JP2008014831A - Edge defect detection method and detection device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an edge defect detection method and a detection device capable of detecting a defect only from a photographed image, and detecting accurately even a defect near the edge. <P>SOLUTION: The edge defect detection device 1 includes a control device 6. The control device 6 includes an edge defect enhancement means 62 for enhancing a defect in contact with a continuous pattern edge by applying an edge defect enhancement filter to the imaged image and acquiring an edge defect enhancement value, and an edge defect detection means 63 for detecting an edge defect based on the edge defect enhancement value acquired by the edge defect enhancement means 62. The edge defect enhancement filter determines respectively each difference between an object pixel selected in the imaged image and each comparison pixel arranged separately from the object pixel as long as a prescribed distance, and the smallest value among the values is used as the edge defect enhancement value. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an edge defect detection method and a detection apparatus.

Conventionally, when inspecting a defect caused by foreign matter adhering to a pattern on a semiconductor wafer or printed circuit board, a captured image obtained by imaging an inspection object with a CCD camera or the like is compared with a reference image without a defect prepared in advance. A method based on pattern matching for inspection is known.
Generally, in the captured image, a minute difference occurs in the density change near the edge of the pattern every time the inspection object is imaged due to the positional relationship between the CCD camera and the inspection object. For this reason, even if the reference image and the captured image are strictly aligned, the defect may not be detected due to this minute difference near the edge of the pattern.

  On the other hand, in order to detect a defect near the edge of such a pattern, a technique for comparing the reference image and the captured image using density gradient information in the reference image and density gradient information in the captured image. Has been proposed (for example, Patent Document 1).

Japanese Patent Laying-Open No. 2005-228062

However, even in the method according to Patent Document 1, since it is necessary to create a reference image in advance in order to detect defects, for example, if there is a design change of an inspection object, a new reference image must be created. There was a problem of not becoming.
An object of the present invention is to provide an edge defect detection method and a detection apparatus that can detect a defect only from a captured image and can accurately detect even a defect near an edge.

  An edge defect detection method according to the present invention includes an edge defect enhancement step of applying an edge defect enhancement filter to a captured image to enhance a defect in contact with an edge of a continuous pattern and obtaining an edge defect enhancement value, and the edge defect enhancement step An edge defect detection step for detecting an edge defect based on the edge defect enhancement value acquired in step (a), wherein the edge defect enhancement filter sandwiches the target pixel selected in the imaging pixel and the target pixel, Two comparison pixels arranged on a straight line continuous in a direction by a predetermined distance, and the target pixel and two comparison pixels arranged by a predetermined distance in a direction orthogonal to the predetermined direction 3 Comparison pixels at three locations, that is, three comparison pixels arranged so that a straight line constituted by the three comparison pixels is parallel to the predetermined direction, Set, determined the difference between the density value of the density value and the target pixel of each comparative pixel each of these values, characterized in that the edge defect emphasizing value of the target pixel the smallest value.

The defect emphasis filter that is generally used obtains the difference between the density value of the target pixel and the density values of the comparison pixels arranged in the eight pixels around the target pixel. For example, among these values, A small value is used as the defect enhancement value of the target pixel.
By applying this defect enhancement filter to the captured image, the defect enhancement value becomes a large value when the difference between the density value of the target pixel and the density values of all the comparison pixels is large.
That is, if the target pixel is in the defective portion and all the comparison pixels are arranged in a region having a density difference from the defective portion, the defect enhancement value is large. For example, the defect enhancement value is greater than a predetermined threshold value. A defect can be detected by setting it as a defect when it is large.

  However, according to such a defect emphasis filter, when an attempt is made to emphasize a defect that touches an edge of a continuous pattern, if the density difference between the pattern and the edge defect is small, even if the target pixel is in the edge defect portion, As a result, a part of the comparison pixel is arranged in the pattern area, and the edge defect cannot be emphasized.

  On the other hand, according to the configuration of the present invention, the edge defect enhancement filter is disposed at a predetermined distance on a straight line that sandwiches the target pixel and continues in a predetermined direction (hereinafter referred to as a reference direction). Two comparison pixels, and three comparison pixels arranged at a predetermined distance in a direction orthogonal to a predetermined direction (hereinafter referred to as an inspection direction) from the target pixel and the two comparison pixels are set. Therefore, comparison pixels are not arranged at three locations in the direction opposite to the inspection direction, which is the comparison pixel in the general defect enhancement filter described above.

Therefore, even if the density difference between the pattern and the edge defect is small, the target pixel is in the edge defect portion, and all the comparison pixels are arranged in areas outside the pattern that have a density difference from the defect portion. For example, the edge defect enhancement value is a large value, and the edge defect enhancement value is a small value if the target pixel and at least one comparison pixel are both in the pattern, out-of-pattern, and edge defect areas.
That is, the edge direction of the continuous pattern (hereinafter referred to as the edge direction) is the reference direction, and the direction in which the edge defect protrudes from the pattern is the inspection direction, and the target pixel is in the edge defect portion. In addition, the edge defect enhancement value increases only when the edge defect falls within the area surrounded by each comparison pixel (hereinafter referred to as an enhancement area).

  For example, an edge defect enhancement filter in which the reference direction is the horizontal direction and the inspection direction is the upward direction sandwiches the target pixel and places two comparison pixels at a predetermined distance in the horizontal direction with respect to the captured image. In this case, the target pixel and three comparison pixels are arranged at a predetermined distance from the two comparison pixels in the upward direction with respect to the captured image, thereby upward from the edge of the pattern continuous in the horizontal direction. A protruding edge defect can be detected.

  Therefore, the edge defect detection step can detect an edge defect by setting the edge defect as an edge defect when the edge defect enhancement value is larger than a predetermined threshold, and accurately detect a defect in contact with the edge of a continuous pattern. Can do. Further, an edge defect can be detected from only the captured image, and it is not necessary to create a new reference image even if there is a design change of the inspection object, and the inspection time can be shortened.

  In the present invention, the edge defect emphasizing step includes an upward edge defect in which the predetermined direction is a horizontal direction with respect to the captured image, and a direction orthogonal to the predetermined direction is an upward direction with respect to the captured image. An enhancement filter, a right edge defect enhancement filter obtained by rotating the upper edge defect enhancement filter 90 degrees clockwise with respect to the captured image, and a right edge defect enhancement filter with respect to the captured image. The captured image includes a downward edge defect enhancement filter rotated 90 degrees clockwise, and a left edge defect enhancement filter obtained by rotating the downward edge defect enhancement filter 90 degrees clockwise with respect to the captured image. It is preferable that the edge defect emphasis value of each filter is obtained by applying to the above, and the largest value among these values is used as the edge defect emphasis value.

  According to such a configuration, the edge direction is a horizontal direction or a vertical direction with respect to the captured image, and the direction in which the defect that touches the edge of these patterns protrudes is any direction. Even so, edge defects are emphasized by the edge defect enhancement filter in either the top, bottom, left, or right direction, so by setting the largest value among these edge defect enhancement values as the edge defect enhancement value, the edge direction and edge Even when the direction in which the defect protrudes is unknown, it is possible to accurately detect the defect in contact with the edge.

In the present invention, it is preferable that the captured image is generated by being reduced according to the size of the edge defect to be detected.
According to such a configuration, even in the case where the size of the edge defect is larger than the enhancement region in the captured image, the captured image is created by being reduced according to the size of the edge defect to be detected. The edge defect can be detected with high accuracy by the edge defect enhancement filter having a size of.

In the present invention, the predetermined distance between the target pixel and each comparison pixel in the edge defect enhancement filter is preferably set according to the size of the edge defect to be detected.
According to such a configuration, since the predetermined distance between the target pixel and each comparison pixel is set according to the size of the edge defect to be detected, the enhancement region of the edge defect enhancement filter eventually detects the edge to be detected. It is set according to the size of the defect, and edge defects of different sizes can be detected with high accuracy from the same captured image.

In the present invention, it is preferable to rotate at least one of the captured image and the edge defect enhancement filter so that the direction of the continuous pattern in the captured image is parallel to a predetermined direction in the edge defect enhancement filter. .
According to such a configuration, since at least one of the captured image and the edge defect enhancement filter is rotated so that the edge direction in the captured image is parallel to the reference direction in the edge defect enhancement filter, the same With the edge defect emphasis filter, it is possible to accurately detect defects that are in contact with the edges of patterns that are continuous in various directions.

  An edge defect detection device according to the present invention applies an edge defect enhancement filter to a captured image to enhance a defect that touches an edge of a continuous pattern and acquires an edge defect enhancement value, and the edge defect enhancement unit Edge defect detection means for detecting an edge defect based on the edge defect enhancement value acquired in step (a), wherein the edge defect enhancement filter sandwiches the target pixel selected in the imaging pixel and the target pixel, Two comparison pixels arranged on a straight line continuous in a direction by a predetermined distance, and the target pixel and two comparison pixels arranged by a predetermined distance in a direction orthogonal to the predetermined direction 3 A comparison pixel at a location is set, the difference between the density value of each comparison pixel and the density value of the target pixel is obtained, and the smallest value among these values is determined as the target pixel. Characterized by an edge defect emphasizing values.

  According to such a configuration, the same operation and effect as those of the edge defect detection method described above can be enjoyed.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a diagram showing an edge defect detection apparatus according to an embodiment of the present invention.
The edge defect detection apparatus 1 detects edge defects of a semiconductor wafer 2, an XY stage 3 that supports the semiconductor wafer 2, a CCD camera 4 that images the semiconductor wafer 2, and a Z stage that installs the CCD camera 4. 5, an XY stage 3, a CCD camera 4, and a control device 6 connected to the Z stage 5.

  As shown in FIG. 1, the XY stage 3 can move in the XY direction, and the Z stage 5 can move in the Z direction. Accordingly, the positions of the semiconductor wafer 2 and the CCD camera 4 can be relatively moved in the XYZ directions.

  The control device 6 operates the XY stage 3, the CCD camera 4, and the Z stage 5 to capture the inspection area of the semiconductor wafer 2 and create a captured image, and the captured image creation unit 61 creates the captured image. Edge defect enhancement means 62 that applies an edge defect enhancement filter to a captured image to emphasize defects that touch the edges of a continuous pattern and obtain an edge defect enhancement value; and edge defects acquired by the edge defect enhancement means 62 And edge defect detection means 63 for detecting an edge defect based on the emphasis value.

Next, an edge defect detection method executed by the control device 6 will be described.
As shown in FIG. 2, the edge defect detection method includes a captured image creation step S1, an edge defect enhancement step S2, and an edge defect detection step S3. Hereinafter, each step will be described in detail with reference to FIGS.
When the edge defect detection is started, the captured image creating means 61 first operates the XY stage 3 and the Z stage 5 and moves the CCD camera 4 to the inspection area of the semiconductor wafer 2 to capture the captured image. Create (for example, FIG. 3).

Here, in the captured image shown in FIG. 3, there is a defect in contact with the edge of the pattern that continues in the horizontal direction in the region surrounded by the dashed-dotted circle 101.
In the present embodiment, a portion that is dark (black) in the captured image of FIG. 3 is a pattern, and a portion that is bright (white) is out of the pattern.

In addition, in the captured image shown in FIG. 3, there are edges of patterns that are continuous in the horizontal direction, and there are edges of patterns that are continuous in the diagonal direction.
In such a case, the captured image creating means 61 rotates the captured image in advance by a technique such as geometric transformation so that the edges of the pattern that continues in the diagonal direction are in the horizontal direction or the vertical direction. A plurality of captured images are created.

When each captured image is created, the edge defect enhancement unit 62 executes an edge defect enhancement step S2.
The edge defect emphasizing means 62 includes four 5 × 5 edge defect emphasis filters for detecting defects whose inspection directions are up, down, left, and right directions shown in FIGS. 4 to 7 in each captured image created by the control device 6. The defect that touches the edge of the pattern by applying is emphasized, and the largest value among the edge defect enhancement values of each filter is acquired as the edge defect enhancement value.

  Specifically, the edge defect enhancement means 62 applies four edge defect enhancement filters to all pixels of each captured image, and the edge defect enhancement values Y-out, Y-down-out, and Y-left calculated by each filter. out and Y right out are acquired, and the largest value among these edge defect enhancement values is set as the edge defect enhancement value of the target pixel.

For example, in the downward edge defect enhancement filter of FIG. 4, when the target pixel is “Y” and the density values of the comparison pixels are “X31, X35, X51, X53, X55”, (6) is used to calculate the edge defect enhancement value Y lower out.
In the present embodiment, a method for calculating the edge defect enhancement value when the density value of the pixels constituting the pattern in the captured image is small and the density value constituting the outside of the pattern is large as compared with this.

A1 = X31−Y (1)
A2 = X35-Y (2)
A3 = X51−Y (3)
A4 = X53−Y (4)
A5 = X55−Y (5)
Y lower out = min {A1 to A5} (6)

  In the present embodiment, in the downward edge defect enhancement filter, the reference direction is the horizontal direction with respect to the captured image, and the inspection direction is the downward direction with respect to the captured image. The predetermined distance for arranging the comparison pixel is two pixels. That is, the two comparison pixels that are arranged at a predetermined distance on a straight line continuous in a predetermined direction with the target pixel interposed therebetween correspond to X31 and X35, and are based on the target pixel and the two comparison pixels. Comparison pixels arranged at a predetermined distance in a direction orthogonal to the predetermined direction, that is, three comparison pixels arranged such that a straight line constituted by the three comparison pixels is parallel to the predetermined direction. Corresponds to X51, X53, and X55.

  At this time, a predetermined distance between the target pixel and the comparison pixel in the edge defect enhancement filter is set according to the size of the edge defect to be detected. Since the edge defect emphasis filter shown in FIGS. 4 to 7 in this embodiment has a predetermined distance of 2 pixels, the emphasis region is 3 × 2 pixels surrounded by “Y, X32, X34, X42, X43, X44”. And edge defects up to this size can be detected. In contrast, for example, an edge defect up to a size of 4 × 3 pixels can be detected by applying an edge defect enhancement filter having a predetermined distance of 3 pixels to the captured image. That is, by setting a predetermined distance between the target pixel and the comparison pixel in the edge defect enhancement filter, edge defects having different sizes can be detected.

  Similarly, using the following equations (7) to (12), the edge defect emphasis value Y out of the upward edge defect enhancement filter in FIG. 5 is calculated, and the equations (13) to (18) in FIG. The edge defect enhancement value Y left out of the right direction edge defect enhancement filter in FIG. 7 is calculated using the edge defect enhancement value Y left out of the left direction edge defect enhancement filter using equations (19) to (24).

A1 = X11−Y (7)
A2 = X13−Y (8)
A3 = X15−Y (9)
A4 = X31−Y (10)
A5 = X35-Y (11)
Y on out = min {A1 to A5} (12)

A1 = X11−Y (13)
A2 = X13−Y (14)
A3 = X31−Y (15)
A4 = X51−Y (16)
A5 = X53−Y (17)
Y left out = min {A1 to A5} (18)

A1 = X13−Y (19)
A2 = X15−Y (20)
A3 = X35-Y (21)
A4 = X53−Y (22)
A5 = X55−Y (23)
Y right out = min {A1 to A5} (24)

  Then, when Y lower out, Y upper out, Y left out, and Y right out are calculated, the largest value among these edge defect enhancement values is calculated by the following equation (25). Calculated as the defect emphasis value Yout.

Yout = max {Y out, Y out, Y left out, Y right out} (25)

Here, for example, the downward edge defect enhancement filter in FIG. 4 is applied to a region surrounded by a thick line in the captured image in which the defect is in contact with the edge of the wiring pattern of the semiconductor wafer 2 illustrated in FIG. In this case, the lower edge defect enhancement value Y out is calculated by the following equations (26) to (31).
It is assumed that the density value of the pixel that images the wiring pattern is 10, the density value of the pixel that images the edge defect is 50, and the density value of the pixel that images the portion outside the wiring pattern is 200.

A1 = 200−50 = 150 (26)
A2 = 200−50 = 150 (27)
A3 = 200−50 = 150 (28)
A4 = 200−50 = 150 (29)
A5 = 200-50 = 150 (30)
Y lower out = min {A1 to A5} = 150 (31)

  On the other hand, the edge defect when the downward edge defect enhancement filter in FIG. 4 is applied to the area surrounded by the thick line in the captured image having no defect in contact with the edge of the wiring pattern of the semiconductor wafer 2 shown in FIG. The enhancement value Y lower out is calculated by the following equations (32) to (37).

A1 = 200−200 = 0 (32)
A2 = 200−200 = 0 (33)
A3 = 200−200 = 0 (34)
A4 = 200−200 = 0 (35)
A5 = 200−200 = 0 (36)
Y lower out = min {A1 to A5} = 0 (37)

  As described above, when the target pixel Y is in the edge defect portion of the wiring pattern of the semiconductor wafer 2 and the edge defect is contained in the enhancement region, the edge defect enhancement value Y lower out increases. On the other hand, when there is no edge defect of the wiring pattern of the semiconductor wafer 2, the out under edge defect emphasis value Y becomes small.

When the edge defect enhancement value Yout is calculated, the edge defect detection unit 63 executes the edge defect detection step S3.
FIG. 9 shows a display image in which the edge defect enhancement value Yout calculated by the edge defect enhancement means 62 is displayed as the density value of each target pixel from the captured image shown in FIG. In FIG. 9, the larger the edge defect enhancement value Yout, the brighter (white), and the smaller, the darker (black).

The edge defect detection means 63 detects an edge defect by applying a predetermined threshold value to the edge defect enhancement value Yout. In the present embodiment, an edge defect is detected when the edge defect enhancement value Yout is greater than a predetermined threshold. That is, in the display image shown in FIG. 9, a portion displayed in white and existing in an area surrounded by a one-dot chain line circle 102 is detected as an edge defect.
Note that the portion detected as the edge defect by the edge defect detection means 63 corresponds to the edge defect existing in the region surrounded by the one-dot chain line circle in the captured image shown in FIG.

The edge defect detection apparatus 1 according to the embodiment has the following effects.
(1) The edge defect enhancement means 62 applies an edge defect enhancement filter to the captured image, whereby the edge direction in the captured image is the reference direction, and the direction in which the edge defect protrudes from the pattern is the inspection direction. In this case, the edge defect enhancement value is increased only when the edge defect fits in the enhancement region. Therefore, the edge defect detection means 63 can detect an edge defect by setting it as an edge defect when the edge defect enhancement value is larger than a predetermined threshold, and accurately detect a defect that touches an edge of a continuous pattern. be able to. Further, an edge defect can be detected from only the captured image, and it is not necessary to create a new reference image even if there is a design change of the inspection object, and the inspection time can be shortened.

(2) The edge defect enhancement means 62 applies the four edge defect enhancement filters whose inspection directions are up, down, left, and right directions to all the pixels of each captured image, and acquires edge defect enhancement values calculated by the filters, Among these edge defect enhancement values, the largest value is set as the edge defect enhancement value of the target pixel. As a result, regardless of whether the edge direction is the horizontal direction or the vertical direction with respect to the captured image, and the direction in which the defect touches the edge of these patterns protrudes, Since the edge defect is emphasized by the edge defect enhancement filter in either the left or right direction, even when the edge direction or the direction in which the edge defect protrudes is unknown, it is possible to accurately detect the defect in contact with the edge.

(3) The edge defect enhancement means 62 sets a predetermined distance between the target pixel and the comparison pixel in the edge defect enhancement filter according to the size of the edge defect to be detected. Thereby, the enhancement region of the edge defect enhancement filter is set according to the size of the edge defect to be detected, and edge defects of different sizes can be detected with high accuracy from the same captured image.

(4) The captured image creating means 61 rotates the captured image in advance by a technique such as geometric transformation so that the edges of the pattern that is continuous in the oblique direction are in the horizontal direction or the vertical direction, and a plurality of captured images are obtained. Thus, the defect that touches the edge of the pattern that continues in various directions can be accurately detected by the vertical / horizontal edge defect enhancement filter.

It should be noted that the present invention is not limited to the above-described embodiment, and modifications, improvements, etc. within a scope that can achieve the object of the present invention are included in the present invention.
For example, the edge defect enhancement filter is a 5 × 5 square filter, but may be a rectangular filter. In short, it may be created according to the shape of the edge defect to be detected.

Further, although the edge defect enhancement filter whose inspection direction is up, down, left, and right is applied to the captured image, eight edge defect enhancement filters whose inspection directions are different by 45 degrees may be applied. In short, in various directions It is only necessary that a defect in contact with an edge of a continuous pattern can be detected with high accuracy.
In addition, the predetermined distance between the target pixel and the comparison pixel in the edge defect enhancement filter is set according to the size of the edge defect to be detected. For example, the density value of the surrounding four pixels is averaged as one pixel. By doing so, the size of the captured image may be reduced, and in short, at least one of the edge defect enhancement filter and the captured image may be changed according to the size of the edge defect to be detected.

  Also, a plurality of captured images have been created by rotating in advance using a method such as geometric transformation so that the edge direction in the captured image is the horizontal direction or the vertical direction. For example, the CCD camera 4 is rotated. Thus, a plurality of captured images may be created. In short, it is only necessary that the captured image can be created by rotating the edge direction in the captured image to be the reference direction of the edge defect enhancement filter.

  The edge defect enhancement value was calculated by subtracting the density value of the target pixel from the density value of each comparison pixel. However, the density value of the pixels constituting the pattern in the captured image is large, and compared with this, it is outside the pattern. If the density value constituting the pixel is small, the density value of each comparison pixel may be calculated by subtracting the density value of the target pixel. In short, depending on the density value of the pixels constituting the pattern, edge defect, and outside of the pattern What is necessary is just to change the calculation method.

It is a figure which shows the edge defect detection apparatus in embodiment of this invention. It is a flowchart which shows the detection method of the edge defect detection apparatus. It is a captured image in the embodiment of the present invention. It is a figure which shows a downward edge defect emphasis filter. It is a figure which shows an upper direction edge defect emphasis filter. It is a figure which shows the left direction edge defect emphasis filter. It is a figure which shows a right direction edge defect emphasis filter. It is a figure which shows the state which applied the downward edge defect emphasis filter to the captured image. It is the display image which displayed edge defect emphasis value as a density value of each object pixel.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Edge defect detection apparatus, 2 ... Semiconductor wafer, 3 ... XY stage, 4 ... CCD camera, 5 ... Z stage, 6 ... Control apparatus, 61 ... Captured image preparation means, 62 ... Edge defect emphasis means, 63 ... Edge defect Detection means, S1 ... captured image creation step, S2 ... edge defect enhancement step, S3 ... edge defect detection step.

Claims (6)

An edge defect enhancement step of applying an edge defect enhancement filter to a captured image to enhance defects that touch the edge of a continuous pattern and obtaining an edge defect enhancement value;
An edge defect detection step for detecting an edge defect based on the edge defect enhancement value acquired in the edge defect enhancement step;
The edge defect enhancement filter includes a target pixel selected in the imaging pixel, two comparison pixels that are disposed at a predetermined distance on a straight line that sandwiches the target pixel and continues in a predetermined direction, and the target A pixel and three comparison pixels arranged at a predetermined distance in a direction orthogonal to the predetermined direction from the two comparison pixels are set, and the density value of each comparison pixel and the density value of the target pixel are set. An edge defect detection method characterized in that a difference is obtained and the smallest value among these values is used as an edge defect enhancement value of a target pixel. The edge defect detection method according to claim 1,
The edge defect enhancement step, wherein the predetermined direction is a horizontal direction with respect to the captured image, and a direction perpendicular to the predetermined direction is an upward direction with respect to the captured image;
A right edge defect enhancement filter obtained by rotating the upward edge defect enhancement filter clockwise by 90 degrees with respect to the captured image;
A downward edge defect enhancement filter in which the right edge defect enhancement filter is rotated 90 degrees clockwise with respect to the captured image;
Applying the left edge defect enhancement filter obtained by rotating the lower edge defect enhancement filter 90 degrees clockwise with respect to the captured image to the captured image to obtain the edge defect enhancement value of each filter, An edge defect detection method characterized in that the largest value among the values is used as the edge defect enhancement value of the target pixel. The edge defect detection method according to claim 1 or 2,
The edge defect detection method according to claim 1, wherein the captured image is generated by being reduced according to a size of the edge defect to be detected. The edge defect detection method according to any one of claims 1 to 3,
The edge defect detection method, wherein the predetermined distance between the target pixel and each comparison pixel in the edge defect enhancement filter is set according to the size of the edge defect to be detected. An edge defect detection method according to any one of claims 1 to 4,
An edge defect characterized by rotating at least one of a captured image and an edge defect enhancement filter so that a direction of a continuous pattern in the captured image is parallel to a predetermined direction in the edge defect enhancement filter Detection method. Edge defect enhancement means for applying an edge defect enhancement filter to a captured image to enhance defects that are in contact with the edges of a continuous pattern and obtaining edge defect enhancement values;
An edge defect detection means for detecting an edge defect based on an edge defect enhancement value acquired by the edge defect enhancement means;
The edge defect enhancement filter includes a target pixel selected in the imaging pixel, two comparison pixels that are disposed at a predetermined distance on a straight line that sandwiches the target pixel and continues in a predetermined direction, and the target A pixel and three comparison pixels arranged at a predetermined distance in a direction orthogonal to the predetermined direction from the two comparison pixels are set, and the density value of each comparison pixel and the density value of the target pixel are set. An edge defect detection apparatus characterized in that a difference is obtained and the smallest value among these values is set as an edge defect enhancement value of a target pixel.