. Even if there is a line defect that passes through two comparison target pixels arranged at point-symmetrical positions with respect to the inspection target pixel in a certain comparison target pixel group and the line defect cannot be detected, another comparison target pixel group Thus, the line defect can be reliably detected.

In the defect detection method of the present invention, the comparison target pixel group setting step includes, as the plurality of comparison target pixels, eight comparison target pixels arranged at intervals of 45 degrees in a circumferential direction centering on the inspection target pixel. These eight comparison target pixels are selected for each of four comparison target pixels arranged at intervals of 90 degrees in the circumferential direction centering on the inspection target pixel, and the first comparison target pixel group and the first comparison target pixel group Two comparison target pixel groups may be set.
For example, a first comparison target pixel group is configured by four comparison target pixels provided on the upper and lower sides and the left and right sides of the inspection target pixel, and the upper right side, the lower right side, the upper left side, and the left side of the inspection target pixel. The second comparison target pixel group may be configured by four comparison target pixels provided obliquely below.
If two comparison target pixel groups each including four comparison target pixels are set, it is possible to detect a spot defect and a line defect with a minimum comparison target pixel. For this reason, the defect detection process by a filter can also be performed in a short time.

  In the defect detection method of the present invention, the minimum luminance difference calculating step detects a bright defect in which the luminance of the defective portion is higher than the surrounding luminance, from the luminance value of the inspection target pixel. When the luminance difference data is obtained by subtracting the luminance value, the minimum luminance difference of the luminance difference data is obtained, and a dark defect in which the luminance of the defective portion is lower than the surrounding luminance is detected, the comparison target pixel It is preferable to obtain luminance difference data by subtracting the luminance value of the pixel to be inspected from the luminance value, and obtain the minimum luminance difference of the luminance difference data.

  In the present invention, the minimum value of each luminance difference data obtained by subtracting the luminance value of the comparison target pixel from the luminance value of the inspection target pixel is used as the defect enhancement value for the bright defect of the inspection target pixel, and the luminance of the comparison target pixel Since the minimum value of each luminance difference data obtained by subtracting the luminance value of the inspection target pixel from the value is used as the defect emphasis value for dark defect of the inspection target pixel, it is possible to detect the bright defect and the dark defect with high accuracy.

A defect detection apparatus according to the present invention includes a defect enhancement processing unit that performs defect enhancement processing on a captured image obtained by capturing an inspection object, and a defect based on a defect enhancement value of each pixel obtained by the defect enhancement processing unit. Defect detection means for detecting, and the defect enhancement processing means performs smoothing processing on the captured image to create a smoothed image, and inspection target pixels in the captured image. A plurality of comparison target pixels that are separated from the target pixel of the smoothed image corresponding to the selected inspection target pixel by a predetermined distance are arranged around the target pixel, and the comparison target pixels are sequentially selected. Luminance difference data that is a difference between a comparison target pixel group setting unit that is set to be divided into a plurality of comparison target pixel groups, a luminance value of each comparison target pixel included in the comparison target pixel group, and a luminance value of the inspection target pixel Seeking Among the brightness difference data, the minimum brightness difference calculating means for obtaining the minimum brightness difference with the smallest value for each comparison target pixel group, and the minimum brightness difference calculated for each comparison target pixel group has the largest value. And a defect enhancement value calculation unit that uses the minimum luminance difference to be a defect enhancement value of the inspection target pixel, and the defect detection unit compares the defect enhancement value at the inspection target pixel with a predetermined threshold value to obtain a defect candidate. A feature is that a pixel is extracted, and a defect is determined from a feature amount of a defect candidate region constituted by the defect candidate pixel.
This defect detection apparatus can also provide the same operational effects as the defect detection method.

FIG. 1 is a block diagram showing a configuration of a defect detection apparatus according to an embodiment of the present invention.
The defect detection apparatus according to the present embodiment detects defects of the inspection object 1 such as a flexible substrate, a liquid crystal panel (TFT panel), a semiconductor wafer, and the like, and inspects the surface of the inspection object for scratches and dirt. Inspection can be performed. The inspection object 1 is placed on the XY stage 2 and configured to be movable in a plane.
The defect detection device includes a microscope 4, a CCD camera 5, a computer device 6, and a display device 7.

The microscope 4 is provided for enlarging the inspection object 1 and photographing it with the CCD camera 5, and a microscope having a sufficient magnification for detecting a defect of the inspection object 1 is used.
The CCD camera 5 is an imaging unit that images the inspection object 1 through the microscope 4.
The computer device 6 is image processing means for controlling the CCD camera 5 and detecting the inspection object 1. The display device 7 is a display device such as a liquid crystal display connected to the computer device 6.

  The computer device 6 includes an image input unit 60, a defect enhancement processing unit 61, a defect extraction unit 62, and a defect determination unit 63.

  Image data of the captured image captured by the CCD camera 5 is input to the image input means 60 of the computer device 6. The captured image is stored in a storage means (not shown). Accordingly, an image acquisition process (imaging process) is performed in which the image input means 60 images the inspection object using the CCD camera 5.

  The defect enhancement processing unit 61 performs a defect enhancement processing step of performing defect enhancement processing on the acquired image, and includes a smoothed image creation unit 610, an inspection target pixel selection unit 611, and a comparison target pixel group setting. Means 612, minimum luminance difference calculation means 613, and defect enhancement value calculation means 614 are provided.

The smoothed image creating means 610 performs a smoothed image creating step of creating a smoothed image by performing a smoothing process on the captured image.
The inspection target pixel selection unit 611 performs an inspection target pixel selection step of sequentially selecting inspection target pixels in the captured image.
The comparison target pixel group setting unit 612 sets the target pixel of the smoothed image corresponding to the inspection target pixel selected in the captured image, and sets the target pixel of the comparison target pixel that is a predetermined distance away from the target pixel in the smoothed image. A plurality of comparison target pixels are arranged around the pixels, and a comparison target pixel group setting step is performed in which these comparison target pixels are divided and set into a plurality of comparison target pixel groups.
The minimum luminance difference calculating unit 613 obtains luminance difference data that is a difference between the luminance value of each comparison target pixel included in the comparison target pixel group and the luminance value of the inspection target pixel, and among the luminance difference data, A minimum luminance difference calculating step for obtaining a minimum luminance difference having a minimum value for each comparison target pixel group is performed.
The defect emphasis value calculating means 614 performs a defect emphasis value calculating step in which the minimum luminance difference having the maximum value among the minimum luminance differences calculated for each comparison target pixel group is used as the defect emphasis value of the inspection target pixel. Is.

  The defect includes a bright defect having a higher luminance value than other pixel portions and a dark defect having a lower luminance value. For this reason, the defect enhancement value calculation means 614 of the present embodiment is configured to separately calculate the defect enhancement value for bright defects and the defect enhancement value for dark defects.

The defect extraction unit 62 extracts defect candidates by comparing the result processed by the defect enhancement processing unit 61 with a predetermined threshold. As the threshold values, a bright defect threshold value and a dark defect threshold value are set, a bright defect region is extracted by comparing the bright defect enhancement result with the bright defect threshold value, and the dark defect enhancement result is compared with the dark defect threshold value. Thus, the dark defect area is extracted.
Alternatively, the defect candidate extraction process by the defect extraction unit 62 may be executed after applying a median filter or the like to the image processed by the defect enhancement processing unit 61 to perform noise removal processing.

  The defect discriminating means 63 discriminates the defect based on the area, average luminance, maximum luminance, etc. of each extracted defect area, further evaluates the rank of the defect, and determines which defect rank the current inspection object corresponds to. A defect discrimination process to be classified is executed.

Next, the operation of the defect detection apparatus according to the embodiment of the present invention will be described.
FIG. 2 is a flowchart for explaining the operation of the defect detection apparatus of this embodiment. The operation shown in FIG. 2 is realized by a program executed on the computer device 6.

First, when the inspection object 1 is set on the XY stage 2, the image input means 60 of the computer device 6 captures an image of the inspection object 1 with the CCD camera 5 and captures an image of the captured data. (Imaging step) is performed (ST1). At this time, the photographing data is taken into the computer device 6 as digital data of 4096 gradations (12 bits) by an A / D converter (not shown).
When the inspection object 1 is a display panel such as a liquid crystal panel, a specific image pattern may be displayed on the display panel so that defects can be easily detected. For example, there are an all-white screen pattern that displays the entire screen in white so that dark defects can be easily detected, an all-black screen pattern that displays all screens in black so that bright defects can be easily detected, and a halftone screen pattern. What is necessary is just to set suitably according to the defect kind to want.

  Next, the defect emphasis processing means 61 performs a defect emphasis processing step for emphasizing defects on the acquired image (ST2). In this defect enhancement processing step ST2, since it is difficult to detect a low-contrast defect as it is, a process for emphasizing the defect in the image is performed. The defect enhancement processing step ST2 is performed according to the processing flow shown in FIG.

First, the defect enhancement processing means 61 applies a smoothing filter to the captured image by the smoothed image creating means 610 to create a smoothed image (ST20).
Here, the smoothing filter applied to the captured image may be a filter that can cut noise components such as a local average filter and a median filter. In addition, the size of the smoothing filter to be applied may be, for example, a 3 × 3 pixel size or a larger size.

Next, the inspection target pixel selection unit 611 executes an inspection target pixel selection step of selecting an inspection target pixel to be inspected in the captured image (ST21).
In the present embodiment, the target pixel is selected for each imaging pixel unit of the CCD camera 5.

Next, the defect enhancement processing means 61 executes the comparison target pixel group setting step by the comparison target pixel group setting means 612 (ST22).
That is, the comparison target pixel group setting unit 612 sets eight comparison target pixels S1 to S8 in the circumferential direction around the target pixel O in the smoothed image as shown in FIG. The comparison target pixels S1 to S8 are divided into two comparison target pixel groups.
Note that the target pixel O is set by selecting a pixel of the smoothed image at the same coordinate position with respect to the inspection target pixel of the captured image selected in the inspection target pixel selection step ST21.

In the present embodiment, each of the comparison target pixels S1 to S8 is arranged at an interval of 45 degrees in the circumferential direction around the target pixel O in the smoothed image.
Specifically, the comparison target pixels S1 and S5 are arranged vertically (vertical direction) with the target pixel O in between, and the comparison target pixels S7 and S3 are arranged right and left (horizontal direction) with the target pixel O in between. . Further, the comparison target pixels S2 and S6 are arranged in an oblique direction (upper right and downward left) with the target pixel O in between, and in an oblique direction (upward left and downward in the right direction) with the target pixel O in between. Comparison target pixels S8 and S4 are arranged.
The comparison target pixels S1 and S5, the comparison target pixels S2 and S6, the comparison target pixels S3 and S7, and the comparison target pixels S4 and S8 are set at point-symmetric positions with the target pixel O as the center.

The distance between the target pixel O and the comparison target pixels S1 to S8 is set according to the size of the defect to be detected. That is, in this embodiment, since the defect is emphasized by the luminance difference between the inspection target pixel corresponding to the target pixel O and the comparison target pixels S1 to S8, the defect is large within an area surrounded by the comparison target pixels S1 to S8. Otherwise it cannot be detected. Therefore, the distance between the target pixel O and the comparison target pixels S1 to S8 may be set according to the size of the defect to be detected.
In the present embodiment, the comparison target pixels S1 to S8 are arranged in a substantially circular shape at a position about 7 pixels away from the pixel of interest O.

  Then, the comparison target pixel group setting unit 612 sets these comparison target pixels S1 to S8 separately into two comparison target pixel groups. That is, the first comparison target pixel group is set by the comparison target pixels S1, S3, S5, and S7 that are alternately arranged in the circumferential direction, and the second comparison is performed by the remaining comparison target pixels S2, S4, S6, and S8. A target pixel group is set.

  Next, the defect enhancement processing unit 61 selects the comparison target pixels S1 to S8 one by one for each comparison target pixel group by the minimum luminance difference calculation unit 613, and the luminance value of the selected pixel (the luminance in the smoothed image). Value) and luminance value data which is the difference between the luminance value of the pixel to be inspected (luminance value in the captured image) and a minimum luminance difference calculating step for obtaining a minimum luminance difference having the smallest value among the luminance difference data Is executed (ST23).

  Specifically, the minimum luminance difference calculating unit 613 first selects each pixel of the first comparison target pixel group one by one, and then compares each comparison target pixel S1, S3, S5 from the luminance value of the inspection target pixel. The luminance difference data F is obtained by subtracting the luminance value of S7. That is, when the luminance value of the inspection target pixel is “O1” and the luminance values of the comparison target pixels S1, S3, S5, and S7 are “S1, S3, S5, and S7”, the following formulas 1 to 4 are used. The luminance difference data F1, F3, F5, and F7 are calculated.

F1 = O1-S1 (Formula 1)
F3 = O1-S3 (Formula 2)
F5 = O1-S5 (Formula 3)
F7 = O1-S7 (Formula 4)

Next, the minimum luminance difference calculation unit 613 uses Equation 5 below to calculate the minimum luminance difference D1 having the smallest value among the luminance difference data F1, F3, F5, and F7 of the first comparison target pixel group. Ask for.
D1 = Min (F1, F3, F5, F7) (Formula 5)

Next, the minimum luminance difference calculation means 613 sequentially selects each pixel of the second comparison target pixel group one by one, and then calculates the luminance of each comparison target pixel S2, S4, S6, S8 from the luminance value of the inspection target pixel. The luminance difference data F is obtained by subtracting the value. That is, when the luminance value of the inspection target pixel is “O1” and the luminance values of the comparison target pixels S2, S4, S6, and S8 are “S2, S4, S6, and S8”, the following formulas 6 to 9 are used. The luminance difference data F2, F4, F6, F8 are calculated.
Further, the minimum luminance difference calculation means 613 uses the following equation 10 to calculate the minimum luminance difference D2 having the minimum value among the luminance difference data F2, F4, F6, and F8 of the second comparison target pixel group. Ask.

F2 = O1-S2 (Formula 6)
F4 = O1-S4 (Formula 7)
F6 = O1-S6 (Formula 8)
F8 = O1-S8 (Formula 9)
D2 = Min (F2, F4, F6, F8) (Formula 10)

  Next, the defect emphasis processing means 61 uses the defect emphasis value calculation means 614 to calculate the smallest luminance difference D1, D2 calculated for each of the first and second comparison target pixel groups as the inspection target pixel. A defect emphasis process is performed with the position defect emphasis value (ST24).

The above formulas 1 to 4 and formulas 6 to 9 are calculation formulas for emphasizing bright defects brighter than the background (surrounding pixels), and the comparison target pixels S1 to S8 are calculated from the luminance value O1 of the inspection target pixel. The luminance difference data F1 to F8 are obtained by subtracting the luminance values S1 to S8.
On the other hand, when dark defects that are darker than the background are emphasized, the luminance difference data F1 to F8 can be obtained by subtracting the luminance value O1 of the inspection target pixel from the luminance values S1 to S8 of the comparison target pixels S1 to S8. That's fine. Specifically, the luminance difference data F1 to F8 may be obtained using the following formulas 11 to 18.

F1 = S1-O1 (Formula 11)
F3 = S3-O1 (Formula 12)
F5 = S5-O1 (Formula 13)
F7 = S7-O1 (Formula 14)
F2 = S2-O1 (Formula 15)
F4 = S4-O1 (Formula 16)
F6 = S6-O1 (Formula 17)
F8 = S8-O1 (Formula 18)

  As shown in FIG. 3, the defect enhancement processing means 61 determines whether or not the defect enhancement processing has been completed over the entire acquired image (ST25). Inspection target pixels are selected (ST21), and a comparison target pixel group setting step ST22, a minimum luminance difference calculation step ST23, and a defect enhancement value calculation step ST24 are performed. That is, since the inspection target pixel is set for each imaging pixel of the CCD camera 5, the inspection target pixel may be sequentially moved for each imaging pixel, and the steps ST21 to ST24 may be sequentially performed.

On the other hand, if the processing has been completed in ST25, the defect enhancement processing means 61 generates a defect enhanced image using the defect enhancement value calculated for each pixel (ST26). That is, a bright defect-enhanced image is generated with a defect enhancement value for bright defects, and a dark defect-enhanced image is generated with a defect enhancement value for dark defects. In addition, the defect emphasis processing unit 61 generates only the defect emphasis image based on the calculated defect emphasis value when only the defect emphasis value of either the bright defect or the dark defect is calculated.
Thus, the defect enhancement processing step ST2 is completed.

When the defect emphasis processing step ST2 is completed, as shown in FIG. 2, the defect extraction means 62 of the defect emphasis processing means 61 sets a threshold for cutting out the defect with respect to the defect enhancement image obtained in the defect enhancement processing step ST2. Then, a defect candidate extraction step for extracting defect candidates is executed (ST3).
That is, the defect extraction unit 62 sets a threshold for cutting out a bright defect when a bright defect emphasized image is generated, and sets a threshold for cutting out a dark defect when a dark defect emphasized image is generated, A region for each defect candidate is cut out from each defect-enhanced image. At this time, the defect extraction means 62 detects a region above the bright defect threshold as a bright defect region for the bright defect enhancement result, and sets a region above the dark defect threshold as the dark defect region for the dark defect enhancement result. To detect.
Here, each threshold value may be set to an optimum value according to the situation of the image. For example, the average value of the defect-enhanced image and its standard deviation may be obtained, and the threshold value may be set using the following equation.

Bright defect threshold wslevel = avr (bright) + α1 · σ (bright) + β1
Dark defect threshold bslevel = avr (dark) + α2 · σ (dark) + β2

Here, avr (bright) and avr (dark) are the average values of the defect-enhanced images, σ (bright) and σ (dark) are the standard deviations of the defect-enhanced images, and α1, α2, β1, and β2 are arbitrary numbers. Thus, it is appropriately determined depending on the situation of the image to be inspected.
Each defect-enhanced image also has a negative part. The negative part is a dark defect component in bright defect enhancement and a bright defect component in dark defect enhancement. When calculating the standard deviation, the negative value is omitted.

Next, the defect discriminating means 63 performs a blob process on the bright defect extracted image and the dark defect extracted image extracted from the emphasized image, and obtains the area of the region cut out as a defect candidate, the average luminance, and the maximum luminance. . Then, the defect discriminating means 63 compares the area in these feature quantities with a preset threshold value, determines that a defect candidate area whose area is equal to or larger than the threshold value is a defect component, and a defect candidate area less than the threshold value is a noise component. A defect discrimination step for judging that is carried out (ST4).
Further, the defect determination means 63 obtains a defect rank based on the area, average luminance, and maximum luminance of the regions separated as defect components in the defect determination step ST4.
The defect component and the defect rank obtained by the defect determination means 63 are displayed on the display device 7 so that the inspector can easily grasp the defect rank of the inspection object 1.

  Next, a verification result obtained by applying the present technique to the inspection target image in order to confirm the detection sensitivity according to the present embodiment will be described. As a comparative example, both the inspection target pixel and the comparison target pixels S1 to S8 are set in the captured image, the difference between the luminance value of the comparison target pixels S1 to S8 and the luminance value of the inspection target pixel is obtained, and the luminance difference Comparative Example 1 in which the minimum value of the data is the defect enhancement value, and Comparative Example in which the median process is performed on the captured image, and the same process as that of Comparative Example 1 is performed on the image after this process to obtain the defect enhancement value 2 will also be described.

FIG. 5 shows an inspection target image 100. The inspection target image 100 has dark defects that are displayed darker than the surrounding area. That is, there are a spot-like dark defect 101 having a relatively large area and an elongated linear dark defect 102. Dark portions other than these dark defects 101 and 102 are noise components.
FIG. 6 is an image of a processing result obtained by performing the processing of Comparative Example 1 on the inspection target image 100. In the processing result of Comparative Example 1, since the dark defects 101 and 102 are not sufficiently detected and the noise component is also detected, the defects 101 and 102 cannot be detected separately from the noise component. .
FIG. 7 is an image of a processing result obtained by performing the processing of Comparative Example 2 on the inspection target image 100. In the processing result of Comparative Example 2, it is possible to remove the noise component and detect the defect 101 having a large area, but the elongated defect 102 is detected by being separated finely and the other part is detected. As a result, the entire defect 102 cannot be detected. For this reason, the comparative example 2 cannot detect a defect depending on the shape of the defect, and particularly cannot detect a line defect reliably.

FIG. 8 shows a dark defect enhanced image 120 obtained by applying the dark defect enhancing filter according to the method of the present embodiment to the inspection target image 100 and executing the defect candidate extracting step ST3 by the defect extracting unit 62. As in this embodiment, if the minimum luminance difference D1, D2 is obtained for each of the first and second comparison target pixel groups and the larger one is used as the defect enhancement value, noise is also detected together. However, the dark defects 101 and 102 of the spot defect and the line defect can also be detected.
Then, when the defect determination step ST4 is performed on the dark defect enhanced image 120 by the defect determination unit 63, as shown in FIG. 9, the noise component can be removed, and only the dark defects 101 and 102 that are detection targets are removed. It was detected and the effectiveness of the present embodiment was verified.

According to this embodiment, there are the following effects.
(1) The defect enhancement processing means 61 arranges the plurality of comparison target pixels S1 to S8 around the pixel of interest O corresponding to the inspection target pixel of the captured image in the smoothed image, and the comparison target. The pixels S1 to S8 are divided into two comparison target pixel groups and set, and the difference between the luminance value of the inspection target pixel in the captured image and the luminance value of each of the comparison target pixels S1 to S8 in the smoothed image is obtained. The minimum luminance differences D1 and D2 having the smallest luminance difference data are calculated for each pixel group, and among these minimum luminance differences D1 and D2, the largest value is used as the defect enhancement value of the inspection target pixel.
For this reason, by obtaining a minimum luminance difference in each comparison target pixel group, a stain defect in the region including the inspection target pixel and surrounded by the comparison target pixel, each inspection target pixel, and any comparison target Line defects other than the line defects passing through the pixels can be emphasized, and even if shading such as illumination unevenness is on the inspection target image, the defects can be detected without any problem, and erroneous detection can be reduced.
Since each comparison target pixel group is different in the position of the comparison target pixel, a line defect that cannot be emphasized in one comparison target pixel group can be emphasized in the other comparison target pixel group. By using a luminance difference having a larger value as the defect enhancement value of the pixel to be inspected, it is possible to emphasize and detect a spot defect and a line defect.
For this reason, in this embodiment, it cannot detect or cannot be detected depending on the angle of the line defect, it is possible to compensate for the difference in detection sensitivity depending on the defect direction, and the defect detection sensitivity can be improved.
In addition, since both a spot defect and a line defect can be detected at the same time, the defect detection process using the spot defect detection filter and the line defect detection process using the line defect detection filter are performed in a different manner. The detection processing time can also be shortened.

(2) Further, since the inspection target pixel is set in the captured image and the luminance value is obtained from the captured image, the defect component is obtained by the smoothing process as compared with the case where the luminance value of the inspection target pixel is obtained from the smoothed image. Will not weaken.
On the other hand, since the comparison target pixels S1 to S8 are set to a smoothed image and their luminance values are also obtained from the smoothed image, the defect component can be emphasized without being affected by the noise component.
Therefore, in this embodiment, the luminance value of the inspection target pixel is acquired from a captured image in which the defect component is not weakened, and the luminance value of the comparison target pixel is acquired from a smoothed image in which the influence of the noise component is reduced. Therefore, the defect component can be surely emphasized, and detection failure of the defect component can be prevented.

(3) Further, in the defect emphasis processing step ST2 by the defect emphasis processing means 61, not only the defect component but also the noise component are simultaneously enhanced to become defect candidates. However, in the defect discrimination process ST4 by the defect discrimination means 63, the defect candidate Since a defect candidate whose feature amount, for example, the area of the defect candidate is less than a threshold, is separated as a noise component, and a defect candidate having an area larger than the threshold is separated as a defect component, the defect component can be identified and detected.

(4) In this embodiment, by appropriately setting the distance between the inspection target pixel (target pixel O) and the comparison target pixels S1 to S8, the size of the spot defect that can be emphasized can be set. Defects can be detected easily and with high accuracy.

(5) In the present embodiment, the defect enhancement value is calculated by the difference between the luminance value of the inspection target pixel and the luminance value of the comparison target pixels S1 to S8, and is between the inspection target pixel and the comparison target pixels S1 to S8. The luminance value of the point is not used. Therefore, if the spot defect is smaller than the area surrounded by the comparison target pixels S1 to S8 and includes the inspection target pixel, even if the size of the defect changes to some extent, the determination by visual inspection of the conventional inspector Defects can be detected in the same way. For this reason, it is not necessary to set the distance size between the inspection target pixel and the comparison target pixels S1 to S8 so finely and can be easily set.

(6) The minimum luminance difference D1 and D2 in each comparison target pixel group is obtained and used as a defect enhancement value. When this defect enhancement value is equal to or greater than a predetermined threshold value, it is recognized as a defect candidate. Can be reduced.

(7) Furthermore, in the minimum luminance difference calculation means 613, since the formula for calculating the bright defect enhancement value and the formula for calculating the dark defect are set separately, the defect portion of the bright defect and the dark defect is determined. Each can be emphasized with high precision, and both the bright defect and the dark defect can be detected easily and accurately by extracting the light defect threshold value and the dark defect threshold value with the defect extraction means 62, respectively. Can do. For this reason, various defects can be efficiently detected by a simple process.

(8) Since the rank of the extracted defect can be classified by the defect discriminating means 63, the objective ranking of the defect can be performed in a short time, and the inspector can easily determine the degree of the defect, and it is a non-defective product. The determination of whether or not can be facilitated in a short time.

The present invention is not limited to the above embodiment.
For example, the distance between the inspection target pixel (target pixel O) and the comparison target pixels S1 to S8 in the filter for emphasizing the defect is not limited to that of the above embodiment, but depends on the size of the spot defect to be detected. Can be set. Further, the defect size to be detected may be changed by performing processing by reducing the size of the image to be detected while the filter size is constant.

  In order to deal with various defect sizes, a plurality of filters having different distances between the inspection target pixel (target pixel O) and the comparison target pixels S1 to S8 are prepared, and these filters are applied to the same image. It is also possible to acquire defect-enhanced images of each size, compare the emphasis values of the pixels at the same position in the defect-enhanced images, select the maximum value, and combine them into one to create a defect-enhanced image. According to such a configuration, spot defects and line defects of a plurality of sizes can be detected together and easily.

Furthermore, in the above embodiment, eight comparison target pixels S1 to S8 are provided, and these comparison target pixels S1 to S8 are divided into two comparison target pixel groups. It is not limited to that of the embodiment.
For example, as shown in FIG. 10, twelve comparison target pixels S11 to S14, S21 to S24, and S31 to S34 are provided and arranged at intervals of 90 degrees in the circumferential direction centering on the inspection target pixel (target pixel O). A defect enhancement filter in which a comparison target pixel group is configured for each of the comparison target pixels S11 to S14, S21 to S24, and S31 to S34 may be used.

  Furthermore, as shown in FIG. 11, 16 comparison target pixels S11 to S14, S21 to S24, S31 to S34, and S41 to S44 are provided, and 90 in the circumferential direction centering on the inspection target pixel (target pixel O). A comparison target pixel group is configured for each of the comparison target pixels S11 to S14, S21 to S24, S31 to S34, and S41 to S44 arranged at intervals of degrees, and a defect enhancement filter provided with four comparison target pixel groups is used. May be.

  In the embodiment and the modifications shown in FIGS. 10 and 11, four comparison target pixels are arranged in each comparison target pixel group, and these comparison target pixels are arranged at intervals of 90 degrees in the circumferential direction. However, the interval between the pixels to be compared is not limited to 90 ° intervals. However, if four comparison target pixels are provided and arranged at intervals of 90 degrees, the respective comparison target pixels can be arranged at equal intervals, and defects can be detected efficiently.

Further, in the modified example shown in the embodiment and FIGS. 10 and 11, four comparison target pixels are arranged in each comparison target pixel group. However, as shown in FIG. S18, S21 to S28 are provided, and a comparison target pixel group is configured for each of the comparison target pixels S11 to S18 and S21 to S28 arranged every other in the circumferential direction centering on the inspection target pixel (target pixel O). Alternatively, two comparison target pixel groups each having eight comparison target pixels may be set.
In this case, since the comparison target pixels S11 to S18 and S21 to S28 are arranged at intervals of 45 degrees in each comparison target pixel group, the line defects that can be detected are compared with the above-described embodiment arranged at intervals of 90 degrees. The width dimension becomes smaller. That is, when detecting a line defect, it is necessary for the line defect to pass through a gap between each comparison target pixel. For this reason, when the comparison target pixels are arranged at intervals of 45 degrees, the width dimension of the line defect passing through the gap is also reduced. Therefore, when limiting the width dimension of the line defect to be detected, a filter as shown in FIG. 12 may be used.

The number of comparison target pixels may be set according to the distance between the inspection target pixel (target pixel O) and the comparison target pixel. That is, when the distance between the inspection target pixel (target pixel O) and the comparison target pixel is increased, the interval between the comparison target pixels is widened, so that the comparison target pixels can be appropriately increased so that the defect can be detected appropriately. It is preferable to do.
However, as the number of comparison target pixels increases, the processing takes time correspondingly, and therefore the number of comparison target pixels is preferably 8, 12, or 16.

  In the above-described embodiment, the defect determination unit 63 determines the defect rank based on the area of the defect portion or the like, but the defect may be determined by another method / procedure. In short, the defect discriminating unit 63 only needs to be able to determine whether or not it corresponds to a defect based on the defect emphasized by the defect enhancement processing unit 61.

  The present invention can detect if the captured image of the object to be inspected 1 has a portion having a luminance difference from the surrounding. For this reason, the present invention can be widely used for detecting foreign object defects on flexible substrates, detecting scratches and dirt on the surface of an object to be inspected, detecting brightness spot defects and color spot defects in various display devices, and the like.

The block diagram of the defect detection apparatus of the screen by embodiment of this invention. The flowchart for demonstrating operation | movement of the defect detection apparatus. The flowchart for demonstrating operation | movement of a defect emphasis processing process. The figure which shows the example of arrangement | positioning of the test object pixel with respect to a captured image, and a comparison object pixel. The figure which shows the example of the captured image detected by this embodiment. FIG. 10 is a diagram illustrating an example of a defect-emphasized image according to Comparative Example 1. The figure which shows the example of the defect emphasis image by the comparative example 2. FIG. The figure which shows the example of the defect emphasis image by this embodiment. The figure which shows the example of the image after the defect candidate extraction process by this embodiment. The figure which shows the modification of a defect emphasis filter. The figure which shows the other modification of a defect emphasis filter. The figure which shows the other modification of a defect emphasis filter.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Inspection object, 2 ... XY stage, 4 ... Microscope, 5 ... CCD camera, 6 ... Computer apparatus, 7 ... Display apparatus, 60 ... Image input means, 61 ... Defect emphasis processing means, 62 ... Defect extraction means, 63 ... defect determination means, 610 ... smoothed image creation means, 611 ... inspection target pixel selection means, 612 ... comparison target pixel group setting means, 613 ... minimum luminance difference calculation means, 614 ... defect enhancement value calculation means.

Claims (5)

A defect enhancement processing step of performing defect enhancement processing on a captured image obtained by imaging the inspection object; and
A defect detection step of detecting a defect based on the defect enhancement value of each pixel obtained in the defect enhancement processing step,
The defect emphasis processing step includes
A smoothed image creating step of creating a smoothed image by performing a smoothing process on the captured image;
An inspection pixel selection step for sequentially selecting inspection pixels in the captured image;
A plurality of comparison target pixels separated from the target pixel of the smoothed image corresponding to the selected inspection target pixel by a predetermined distance are arranged around the target pixel, and the comparison target pixels are divided into a plurality of comparison target pixel groups and set. A comparison target pixel group setting step,
The luminance difference data that is the difference between the luminance value of each comparison target pixel included in the comparison target pixel group and the luminance value of the inspection target pixel is obtained, and the minimum luminance difference that minimizes the value among the luminance difference data Calculating a minimum luminance difference for each comparison target pixel group,
A defect enhancement value calculating step in which, among the minimum luminance differences calculated for each comparison target pixel group, a minimum luminance difference having a maximum value is a defect enhancement value of the inspection target pixel, and
The defect detection step extracts a defect candidate pixel by comparing a defect emphasis value at the inspection target pixel with a predetermined threshold, and determines a defect from a feature amount of a defect candidate region constituted by the defect candidate pixel. A defect detection method characterized by the above. The defect detection method according to claim 1,
The comparison target pixel group setting step includes:
As the plurality of comparison target pixels, 4 × n (n is an integer of 2 or more) comparison target pixels are selected,
A defect characterized in that these comparison target pixels are selected for each of four comparison target pixels arranged at intervals of 90 degrees in the circumferential direction centering on the inspection target pixel, and each comparison target pixel group is set. Detection method. The defect detection method according to claim 2,
The comparison target pixel group setting step includes:
As the plurality of comparison target pixels, eight comparison target pixels arranged at intervals of 45 degrees in a circumferential direction centering on the inspection target pixel are selected.
These eight comparison target pixels are selected for each of four comparison target pixels arranged at intervals of 90 degrees in the circumferential direction centering on the inspection target pixel, and the first comparison target pixel group and the second comparison target pixel A defect detection method characterized by setting a pixel group. In the defect detection method in any one of Claims 1-3,
The minimum luminance difference calculation step includes:
When detecting a bright defect in which the luminance of the defective portion is higher than the surrounding luminance, the luminance difference data is obtained by subtracting the luminance value of the comparison target pixel from the luminance value of the inspection target pixel. Find the minimum brightness difference of
When detecting a dark defect in which the luminance of the defective portion is lower than the surrounding luminance, the luminance difference data is obtained by subtracting the luminance value of the pixel to be inspected from the luminance value of the comparison target pixel. A defect detection method characterized in that a minimum luminance difference is obtained. Defect enhancement processing means for performing defect enhancement processing on a captured image obtained by imaging the inspection object;
A defect detection means for detecting a defect based on the defect enhancement value of each pixel obtained by the defect enhancement processing means,
The defect enhancement processing means includes
A smoothed image creating means for creating a smoothed image by performing a smoothing process on the captured image;
Inspection target pixel selection means for sequentially selecting inspection target pixels in the captured image;
A plurality of comparison target pixels separated from the target pixel of the smoothed image corresponding to the selected inspection target pixel by a predetermined distance are arranged around the target pixel, and the comparison target pixels are divided into a plurality of comparison target pixel groups and set. Comparison target pixel group setting means,
The luminance difference data that is the difference between the luminance value of each comparison target pixel included in the comparison target pixel group and the luminance value of the inspection target pixel is obtained, and the minimum luminance difference that minimizes the value among the luminance difference data A minimum luminance difference calculating means for obtaining for each comparison target pixel group,
Among the minimum luminance differences calculated for each comparison target pixel group, a defect enhancement value calculation unit that sets a minimum luminance difference that has a maximum value as a defect enhancement value of the inspection target pixel,
The defect detection means extracts a defect candidate pixel by comparing a defect emphasis value at the inspection target pixel with a predetermined threshold value, and determines a defect from a feature amount of a defect candidate region constituted by the defect candidate pixel. A defect detection apparatus characterized by the above.

Images