JP2009222513A - Flaw detector and flaw detection method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a flaw detector for properly detecting flaws, and to provide a flaw detection method. <P>SOLUTION: The flaw detector 1 includes an imaging part 3 for imaging an inspection target X to acquire an image taken; an image dividing means for dividing the taken image into a plurality of inspection regions; a standard deviation operating means for calculating the standard deviation of the respective inspection regions; a shadow part extraction means for extracting the inspection region of the inspection target, on the basis of the standard deviation; a shadow direction determining means for setting the scanning direction along the shadow direction, with respect to the inspection region; and a flaw detection means for calculating the luminance change quantity for the pixel of the inspection target, along the scanning direction and detecting the pixel of the inspection target becoming a predetermined flaw threshold or larger in the luminance change quantity. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a defect detection apparatus and a defect detection method for detecting a defect to be inspected.

  Conventionally, a detection method is known in which an inspection target such as a wafer is imaged and a defect such as a foreign object is detected from the captured image (see, for example, Patent Documents 1 and 2).

  The one described in Patent Document 1 captures a grayscale image of an inspection object placed on a stage and illuminated by illumination by a CCD camera, and each pixel is detected by an image detection unit based on the captured image. The inspection target density data is generated. Thereafter, the inspection calculation processing unit masks the inspection target density data using the masking data to exclude density data other than the inspection area, and divides the inspection target density data into a plurality of local areas. Then, the average density in each local region is calculated, and when the average density is outside the predetermined reference density range, it is determined that the defect is present.

  In addition, the device described in Patent Document 2 uses a pixel that is equidistant from the inspection target pixel along the scanning line direction in the scanning line direction as a comparison pixel, and determines the degree of defect from the density values of the inspection target pixel and this comparison pixel. A defect detection method for calculating and performing defect determination from this defect degree is employed.

JP 2000-258353 A JP-A-08-101139

  By the way, in the defect detection method described in Patent Document 1, if density unevenness occurs due to the influence of illumination or the like, it may be determined that the density is far from the average density even if it is a non-defective product. There is a problem of causing detection. In addition, in the defect detection method described in Patent Document 2, when a shadow appears due to illumination in the set scanning line direction, it reacts to the difference in density and causes a false detection. There is.

  In view of the above problems, an object of the present invention is to provide a defect detection apparatus and a defect detection method for accurately detecting defects.

  The defect detection apparatus according to the present invention includes an imaging unit that captures an inspection target and obtains a captured image, an image dividing unit that divides the captured image into a plurality of inspection regions, and all of the inspection target pixels in each inspection region. A standard deviation calculating means for calculating a standard deviation of brightness, a defect candidate area detecting means for detecting an area where the standard deviation is equal to or greater than a predetermined area standard deviation threshold as a defect candidate area, and a shadow on the defect candidate area. A scanning direction setting means for setting a scanning direction along the direction, and a luminance change amount of each inspection target pixel along the scanning direction, and an inspection target pixel whose luminance change amount is equal to or greater than a predetermined defect threshold And defect detection means for detecting as pixels.

  According to the present invention, the captured image obtained by imaging the inspection object by the image dividing means is divided into a plurality of inspection areas, and the standard deviation of the luminance in these inspection areas is calculated by the standard deviation calculating means. And a defect candidate area | region detects the test | inspection area | region which has a standard deviation below a predetermined area | region standard deviation threshold value as a defect candidate area | region. Thereby, in the inspection object, it is possible to extract a portion where a shadow is caused by an edge portion or the like. Here, in these extracted defect candidate areas, when there is an influence of a shadow, the luminance gradually changes along the direction of the shadow. Here, by setting the scanning direction substantially along the direction of the shadow by the scanning direction setting means, the luminance change along the scanning direction becomes substantially constant in the range where there is no defect. Therefore, the defect detection means calculates the amount of change in luminance along the scanning direction, and detects a pixel in which the amount of change in luminance is equal to or greater than the defect threshold value due to the influence of the shadow. Implementation becomes possible. Accordingly, even when a shadow is reflected in the captured image when the inspection object is imaged, the influence of the shadow can be reduced, and a defect can be detected with high accuracy.

In the defect detection apparatus of the present invention, the image dividing unit equally divides the inspection target portion of the captured image into a horizontal direction and a vertical direction orthogonal to the horizontal direction, and each of the equally divided regions. Is preferably the inspection region.
According to the present invention, since each inspection area has the same size, processing such as calculation of the standard deviation in each inspection area, calculation of the luminance change amount, and setting of the scanning direction by the standard deviation calculating means can be simplified. And the processing load can be reduced. In addition, by dividing the captured image into a plurality of inspection areas, it is possible to clearly separate the area affected by the shadow and the area without the shadow, and appropriately extract the area affected by the shadow. it can.

In the defect detection apparatus of the present invention, the scanning direction setting means includes a maximum luminance pixel having the maximum luminance and a minimum luminance pixel having the minimum luminance among the pixels arranged on the outer peripheral edge of the defect candidate region. When the maximum luminance pixel and the minimum luminance pixel are located at point symmetry positions with respect to the center point of the defect candidate area, the line direction connecting the maximum luminance pixel and the minimum luminance pixel is set as the scanning direction. It is preferable to do.
According to the present invention, the scanning direction setting means recognizes the luminance of each pixel arranged at the outer periphery of the inspection area, detects the maximum luminance pixel and the minimum luminance pixel, and detects the maximum luminance pixel and the minimum luminance pixel. When it is in a symmetrical position that is point-symmetric with respect to the center point of the inspection area, a line direction connecting these maximum luminance pixel and minimum luminance pixel is set as a scanning direction. That is, when a shadow is generated on an inspection target due to an edge portion or a concave portion, the luminance of the shadow direction of these shadows changes along one direction. Therefore, in general, in an inspection area in which a captured image to be inspected is sufficiently subdivided, if there is a shadow along a predetermined shadow direction, the shadow density is the highest at one of the outer peripheral edges in the inspection area. There is a point where the brightness is minimum and a point where the density of the shadow is the smallest and the brightness is the maximum. In general, when there is no bright defect or dark defect at the outer periphery of the inspection area, the maximum luminance pixel and the minimum luminance pixel exist at positions that are substantially point-symmetric with respect to the center point of the inspection area. The direction connecting the pixel and the minimum luminance pixel substantially coincides with the shadow direction. Therefore, by setting the direction connecting the maximum luminance pixel and the minimum luminance pixel as the scanning direction and performing defect detection along the scanning direction, defect detection substantially along the shadow direction can be performed. Therefore, it is possible to carry out accurate defect detection excluding a sudden luminance change due to the shading.

  At this time, in the defect detection apparatus of the present invention, the defect detection means scans along the scanning direction from the maximum luminance pixel toward the minimum luminance pixel, and the luminance change amount is larger than other inspection target pixels. It is preferable to detect a light defect and scan along the scanning direction from the minimum luminance pixel to the maximum luminance pixel to detect a dark defect whose luminance is smaller than that of other inspection target pixels.

According to the present invention, in detecting a bright defect, scanning is performed along the scanning direction from the maximum luminance pixel toward the minimum luminance pixel, and, for example, in the pixels adjacent to each other along the scanning direction, the positive direction in the scanning direction A difference value obtained by subtracting the luminance value of the pixel on the negative direction side in the scanning direction from the luminance value of the pixel on the side is calculated. Here, when there is no defect, the calculated difference value is a substantially constant value because the luminance difference due to the influence of the shadow is calculated.
On the other hand, when a bright defect exists, a difference value between a pixel corresponding to a portion without a defect (hereinafter referred to as a normal pixel) and the bright defect (the bright defect is on the positive side in the scanning direction, (When there is a normal pixel on the negative direction side) is a positive value. Further, when the difference value between pixels along the scanning direction is calculated, a difference value that is much larger than the shadow luminance change value is calculated when changing from a bright defect to a normal pixel along the scanning direction. . Therefore, it is possible to accurately detect the position of the bright defect without being affected by the shadow.
The same applies to the detection of dark defects, and by performing defect detection in the direction opposite to the scanning direction of bright defects, a positive difference value is calculated at the pixel position of the dark defect, and it is usually detected from the dark defect in the scanning direction. A difference value that is much larger than the shadow luminance change value is calculated at the transition to a pixel. Therefore, even in the case of dark defect detection, the position of the dark defect can be detected easily and accurately without being affected by the shadow.

  Furthermore, in the defect detection apparatus of the present invention, the scanning direction setting unit is configured such that the maximum luminance pixel and the minimum luminance pixel detected from the pixels arranged on the outer periphery of the defect candidate region are the center points of the defect candidate region. The maximum luminance symmetric pixel that is point symmetric with respect to the central point of the defect candidate region of the maximum luminance pixel, and the point with respect to the central point of the defect candidate region of the minimum luminance pixel Detecting a symmetric minimum luminance symmetric pixel, calculating a luminance difference between the maximum luminance pixel and the maximum luminance symmetric pixel, a luminance difference between the minimum luminance pixel and the minimum luminance symmetric pixel, and calculating the maximum luminance pixel and the maximum luminance pixel; The direction connecting any one of the pixels with the luminance symmetric pixel and between the pixels with the smallest luminance symmetric pixel and the smallest luminance symmetric pixel with a smaller value of the luminance difference is set in front. It is preferable to set the scanning direction.

  According to the present invention, when the arrangement positions of the maximum luminance pixel and the minimum luminance pixel are not point-symmetric with respect to the center point of the inspection area, the luminance difference between the maximum luminance pixel and the maximum luminance symmetric pixel, the minimum luminance pixel, and the minimum luminance The luminance difference of the symmetric pixel is calculated. One of the luminance differences having a smaller value is set as the scanning direction. That is, when the maximum luminance pixel and the minimum luminance pixel are not point-symmetric, for example, there may be a bright defect or a dark defect in any of the pixels arranged at the outer periphery of the inspection area. If the shadow direction is determined based on the pixel, it becomes a factor of erroneous detection. On the other hand, in the present invention, as described above, the luminance difference between the maximum luminance pixel and the maximum luminance symmetric pixel and the luminance difference between the minimum luminance pixel and the minimum luminance symmetric pixel are calculated, and one of these luminance differences is selected. Thus, the scanning direction is set. That is, since the brightness value of the bright defect and the dark defect are greatly different from those of the other pixels having no defect, the defect pixels and the defect pixels and the center point of the inspection area are point symmetric. The value of the luminance difference from the minimum luminance symmetric pixel or the maximum luminance symmetric pixel is much larger than the luminance difference due to the influence of the shadow. Therefore, by selecting one with a small luminance difference and setting it as the scanning direction, it is possible to appropriately set the scanning direction along the shadow direction, and to perform accurate defect detection excluding the influence of the shadow. it can.

  Here, according to the present invention, the defect detection unit is configured to detect the luminance difference between a pixel between the maximum luminance pixel and the maximum luminance symmetric pixel and between a pixel between the minimum luminance pixel and the minimum luminance symmetric pixel. It is preferable to detect any one of the maximum luminance pixel and the minimum luminance pixel corresponding to a pixel having a large value as a defective pixel.

According to this invention, as in the above invention, when the maximum luminance pixel and the minimum luminance pixel are not point-symmetric with respect to the center point of the defect candidate area, the luminance difference between the maximum luminance pixel and the maximum luminance symmetry pixel. Of the minimum luminance pixel and the luminance difference between the minimum luminance symmetric pixel and the larger one of the maximum luminance pixel or the minimum luminance pixel is recognized as a defective pixel. For example, when the luminance difference between the maximum luminance pixel and the maximum luminance symmetric pixel is larger than the luminance difference between the minimum luminance pixel and the minimum luminance symmetric pixel, the maximum luminance pixel is recognized as a bright defect having a higher luminance than other normal pixels. . Further, when the luminance difference between the minimum luminance pixel and the minimum luminance symmetric pixel is larger than the luminance difference between the maximum luminance pixel and the maximum luminance symmetric pixel, the minimum luminance pixel is recognized as a dark defect having a lower luminance than other normal pixels. .
As described above, when a defect is detected, a difference value in luminance between adjacent pixels along the scanning direction is calculated, and the defective pixel is recognized based on the calculated value. At this time, since the pixel arranged at the outer periphery of the inspection region is only one adjacent pixel, for example, when the pixel is located at the final end in the scanning direction, the comparison pixel for calculating the difference value Therefore, normal defect detection cannot be performed. In contrast, in the present invention, as described above, when the maximum luminance pixel and the minimum luminance pixel are not point-symmetric with respect to the center point of the defect candidate area, Of the luminance difference and the luminance difference between the minimum luminance pixel and the minimum luminance symmetrical pixel, the largest or minimum luminance pixel having a larger value is defined as a defective pixel. For this reason, it is possible to detect defective pixels with high accuracy even at the outer periphery of the inspection region.

  The defect detection method of the present invention is a defect detection method for detecting a defect to be inspected, images the inspection object, acquires a captured image, divides the captured image into a plurality of inspection regions, and The standard deviation of all the luminances of the inspection target pixels in the inspection area is calculated, and an area where the calculated standard deviation is equal to or greater than a predetermined area standard deviation threshold is detected as a defect candidate area. A scanning direction is set for the defect candidate area, a change in luminance of each pixel to be inspected along the scanning direction is recognized, a luminance change amount is calculated, and the calculated luminance change amount is a predetermined defect. A pixel to be inspected having a threshold value or more is detected as a defective pixel.

  According to this invention, similarly to the above-described invention, by setting the scanning direction along the shadow direction in the inspection object, it is possible to carry out defect detection excluding the influence of shadows, and to detect pixel defects with high accuracy. be able to.

Hereinafter, an embodiment according to the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing an outline of a defect detection apparatus according to an embodiment of the present invention.
FIG. 2 is a block diagram showing an outline of various programs developed by the CPU of the defect detection apparatus.
FIG. 3 is a diagram illustrating a state in which the captured image of the inspection target is divided into a plurality of inspection regions.
FIG. 4 is a diagram illustrating the minimum luminance pixel, the maximum luminance pixel, and the shadow direction in the inspection area.
FIG. 5 is a diagram illustrating a minimum luminance pixel, a maximum luminance pixel, and a shadow direction in another inspection area.

In FIG. 1, reference numeral 1 denotes a defect detection device. The defect detection device 1 is an apparatus that images an inspection object and detects a foreign matter attached to the inspection object from the captured image. As an inspection object, for example, precision parts such as a semiconductor wafer can be targeted.
And this defect detection apparatus 1 is provided with the stage 2, the imaging part 3, and the detection control apparatus 10, as shown in FIG.

  The stage 2 is formed in a substantially plate shape and includes a mounting surface 2A on which an inspection object X can be mounted. Further, the stage 2 includes a moving mechanism (not shown). This moving mechanism moves the stage 2 in the XY direction, which is the surface direction of the mounting surface 2A, by a control signal input from the detection control device 10.

The imaging unit 3 is provided facing the mounting surface 2A of the stage 2, and has an objective lens 31 facing the mounting surface 2A and a CCD (Charge Coupled Device) camera 32 that captures image light that has passed through the objective lens 31. And a focus driving unit 33.
The focus driving unit 33 holds the entire imaging unit 3 so as to be able to advance and retreat in the Z-axis direction orthogonal to the mounting surface 2A of the stage 2, and moves the imaging unit 3 from the detection control device 10 in the Z-axis direction. When the control signal is input, the imaging unit 3 is moved back and forth in a direction corresponding to the control signal.
In the imaging unit 3, the image light of the inspection object X incident from the objective lens 31 is imaged by the CCD camera 32, and the image data of the captured image is output to the detection control device.

  As shown in FIG. 1, the detection control device 10 includes an input / output unit 11, an input unit 12, a memory 13, and a CPU (Central Processing Unit) 14. As the detection control device 10, for example, a general-purpose personal computer may be used, or a processing device dedicated to defect detection processing may be used.

  The input / output unit 11 is connected to the moving mechanism of the stage 2 and the imaging unit 3, and outputs a predetermined control signal to the stage 2 and the imaging unit 3. Further, when image data of a captured image is input from the imaging unit 3, the input / output unit 11 outputs this image data to the CPU 14. In addition, the input / output unit 11 can be connected to output means such as a monitor and a printer and other external devices, for example, and can output data such as defect detection results to these external devices. In the present embodiment, an example in which output means such as a monitor or a printer is connected via the input / output unit 11 is shown. However, for example, a configuration in which these output means are integrally provided may be employed.

  The input unit 12 includes input means such as a keyboard and a mouse, for example, and outputs input signals input from these input means to the CPU. As this input signal, for example, a start request signal for starting defect detection processing by the defect detection device 1, a stage movement request signal for moving the stage 2 to a predetermined position, and focus adjustment by moving the imaging unit 3 For example, a focus request signal for performing the above.

  The memory 13 stores various programs developed on an OS (Operating System) that controls the overall operation of the detection control apparatus 10. The memory 13 stores various data used in various programs.

  The CPU 14 develops various programs recorded in the memory 13 and performs various arithmetic processes. As a program executed by the CPU 14, as shown in FIG. 2, an image recognition means 141, an area dividing means 142 as an image dividing means, a standard deviation calculating means 143, and a shadow portion as a defect candidate area detecting means. Examples include an extraction unit 144, a shadow direction determination unit 145 that also functions as a scanning direction setting unit, a defect detection unit 146, and an output control unit 147.

  The image recognition unit 141 recognizes image data of a captured image input from the imaging unit 3.

The area dividing unit 142 sets a range A to be inspected in the captured image in the captured image, and further performs processing to divide the range A into a plurality of inspection areas 100. Specifically, the area dividing unit 142 sets an inspection target range A as shown in FIG. 3 based on, for example, range information set and input by an input operation of the input unit 12 by a user. Note that, for example, a range within a predetermined pixel from the outer peripheral edge in the captured image may be set as the inspection target range A.
Then, the area dividing unit 142 divides the captured image recognized by the image recognizing unit 141 into a plurality of inspection areas as shown in FIG. Specifically, the area dividing unit 142 draws a plurality of virtual lines equally in the X direction and the Y direction orthogonal to the X direction in the captured image, and each square area surrounded by these virtual lines is inspected area 100. Set as.

  The standard deviation calculation means 143 calculates the standard deviation of luminance in each inspection area 100. Specifically, the standard deviation calculation means 143 recognizes all the luminance values of the inspection target pixels in each inspection region 100 and calculates the standard deviation of all these inspection target pixels.

  The shadow extraction unit 144 compares the standard deviation in each inspection region 100 calculated by the standard deviation calculation unit 143 with a preset brightness threshold, and the calculated standard deviation is smaller than the brightness threshold. (Inspection area 100 indicated by a thick line in FIG. 3) is extracted as an area with a shadow. Here, the brightness threshold is a value set for each inspection object X, for example, and may be stored in advance in the memory 13 so as to be appropriately readable. May be a value that is set and input when the defect detection process is performed.

The shadow direction determination unit 145 determines the direction of the shadow part in each inspection region 100 extracted by the shadow part extraction unit 144, and sets the scanning direction d, which is the direction in which the defect detection process is performed.
Specifically, as shown in FIG. 4, the shadow direction determination unit 145 has a minimum luminance value among the inspection target pixels arranged on the outer periphery in each inspection region 100 extracted by the shadow part extraction unit 144. The minimum luminance pixel 102 and the maximum luminance pixel 101 having the maximum luminance value are recognized.
Here, when the minimum luminance pixel 102 and the maximum luminance pixel 101 are arranged at positions that are point-symmetric with respect to the center point O in the inspection region 100, the shadow direction determination unit 145 determines that the minimum luminance pixel 102 and the maximum luminance pixel 101 are minimum luminance. The direction toward the pixel 102 is recognized as a shadow direction (a direction in which the shadow gradually darkens), and this shadow direction is set as the scanning direction d.

On the other hand, when the shadow direction determination unit 145 determines that the positions of the minimum luminance pixel 102 and the maximum luminance pixel 101 are not point-symmetric with respect to the center point O in the inspection region 100, as shown in FIG. A minimum luminance symmetric pixel 104 that is point-symmetric with the minimum luminance pixel 102 and a maximum luminance symmetric pixel 103 that is point symmetric with the maximum luminance pixel 101 are recognized with respect to the center point O of 100. Then, the shadow direction determination means calculates the luminance difference between the maximum luminance pixel 101 and the maximum luminance symmetric pixel 103 and the luminance difference (first luminance difference and second luminance difference) between the minimum luminance pixel 102 and the minimum luminance symmetric pixel 104, respectively. Among these, the pixel between one of the smaller values is determined as the shadow direction. For example, when the first luminance difference that is the difference between the luminance values of the maximum luminance pixel 101 and the maximum luminance symmetric pixel 103 is smaller than the second luminance difference that is the difference between the luminance values of the minimum luminance pixel 102 and the minimum luminance symmetric pixel 104. The direction from the maximum luminance pixel 101 to the maximum luminance symmetric pixel 103 is recognized as the shadow direction, and this shadow direction is set as the scanning direction d. When the second luminance difference is smaller than the first luminance difference, the direction from the minimum luminance symmetric pixel 104 toward the minimum luminance pixel 102 is recognized as the shadow direction, and this shadow direction is set as the scanning direction d.
That is, when the captured image is divided into a plurality of minute inspection areas 100, it can be considered that the shadow portion is always formed in a certain shadow direction in the inspection area 100. Therefore, when a luminance difference occurs due to the influence of the shadow portion, the minimum luminance pixel 102 with the darkest shadow and the lowest luminance exists at any of the outer peripheral edges that are the outermost portions of the inspection region 100, and There is a maximum luminance pixel 101 in which a pixel symmetric with respect to the central point O and the minimum luminance pixel 102 has the smallest shadow density. Therefore, when the maximum luminance pixel 101 and the minimum luminance pixel 102 are not point-symmetric, it can be determined that one of them is a bright defect or a dark defect, and a pixel or shadow having no such bright defect, dark defect, or other defect. The brightness value is clearly different compared to the amount of influence. Therefore, as described above, by comparing the first luminance difference and the second luminance difference and setting one of the smaller values in the scanning direction d, the scanning direction d can be substantially matched with the shadow direction.

The defect detection unit 146 scans each pixel (inspection target pixel) in the inspection area 100 along the scanning direction d set by the shadow direction determination unit 145 or the reverse scanning direction −d that is opposite to the scanning direction d. Scan and detect defective pixels.
Specifically, the defect detection means 146 performs a bright defect detection process for detecting a bright defect and a dark defect detection process for detecting a dark defect. Hereinafter, the light defect detection process and the dark defect detection process of the defect detection means will be described with reference to FIGS. FIG. 6 shows the luminance value of each inspection target pixel when there is a bright defect in a part of the inspection target pixel along the scanning direction d, the luminance difference between adjacent pixels when scanning along the scanning direction d, It is a figure which shows the luminance difference between the adjacent pixels at the time of scanning along reverse scan direction -d. FIG. 7 shows the luminance value of each inspection target pixel when there is a dark defect in a part of the inspection target pixel along the scanning direction d, the luminance difference between adjacent pixels when scanning along the scanning direction d, It is a figure which shows the luminance difference between the adjacent pixels at the time of scanning along reverse scan direction -d.

  In the bright defect detection process, the defect detection unit 146 scans the inspection target pixel along the scanning direction d. That is, the defect detection unit 146 calculates a luminance difference obtained by subtracting the luminance value of the inspection target pixel on the negative direction side in the scanning direction d from the luminance value of the inspection target pixel on the positive direction side in the scanning direction d. Here, as shown in the pixels M1 to M3 and M5 to M6 shown in FIG. 6, when there is no bright defect, the difference value is the amount of change in luminance due to the shade density difference (hereinafter referred to as the shadow influence amount). ), And scanning along the scanning direction d as described above results in a substantially negative value.

On the other hand, when there is a bright defect, if a normal pixel and a bright defective pixel continue along the scanning direction d, the luminance difference between these pixels is positive as shown, for example, between the pixel M3 and the pixel M4 in FIG. Value. Furthermore, when the light defect and the normal pixel are successively arranged in the scanning direction d in this order, for example, as shown between the pixel M4 and the pixel M5 in FIG. It is much smaller (absolute value is larger).
As described above, when the luminance difference between the inspection target pixels changes from a negative shadow influence amount to a positive value, the defect detection unit 146 corrects the positive in the scanning direction d among these inspection target pixels. The direction side pixel is detected as a bright defect. Further, the defect detection unit 146 determines the inspection target immediately after the bright defect when the luminance difference between the bright defect along the scanning direction d and the pixel to be inspected immediately after the bright defect substantially matches the shadow effect amount. Pixels are also detected as bright defects. Further, the defect detection unit 146 determines the inspection target immediately after the bright defect when the luminance difference between the bright defect and the pixel to be inspected immediately after the bright defect is a value that exceeds a predetermined value by the shadow influence amount. The pixel is recognized as a normal pixel.

Further, in this bright defect detection process, when a dark defect exists immediately after a normal pixel along the scanning direction d as shown by the pixels M3 to M4 in FIG. A small value (a value having a large absolute value) exceeding a predetermined threshold value is calculated from the partial influence amount. Further, when a normal pixel exists immediately after a dark defect along the scanning direction d, a positive value is calculated.
As described above, when the defect detection unit 146 changes from a shadow influence amount in which the luminance difference between pixels is a negative value to a small negative value exceeding a predetermined threshold value than the shadow influence amount, these inspections are performed. Among the target pixels, pixels on the positive direction side in the scanning direction d are recognized as dark defect candidates. In addition, when the luminance difference between the dark defect candidate along the scanning direction d and the pixel to be inspected immediately after the dark defect candidate substantially coincides with the shadow area influence amount, the defect detection unit 146 immediately follows the dark defect candidate. Are also recognized as dark defect candidates. Further, when the luminance difference between the dark defect candidate and the inspection target pixel immediately after the dark defect candidate becomes a positive value, the defect detection unit 146 sets the inspection target pixel immediately after the dark defect candidate as a normal pixel. recognize. That is, even when the luminance difference between the inspection target pixels becomes a positive value, the defect detection unit 146 determines the pixel on the positive direction side in the scanning direction d when the pixel on the negative direction side in the scanning direction d is a dark defect candidate. Since it is not detected as a bright defect, erroneous detection can be prevented.

  In the dark defect detection process, the defect detection unit 146 scans the inspection target pixel along the reverse scanning direction −d that is opposite to the scanning direction d. In other words, the defect detection means 146 determines the negative value in the reverse scanning direction −d (the positive direction in the scanning direction d) from the luminance value of the pixel to be inspected on the positive direction side in the reverse scanning direction −d (the negative direction in the scanning direction d). A difference value obtained by subtracting the luminance value of the pixel to be inspected is calculated. Here, as shown in the pixels M1 to M3 and M5 to M6 shown in FIG. 7, when there is no dark defect, this difference value becomes a shadow area influence amount due to the shade density difference, and the reverse scanning direction as described above. By scanning along -d, the value becomes a substantially constant positive value.

On the other hand, when there is a dark defect, if a normal pixel and a dark defect continue along the reverse scanning direction -d, for example, as shown between the pixel M5 and the pixel M4 in FIG. Is a negative value. Further, when the dark defect and the normal pixel continue in this order along the reverse scanning direction -d, for example, as shown between the pixel M4 and the pixel M3 in FIG. It is much larger (absolute value is larger) than the amount.
Therefore, when the luminance difference between the inspection target pixels changes from a shadow influence amount that is a positive value to a negative value, the defect detection unit 146 has a positive direction in the reverse scanning direction −d among the inspection target pixels. The inspection target pixel on the side is detected as a dark defect. In addition, when the luminance difference between the dark defect and the pixel to be inspected immediately after the dark defect substantially coincides with the shadow influence amount, the defect detecting unit 146 also sets the pixel to be inspected immediately after the dark defect as a dark defect. To detect. Further, the defect detection unit 146 determines that the luminance difference between the dark defect and the pixel to be inspected immediately after the dark defect is a large value (absolute value is large) exceeding a predetermined threshold value than the shadow effect amount. The pixel to be inspected immediately after the dark defect is recognized as a normal pixel.

Further, in this dark defect detection process, when a bright defect is present immediately after a normal pixel along the reverse scanning direction -d as shown by pixels M5 to M4 in FIG. 6, the luminance difference between these pixels is A large value (a value having a large absolute value) exceeding a predetermined threshold value is calculated from the shadow effect amount. Further, when a normal pixel exists immediately after a bright defect along the scanning direction, a negative value is calculated.
Therefore, the defect detection means 146, during the dark defect detection process, from the shadow effect amount in which the luminance difference between the inspection target pixels is a positive value, as described above, is larger than the shadow effect amount. In this case, the pixels on the positive direction side in the reverse scanning direction −d between the inspection target pixels are recognized as bright defect candidates. The defect detection means 146 also determines the bright defect candidate when the luminance difference between the bright defect candidate along the reverse scanning direction −d and the inspection target pixel immediately after the bright defect candidate substantially matches the shadow effect amount. The pixel to be inspected immediately after is recognized as a bright defect candidate. Further, when the luminance difference between the bright defect candidate and the inspection target pixel immediately after the bright defect candidate is a negative value, the defect detection unit 146 sets the inspection target pixel immediately after the bright defect candidate as a normal pixel. recognize. In other words, the defect detection unit 146 reverses when the pixel on the negative direction side in the reverse scanning direction −d is a bright defect candidate even when the luminance difference between the inspection target pixels becomes a negative value during the dark defect detection process. The inspection target pixel on the positive direction side in the scanning direction −d is not detected as a dark defect, and erroneous detection can be prevented.

  In addition, when the shadow direction determination unit 145 determines that the maximum luminance pixel 101 and the minimum luminance pixel 102 are not in a point-symmetrical position with respect to the center point O of the inspection area 100, the defect detection unit 146 One of the pixel 101 and the minimum luminance pixel 102 is recognized as a defective pixel. Specifically, when the shadow direction determination unit 145 determines that the first luminance difference is smaller than the second luminance difference, the defect detection unit 146 detects the minimum luminance pixel 102 as a dark defect. When the shadow direction determination unit determines that the second luminance difference is smaller than the first luminance difference, the defect detection unit 146 detects the maximum luminance pixel 101 as a bright defect.

  Furthermore, the defect detection means 146 generates a binarized image as shown in FIG. 8 in which defective pixels and normal pixels in the captured image are separated by binarization. Specifically, the defect detection means 146 sets the bright defect detected by the bright defect detection process and the dark defect detected by the dark defect detection process to “+1”, and other normal pixels to “0”. Generate a digitized image.

  The output control unit 147 performs control to output the pixel position of the defective pixel detected by the defect detection unit 146 and the binarized image to, for example, an output unit connected to the input / output unit 11.

(Defect detection operation)
Next, the defect detection operation in the defect detection apparatus 1 as described above will be described with reference to the drawings.
FIG. 9 is a flowchart of the defect detection operation of the defect detection apparatus 1.

  In the defect detection operation of the defect detection apparatus 1, first, the detection control apparatus 10 outputs a control signal to the imaging unit 3 and the stage 2, and the stage 2 and the imaging unit 3 are brought into a state where the inspection region of the inspection target X can be imaged. Move the to adjust the focus. Thereafter, the imaging unit 3 captures an inspection target area of the inspection target X, and outputs image data of the captured image to the detection control device 10 (step S101).

  When the image recognition unit 141 recognizes the input of image data from the imaging unit 3, the detection control device 10 sets the inspection target range A from the captured image of the image data by the region dividing unit 142. The area dividing unit 142 draws a plurality of virtual lines on the captured image and divides the picked-up image into a plurality of inspection areas 100 (step S102).

  Thereafter, the standard deviation calculation means 143 recognizes the luminance value of the pixel to be inspected in each inspection region 100, and calculates the standard deviation of the luminance value for each inspection region 100 (step S103). Then, the shadow extraction unit 144 compares the standard deviation of each inspection region 100 calculated in step S103 with a preset luminance threshold value, and shadows the inspection region 100 having a standard deviation smaller than the luminance threshold value. Are extracted as the inspection area 100 affected by (step S104).

After this step S104, the shadow direction determination unit 145 measures the luminance value of the inspection target pixel arranged at the outer peripheral edge of each inspection region 100, and among these inspection target pixels, the maximum luminance value is maximized. The brightness pixel 101 and the minimum brightness pixel 102 having the minimum brightness value are recognized.
The shadow direction determination unit 145 recognizes the positional relationship between the maximum luminance pixel 101 and the minimum luminance pixel 102, and the maximum luminance pixel 101 and the minimum luminance pixel 102 are point-symmetric with respect to the center point O of the inspection region 100. It is determined whether or not it is located at a position (step S106).

  In this step S106, when it is determined that the maximum luminance pixel 101 and the minimum luminance pixel 102 are positioned point-symmetrically with respect to the center point O, the shadow direction determination unit 145 directs the shadow direction from the maximum luminance pixel 101 toward the minimum luminance pixel 102. And the scanning direction d is set in parallel with the shadow direction (step S107).

On the other hand, when the shadow direction determination unit 145 determines in step S106 that the maximum luminance pixel 101 and the minimum luminance pixel 102 are not in point symmetry with respect to the center point O, the maximum luminance pixel 101 with respect to the center point O. And the minimum luminance symmetric pixel 104 that is point-symmetric with respect to the minimum luminance pixel 102 with respect to the center point O. Then, the shadow direction determination unit 145 determines the first luminance difference, which is the difference between the luminance value of the maximum luminance pixel 101 and the luminance value of the maximum luminance symmetric pixel 103, the luminance value of the minimum luminance pixel 102, and the luminance value of the minimum luminance symmetric pixel 104. A second luminance difference that is a difference between the two is calculated (step S108).
Further, the shadow direction determination unit 145 compares the first luminance difference and the second luminance difference calculated in step S108 (step S109). In step S109, when the shadow direction determination unit 145 determines that the first luminance difference is smaller than the second luminance difference, the shadow direction determination unit 145 determines that the direction from the maximum luminance pixel 101 toward the maximum luminance symmetric pixel 103 is the shadow direction. A scanning direction d parallel to the shadow direction is set (step S110).
At this time, the defect detection means 146 detects the minimum luminance pixel 102 as a dark defect (step S111).

In step S109, when the shadow direction determination unit 145 determines that the second luminance difference is smaller than the first luminance difference, the shadow direction determination unit 145 determines that the direction from the minimum luminance symmetric pixel 104 toward the minimum luminance pixel 102 is the shadow direction. Then, a scanning direction d parallel to the shadow direction is set (step S112).
In this case, the defect detection means 146 detects the maximum luminance pixel 101 as a bright defect (step S113).

  Thereafter, the defect detection means 146 of the detection control apparatus 10 performs a light defect detection process (step S114) and a dark defect detection process (step S115) on each inspection area 100, and the inspection area 100 Detect bright and dark defects. Further, the defect detection means 146 generates a binarized image in which these bright defects and dark defects are set to +1 and other normal pixels are set to 0 (step S116).

  Thereafter, the output control unit 147 outputs, to the preset output unit, the binarized image generated in Step S116, the address indicating the pixel position of the defective pixel, and the like, and outputs the defect detection result. (Step S117).

[Light defect detection processing]
Here, the bright defect detection process of step S114 in the defect detection operation of the defect detection apparatus 1 will be described. FIG. 10 is a flowchart of the bright defect detection process.

In the description of the bright defect detection process in step S114, the pixel position of each inspection target pixel in the inspection region 100 is defined as shown in FIG. FIG. 11 is a diagram illustrating the pixel position of the inspection region 100.
That is, mmax virtual scanning lines D are drawn in order along the scanning orthogonal direction d ′ orthogonal to the scanning direction d in the inspection region 100, and numbers 1 to mmax are sequentially assigned from the negative direction side of the scanning orthogonal direction d ′. Assume that there is an mth virtual scanning line D that overlaps the pixel to be inspected. Further, it is assumed that numbers from 1 to nmax are assigned to pixels on the mth virtual scanning line D (virtual scanning line Dm) from the negative direction side in the scanning direction, and the pixel to be inspected is nth. In the inspection area 100 as described above, as shown in FIG. 11, the pixel position of the inspection target pixel can be displayed as M (n, m) using the scanning variable n and the scanning axis variable m.
In the bright defect detection procedure, the defect detection unit 146 initializes the scanning variable n and the scanning axis variable m, and sets n = 1 and m = 1 (step S201).

  Then, the defect detection means 146 calculates the luminance difference P between the pixel Mn located at M (n, m) and the pixel Mn + 1 located at M (n + 1, m) (step S202). That is, the defect detection unit 146 calculates a luminance difference P obtained by subtracting the luminance value of the pixel Mn from the luminance value of the pixel Mn + 1.

  Here, the defect detection means 146 determines whether or not the brightness difference P substantially matches, for example, a shadow influence amount Q set in advance (step S203). In step S203, when the difference between the luminance difference P and the shadow influence amount Q is within a predetermined threshold value, the defect detection unit 146 determines that they are substantially the same, and determines that the pixel Mn + 1 is a normal pixel.

On the other hand, if the luminance difference P does not coincide with the shadow influence amount Q in step S203, it is determined whether the luminance difference P is a positive value (step S204).
In step S204, when the defect detection unit 146 determines that the luminance difference P is a positive value, the defect detection unit 146 detects the pixel Mn + 1 as a bright defect (step S205).
In addition, after step S205, the defect detection unit 146 adds 1 to the scanning variable n (step S206), moves the inspection target pixel, and calculates the luminance difference P in the same manner as the process of step S202 (step S202). S207).

  After this step S207, the defect detection means 146 determines whether or not the luminance difference P matches the shadow area influence amount Q (step S208). If it is determined that the brightness difference P substantially matches, the processing of step S205 is performed. That is, since the luminance value of the inspection target pixel adjacent to the pixel that becomes the bright defect has substantially the same luminance value as the pixel of the bright defect, the defect detection unit 146 has the inspection target adjacent to the pixel that becomes the bright defect. The pixel is also judged to be a light defect.

  Further, when it is determined in step S208 that the luminance difference P does not substantially coincide with the shadow portion influence amount Q, the defect detection unit 146 determines whether the luminance difference P is a negative value (step S209). In step S209, when the defect detection unit 146 determines that the luminance difference P is a positive value, the defect detection unit 146 performs the process of step S205. That is, the defect detection means 146 has an inspection target pixel adjacent to the pixel that is the bright defect because the luminance value of the inspection target pixel adjacent to the pixel that is the bright defect is larger than the bright defect pixel. Judged as a light defect.

Further, when the defect detection unit 146 determines in step S209 that the luminance difference P is a negative value, the absolute value of the luminance difference P is a value sufficiently larger than the absolute value of the shadow influence amount Q. It is determined whether or not there is (step S210). Specifically, the defect detection means 146 determines whether or not the absolute value of the luminance difference P is a value within a predetermined range centered on the absolute value of the luminance difference P calculated in step S202. In step S210, when the defect detection unit 146 determines that the absolute value of the luminance difference P is smaller than a value within a predetermined range centered on the absolute value of the luminance difference P calculated in step S202, the step The process of S205 is performed. That is, the defect detection unit 146 has a luminance value of an inspection target pixel adjacent to a pixel that becomes a bright defect smaller than that of a pixel that becomes a bright defect, but has a higher luminance value than a normal pixel. The pixel to be inspected adjacent to the pixel that becomes the bright defect is also determined to be the bright defect.
On the other hand, when it is determined in step S210 that the absolute value of the brightness difference P is a value within a predetermined range centered on the absolute value of the brightness difference P calculated in step S202, the defect detection means 146 It is determined that Mn + 1 is a normal pixel.

Returning to the description of the processing in step S204, when the defect detection unit 146 determines that the luminance difference P is a negative value, the defect detection unit 146 detects the pixel Mn + 1 as a dark defect candidate (step S211).
After this step S211, the defect detection means 146 adds 1 to the scanning variable n (step S212), moves the inspection target pixel, and calculates the luminance difference P in the same manner as the process of step S202 (step S213). ).

  After this step S213, the defect detection means 146 determines whether or not the luminance difference P matches the shadow area influence amount Q (step S214). If it is determined that they substantially match, the processing of step S211 is performed. That is, the defect detection unit 146 determines that the inspection target pixel adjacent to the pixel that is a dark defect candidate has substantially the same luminance value as the dark defect candidate pixel. It is judged that.

  Further, when the defect detection unit 146 determines in step S214 that the luminance difference P does not substantially coincide with the shadow influence amount Q, the defect detection unit 146 determines whether the luminance difference P is a positive value (step S215). In step S215, when the defect detection unit 146 determines that the luminance difference P is a negative value, the defect detection unit 146 performs the process of step S211. That is, the defect detection unit 146 determines that the inspection target pixel is also a dark defect candidate because the luminance value of the inspection target pixel adjacent to the pixel that is the dark defect candidate is smaller than the dark defect pixel. To do.

Further, when the defect detection means 146 determines in this step S215 that the luminance difference P is a positive value, the absolute value of the luminance difference P is a value sufficiently larger than the absolute value of the shadow influence amount Q. It is determined whether or not there is (step S216). Specifically, the defect detection means 146 determines whether or not the absolute value of the luminance difference P is a value within a predetermined range centered on the absolute value of the luminance difference P calculated in step S202. In step S216, when the defect detection means 146 determines that the absolute value of the luminance difference P is smaller than a value within a predetermined range centered on the absolute value of the luminance difference P calculated in step S202, The process of S211 is performed. That is, the defect detection unit 146 has a luminance value of the inspection target pixel adjacent to the pixel that is a dark defect candidate, which is larger than that of the pixel that is a dark defect candidate, but is a luminance value compared to a pixel corresponding to a normal portion. Therefore, it is determined that this pixel to be inspected is also a dark defect candidate.
On the other hand, if it is determined in step S216 that the absolute value of the brightness difference P is a value within a predetermined range centered on the absolute value of the brightness difference P calculated in step S202, the defect detection means 146 It is determined that Mn + 1 is a normal pixel.

When the defect detection unit 146 determines that the pixel Mn + 1 is a pixel corresponding to a normal part having no defect after step S203, step S210, and step S216, 1 is added to the scanning variable n (step S217), it is determined whether or not the scanning variable n has reached nmax (step S218). If it is determined in step S218 that n = nmax is not satisfied, the process of step S202 is performed. Thereby, the defect detection process of step S202-step S216 is implemented sequentially with respect to the test object pixel adjacent along a scanning direction.
When the inspection target pixel to be inspected reaches the end of the inspection region 100 and it is determined in step S205 that the scanning variable n = nmax, 1 is added to the scanning axis variable m (step S219), and the scanning axis It is determined whether or not the variable m has reached mmax (step S220). If it is determined in step S220 that m = mmax is not satisfied, the scan variable n is initialized, that is, 1 is set in the scan variable n (step S221), and the process of step S202 is performed again. As a result, the defect detection process of step S202 to step S216 is sequentially performed on the inspection target ◯ on all virtual scanning lines D in the inspection region 100. If the defect treatment means ◯ determines that the scanning axis variable m has reached mmax in the process of step S220, the series of bright defect detection processes is terminated.

[Dark defect detection processing]
Next, the dark defect detection process in step S115 in the defect detection method of the defect detection apparatus 1 will be described with reference to the drawings. FIG. 12 is a flowchart of dark defect detection processing in the defect detection operation of the defect detection apparatus 1.
The dark defect detection process is simplified since the process is substantially the same as the bright defect detection process.

In the dark defect detection process, the defect detection means 146 first initializes the scanning variable n and the scanning axis variable m, and sets n = nmax and m = 1 (step S301).
Then, the defect detection means 146 calculates the luminance difference P between the pixel Mn located at M (n, m) and the pixel Mn-1 located at M (n−1, m) (step S302). That is, the defect detection unit 146 calculates a luminance difference P obtained by subtracting the luminance value of the pixel Mn from the luminance value of the pixel Mn-1.

  Here, the defect detection means 146 determines whether or not the luminance difference P substantially matches, for example, a shadow influence amount Q set in advance (step S303). In step S303, when the difference between the luminance difference P and the shadow influence amount Q is within a predetermined threshold value, the defect detection unit 146 determines that they are substantially the same, and determines that the pixel Mn-1 is a normal pixel.

On the other hand, if the luminance difference P does not coincide with the shadow influence amount Q in step S303, it is determined whether the luminance difference P is a negative value (step S304).
In step S304, when the defect detection unit 146 determines that the luminance difference P is a negative value, the defect detection unit 146 detects the pixel Mn−1 as a dark defect (step S305).
After step S305, the defect detection unit 146 subtracts 1 from the scanning variable n (step S306), moves the inspection target pixel, and calculates the luminance difference P in the same manner as the process of step S302 (step S302). S307).

  After this step S307, the defect detection means 146 determines whether or not the luminance difference P matches the shadow area influence amount Q (step S308). If it is determined that the brightness difference P substantially matches, the processing of step S305 is performed. That is, since the luminance value of the inspection target pixel adjacent to the pixel that is the dark defect has substantially the same luminance value as the pixel of the dark defect, the defect detection unit 146 determines that the inspection target pixel is also a dark defect. to decide.

  Further, when the defect detection unit 146 determines in step S308 that the luminance difference P does not substantially coincide with the shadow effect amount Q, the defect detection unit 146 determines whether the luminance difference P is a positive value (step S309). In step S309, when the defect detection unit 146 determines that the luminance difference P is a negative value, the defect detection unit 146 performs the process of step S305. That is, the defect detection unit 146 determines that the inspection target pixel is also a dark defect because the luminance value of the inspection target pixel adjacent to the pixel that is a dark defect is smaller than the dark defect.

Further, when the defect detection means 146 determines in this step S309 that the luminance difference P is a negative value, the absolute value of the luminance difference P is sufficiently larger than the absolute value of the shadow effect amount Q. It is determined whether or not there is (step S310). Specifically, the defect detection means 146 determines whether or not the absolute value of the luminance difference P is a value within a predetermined range centered on the absolute value of the luminance difference P calculated in step S302. In step S310, when the defect detection unit 146 determines that the absolute value of the luminance difference P is smaller than a value within a predetermined range centered on the absolute value of the luminance difference P calculated in step S302, the step The process of S305 is performed. In other words, the defect detection unit 146 has a luminance value of the inspection target pixel adjacent to the pixel that becomes the dark defect larger than that of the pixel that becomes the dark defect, but the luminance value is smaller than that of the normal pixel. The pixel to be inspected is also determined to be a dark defect.
On the other hand, when it is determined in step S310 that the absolute value of the luminance difference P is a value within a predetermined range centered on the absolute value of the luminance difference P calculated in step S302, the defect detection unit 146 It is determined that Mn-1 is a normal pixel.

Returning to the description of the processing in step S304, when the defect detection unit 146 determines that the luminance difference P is a negative value, the defect detection unit 146 detects the pixel Mn−1 as a bright defect candidate (step S311).
After this step S311, the defect detection means 146 subtracts 1 from the scanning variable n (step S312), moves the inspection target pixel, and calculates the luminance difference P as in the process of step S302 (step S313). ).

  After this step S313, the defect detection means 146 determines whether or not the luminance difference P matches the shadow portion influence amount Q (step S314). If it is determined that the brightness difference P substantially matches, the processing of step S311 is performed. That is, the defect detection means 146 has the luminance value of the pixel to be inspected adjacent to the pixel to be a bright defect candidate substantially the same as the pixel of the bright defect candidate. It is judged that.

  Further, when the defect detection unit 146 determines in step S314 that the luminance difference P does not substantially coincide with the shadow effect amount Q, the defect detection unit 146 determines whether the luminance difference P is a negative value (step S315). In step S315, when the defect detection unit 146 determines that the luminance difference P is a negative value, the defect detection unit 146 performs the process of step S311. That is, the defect detection unit 146 determines that the inspection target pixel is also a bright defect candidate because the luminance value of the inspection target pixel adjacent to the pixel that is a bright defect candidate is larger than the bright defect pixel. To do.

Further, when the defect detection means 146 determines in this step S315 that the luminance difference P is a positive value, the absolute value of the luminance difference P is sufficiently larger than the absolute value of the shadow effect amount Q. It is determined whether or not there is (step S316). Specifically, the defect detection means 146 determines whether or not the absolute value of the luminance difference P is a value within a predetermined range centered on the absolute value of the luminance difference P calculated in step S302. In step S316, when the defect detection unit 146 determines that the absolute value of the luminance difference P is smaller than a value within a predetermined range centered on the absolute value of the luminance difference P calculated in step S302, The process of S311 is performed. That is, the defect detection unit 146 has a luminance value of an inspection target pixel adjacent to a pixel that is a bright defect candidate smaller than that of a pixel that is a bright defect candidate. Therefore, it is determined that this inspection target pixel is also a bright defect candidate.
On the other hand, when it is determined in step S316 that the absolute value of the luminance difference P is a value within a predetermined range centered on the absolute value of the luminance difference P calculated in step S302, the defect detection unit 146 It is determined that Mn-1 is a normal pixel.

When the defect detection unit 146 determines that the pixel Mn-1 is a normal pixel corresponding to a normal part having no defect after step S303, step S310, and step S316, 1 is subtracted from the scan variable n. (Step S317), it is determined whether or not the scanning variable n is 1 (Step S318). If it is determined in step S318 that n = 1 is not satisfied, the process of step S302 is performed. Thereby, the defect detection process of step S302-step S316 is implemented sequentially with respect to the test object pixel adjacent along a scanning direction.
When the inspection target pixel to be inspected reaches the end of the inspection region 100 and it is determined in step S305 that the scanning variable n = 1, 1 is added to the scanning axis variable m (step S319), and the scanning axis It is determined whether or not the variable m has reached mmax (step S320). If it is determined in step S320 that m = mmax is not satisfied, the scan variable n is initialized, that is, nmax is set in the scan variable n (step S321), and the process of step S302 is performed again. Thereby, the defect detection process of step S302-step S316 is implemented sequentially with respect to the test object (circle) on all the virtual scanning lines D in the test | inspection area | region 100. FIG. If the defect treatment means ◯ determines that the scanning axis variable m has reached mmax in the process of step S320, the series of bright defect detection processes is terminated.

  In the defect detection apparatus 1 described above, the description of the configuration and the defect detection method for performing the defect detection process on the inspection area 100 determined to have no shadow is omitted, but actually, the inspection without the shadow is not performed. The detection process of the bright defect and the dark defect is also performed on the region 100. This defect detection process can be performed by a conventional defect detection method because there is no change in luminance value due to the influence of shadows.

[Effects of defect detection device]
As described above, in the defect detection apparatus 1 of the above-described embodiment, the captured image is divided into a plurality of minute inspection areas 100, the standard deviation of these inspection areas 100 is obtained, and this standard deviation is equal to or less than the luminance threshold value. An inspection area 100 having a shadow portion is detected. Then, a scanning direction along the shadow direction is set in these inspection regions 100, and luminance values are compared along the scanning direction to detect pixels (bright defect, dark defect) corresponding to the defect.
By making the shadow direction of the shadow part parallel to the scanning direction, the amount of change in luminance accompanying the change in shade of the shadow part can be made substantially constant. Therefore, by subtracting this luminance change amount, the influence of shadow shading can be easily excluded, so that erroneous detection due to the influence of shadow can be prevented and defect detection accuracy can be improved. Therefore, foreign matter adhering to the wafer or the like can be detected well, and production management and manufacturability can be further improved.

  In addition, the region dividing unit 142 draws a horizontal line and a vertical line evenly on the captured image, and divides the image into a plurality of small square inspection regions 100. For this reason, various calculations such as the calculation of the standard deviation in each inspection region 100 can be easily performed, and the processing load can be reduced. Further, since the shadow direction in each inspection region 100 can be set to a certain direction by dividing the image into a plurality of minute regions, the shadow direction can be easily recognized and the scanning direction d can be easily set. Since the influence amount of the shadow portion in 100 becomes small, the influence of the shadow portion can be more appropriately excluded, and the defect detection with high accuracy can be performed.

Then, the shadow direction determination unit 145 detects the maximum luminance pixel 101 and the minimum luminance pixel 102 at the outer periphery of the inspection region 100, and these maximum luminance pixel 101 and minimum luminance pixel 102 are at the center point O of the inspection region 100. On the other hand, when it is determined that the pixel exists at a point-symmetrical position, the direction from the maximum luminance pixel 101 to the minimum luminance pixel 102 is a shadow direction (a direction in which the shadow gradually increases), and the scanning direction d is set along the shadow direction. Set.
In the inspection object X, when there is an edge portion or a concave portion and a shadow is generated in the inspection object range, when the inspection object range is divided into minute regions, as described above, the shadow is along a predetermined shadow direction. The shade changes. Accordingly, in the inspection area 100, a portion having the darkest shadow density (a pixel having the lowest luminance) and a portion having a light shadow density (the pixel having the highest luminance) in any of the pixels arranged on the outer periphery of the inspection area 100. ) Occurs. Therefore, by setting the direction from the maximum luminance pixel to the minimum luminance pixel as the scanning direction d, it is possible to easily set the scanning direction d substantially coincident with the shadow direction, and the shadow portion influence amount in the scanning direction d can be set as follows. It can be made substantially constant.

Further, the defect detection unit 146 calculates a brightness defect by calculating a luminance difference obtained by subtracting the luminance value of the pixel Mn on the negative direction side from the luminance value of the pixel Mn + 1 on the positive direction side in the scanning direction along the scanning direction d. A dark defect is detected by calculating a luminance difference obtained by subtracting the luminance value of the pixel on the negative direction side from the luminance value of the pixel on the positive direction side in the reverse scanning direction.
For this reason, when the luminance value of the pixel Mn is subtracted from the luminance value of the pixel Mn + 1 as described above in the detection of the bright defect, the luminance value tends to gradually decrease due to the influence of the shadow portion. Thereby, in a normal pixel, a luminance difference becomes a negative value. On the other hand, when there is a bright defect, the brightness value of the bright defect is large and fluctuates to the positive value side, so the brightness difference becomes a positive value. Therefore, the defect detection unit 146 can easily detect a bright defect by detecting a pixel in which the luminance difference has a positive value.
The same applies to the detection of a dark defect. When a dark defect is detected, if the luminance value of the pixel Mn-1 is subtracted from the luminance value of the pixel Mn, the luminance difference tends to gradually increase due to the influence of the shadow portion. is there. Therefore, the luminance difference is a positive value in the normal pixel. On the other hand, when there is a dark defect, the luminance value of the dark defect is large and fluctuates to the negative value side, so the luminance difference becomes a negative value. Thereby, the defect detection means 146 can detect a dark defect easily by detecting the pixel from which this brightness | luminance difference becomes a negative value.

When the shadow direction determination unit 145 determines that the maximum luminance pixel 101 and the minimum luminance pixel 102 are present at positions that are not point-symmetric, the shadow direction determination unit 145 recognizes the maximum luminance symmetric pixel 103 and the minimum luminance symmetric pixel 104, and The first luminance difference, which is the difference between the luminance value of the first luminance symmetry and the luminance value of the maximum luminance symmetric pixel 103, and the second luminance difference, which is the difference between the luminance value of the minimum luminance pixel 102 and the luminance value of the minimum luminance symmetric pixel 104, are calculated. Then, when the first luminance difference is smaller than the second luminance difference, the shadow determination unit ○ sets the direction from the maximum luminance pixel 101 to the maximum luminance symmetric pixel 103 as the scanning direction d, and the second luminance difference is the first luminance difference. When the difference is smaller than the luminance difference, the direction from the minimum luminance symmetric pixel 104 toward the minimum luminance pixel 102 is set as the scanning direction d.
For this reason, when any of the inspection target pixels arranged on the outer periphery of the inspection region 100 has a bright defect or a dark defect, the luminance value of the pixel corresponding to the defective portion is significantly different from that of the normal pixel. Become. On the other hand, as described above, since the shadow direction is constant in the minute inspection region 100, when there is no defect, the maximum luminance pixel 101 and the minimum luminance pixel 102 are arranged at positions that are point-symmetric with respect to each other. Therefore, when the maximum luminance pixel 101 and the minimum luminance pixel 102 are not point-symmetric with respect to the center point O, it can be considered that one of them is a defect. Therefore, as described above, when the first luminance difference and the second luminance difference are calculated, one of the first luminance difference and the second luminance difference calculated by the defective pixel exceeds the luminance difference due to the influence of the shadow portion. Large value. That is, as in the present invention, between the first luminance difference and the second luminance difference, one of the pixels with the smaller value is set as the shadow direction, and the scanning direction is set, so that the scanning direction is in a direction substantially coincident with the shadow direction. Can be set, and good defect detection excluding the influence of shadows can be performed.

At this time, the defect detection means 146 detects the maximum luminance pixel 101 or the minimum luminance pixel 102 having a larger value among the first luminance difference and the second luminance difference as a defect.
For this reason, even when there is a defective pixel on the outer periphery of the inspection region 100, it can be detected with high accuracy. That is, as described above, when calculating the luminance difference between adjacent pixels along the scanning direction, if the defective pixel is located at one end in the scanning direction d (the end of the detection scanning start position), For example, a separate defect detection method is required, such as comparing the luminance value of a pixel with a threshold value. On the other hand, in the configuration as described above, it is possible to easily detect a defective portion existing on the outer periphery of the inspection region 100.

Further, when performing the bright defect detection along the scanning direction d, the defect detection unit 146 regards the defect as a dark defect candidate if the luminance difference P changes from the shadow effect amount Q to a value smaller than the shadow effect amount Q. Even if a positive value is calculated when a bright defect is detected in the next pixel, it is not detected as a bright defect. Similarly, when performing the dark defect detection process along the reverse scanning direction −d, if the luminance difference P changes from the shadow influence amount Q to a value larger than the shadow influence amount Q, it is regarded as a bright defect candidate, Even if a negative value is calculated when a dark defect is detected in the next pixel, it is not detected as a dark defect.
For this reason, when a positive value is generated due to the influence of a dark defect candidate during the bright defect detection process, it is not detected as a bright defect, and thus erroneous detection can be prevented. Similarly, when a negative value appears due to the influence of a bright defect candidate during dark defect detection processing, it is not detected as a dark defect, and erroneous detection can be prevented. Therefore, a more accurate defect detection process excluding the influence of the shadow portion can be performed.

Further, when performing the bright defect detection along the scanning direction d, the defect detection unit 146 determines that the dark defect candidate is detected when the luminance difference P changes from the shadow effect amount Q to a value smaller than the shadow effect amount Q. Therefore, even when the shadow influence amount Q is calculated when a bright defect is detected in the next pixel, this pixel is recognized as a dark defect candidate. Similarly, when performing the dark defect detection process along the reverse scanning direction −d, if the luminance difference P changes from the shadow influence amount Q to a value larger than the shadow influence amount Q, it is regarded as a bright defect candidate, When the shadow influence amount Q is calculated when a dark defect is detected in the next pixel, it is recognized as a bright defect candidate.
For this reason, even when the shadow influence amount Q is calculated as the luminance difference P, if the previous pixel is recognized as a bright defect candidate or a dark defect candidate, as described above, these bright defect candidates and dark defect candidates. Can be recognized as continuous. Therefore, when a positive value is calculated as the luminance difference P at the time of detecting a bright defect in a pixel next to a dark defect candidate, it is not recognized as a bright defect, and erroneous detection can be better prevented. Similarly, even when a negative value is calculated as the luminance difference P when a dark defect is detected in a pixel next to a bright defect candidate, it is not detected as a dark defect, and erroneous detection can be prevented better.

  Then, after the positive value indicating the bright defect is calculated as the luminance difference P during the bright defect detection process, the defect detection unit 146 calculates the shadow effect amount Q when the next pixel is detected. Pixels are also detected as bright defects. Similarly, after the negative value indicating the dark defect is calculated as the luminance difference P during the dark defect detection process, the defect detection unit 146 calculates the shadow effect amount Q when detecting the next pixel. This pixel is also detected as a dark defect. For this reason, even when a bright defect or a dark defect continues, these defects can be detected accurately, and the defect detection accuracy can be further improved.

[Other embodiments]
It should be noted that the present invention is not limited to the above-described embodiment, and modifications, improvements, and the like within the scope that can achieve the object of the present invention are included in the present invention.

  For example, in the defect detection apparatus 1 of the above embodiment, a structure such as a semiconductor wafer is exemplified as the inspection target X. For example, an image is projected from a liquid crystal projector onto the screen, and the image on the screen is captured by the imaging unit. It is good also as a structure which images and detects a defective pixel of a liquid crystal. In this case, it is possible to reduce the influence of shadows due to creases, wrinkles, and deflections on the screen.

  Further, the defect detection unit 146 may be configured to calculate the shadow influence amount in the normal pixel along the scanning direction d, and to calculate an ideal luminance value in each pixel based on the shadow influence amount. In this configuration, a difference value between the calculated ideal luminance value and the luminance value of each pixel is actually calculated, and when this difference value is equal to or larger than a predetermined value set in advance than the ideal luminance value, When the difference value is less than or equal to a predetermined value set in advance from the ideal luminance value, it is detected as a dark defect. Even in this case, as in the above embodiment, defect detection that excludes the influence of shadows can be performed, and light and dark defects can be detected simultaneously, so that defects can be detected more efficiently. Can do. In addition, it is possible to detect a defect with high accuracy even when a bright defect and a dark defect are adjacent to each other along the scanning direction d, or with respect to pixels on the outer peripheral edge of the inspection region 100.

Further, the defect detection unit 146 performs defect detection of each inspection target pixel along the scanning direction d during the bright defect detection processing, and detects defects of the inspection target pixel along the reverse scanning direction −d during the dark defect detection processing. Although the detection is performed, for example, a configuration may be adopted in which processing is performed along the reverse scanning direction -d during the bright defect detection processing and along the scanning direction d during the dark defect detection processing, and further, defect detection is performed in the same direction in both. It is good.
When the dark defect detection process is detected along the scanning direction d, a dark defect is detected when the luminance difference P is a negative value and the absolute value is larger than the absolute value of the shadow influence amount Q by a predetermined value or more. . With this configuration, dark defects can be detected as in the above embodiment.

  In addition, the specific structure and procedure for carrying out the present invention can be appropriately changed to other structures and the like within a range in which the object of the present invention can be achieved.

It is a block diagram which shows the outline of the defect detection apparatus of one Embodiment which concerns on this invention. It is a block diagram which shows the outline of the various programs expand | deployed by CPU of the defect detection apparatus of the said embodiment. In the said embodiment, it is a figure which shows the state which divided | segmented the picked-up image of the test target object into several test area | region. In the said embodiment, it is a figure which shows an example which shows the minimum brightness | luminance pixel in a test | inspection area | region, the maximum brightness | luminance pixel, and a shadow direction. In the said embodiment, it is a figure which shows the minimum luminance pixel, the maximum luminance pixel, and shadow direction in another test | inspection area | region. When there is a bright defect in a part of the inspection target pixel along the scanning direction, the luminance value of each inspection target pixel, the luminance difference between adjacent pixels when scanned along the scanning direction, and along the reverse scanning direction It is a figure which shows an example of the luminance difference between the adjacent pixels at the time of scanning. When there is a dark defect in a part of the inspection target pixel along the scanning direction, the luminance value of each inspection target pixel, the luminance difference between adjacent pixels when scanned along the scanning direction, and along the reverse scanning direction It is a figure which shows the luminance difference between the adjacent pixels at the time of scanning. It is a figure which shows the example of the binarized image showing a defect site | part. It is a flowchart of the defect detection operation | movement of the defect detection apparatus of the said embodiment. It is a flowchart of the bright defect detection process in the defect detection operation | movement of the said embodiment. It is a figure which shows the pixel position of a test | inspection area | region. It is a flowchart of the dark defect detection process in the defect detection operation | movement of the said embodiment.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Defect detection apparatus, 3 ... Imaging part as imaging means, 100 ... Inspection area | region, 101 ... Maximum brightness pixel, 102 ... Minimum brightness pixel, 103 ... Maximum brightness symmetry pixel, 104 ... Minimum brightness symmetry pixel, 142 ... Image division Area dividing means as means, 143 ... standard deviation calculating means, 144 ... shadow part extracting means that also functions as defect candidate area detecting means, 145 ... shadow direction determining means that also functions as scanning direction setting means, 146 ... defect detecting means , D: scanning direction, -d: reverse scanning direction, O: center point of inspection region, X: inspection object.

Claims (7)

Imaging means for capturing an image of an inspection object and acquiring a captured image;
Image dividing means for dividing the captured image into a plurality of inspection regions;
A standard deviation calculating means for calculating the standard deviation of all the luminances of the inspection target pixels in each inspection region;
A defect candidate area detecting means for detecting an area where the standard deviation is equal to or greater than a predetermined area standard deviation threshold as a defect candidate area;
A scanning direction setting means for setting a scanning direction along a shadow direction for the defect candidate region;
Defect detection means for calculating a luminance change amount of each inspection target pixel along the scanning direction and detecting an inspection target pixel whose luminance change amount is a predetermined defect threshold or more as a defective pixel;
A defect detection apparatus comprising: In the defect detection apparatus according to claim 1,
The image dividing means equally divides the inspection target portion of the captured image into a horizontal direction and a vertical direction orthogonal to the horizontal direction, and each of the equally divided areas is set as the inspection area. Defect detection device. In the defect detection apparatus according to claim 1 or 2,
The scanning direction setting means detects a maximum luminance pixel having the maximum luminance and a minimum luminance pixel having the minimum luminance among the pixels arranged at the outer peripheral edge of the defect candidate region, and these maximum luminance pixels and When the minimum luminance pixel is located at a point-symmetrical position with respect to the center point of the defect candidate area, a line direction connecting the maximum luminance pixel and the minimum luminance pixel is set as the scanning direction. The defect detection apparatus according to claim 3,
The defect detection means scans along the scanning direction from the maximum luminance pixel toward the minimum luminance pixel, detects a bright defect whose luminance change amount is larger than other inspection target pixels, and the minimum luminance pixel A defect detection apparatus that scans along the scanning direction from the pixel toward the maximum luminance pixel to detect a dark defect whose luminance is smaller than that of other inspection target pixels. In the defect detection apparatus according to claim 3 or claim 4,
The scanning direction setting means, when the maximum luminance pixel and the minimum luminance pixel detected from pixels arranged at the outer periphery of the defect candidate region are not point-symmetric with respect to the center point of the defect candidate region, Detect a maximum luminance symmetric pixel that is point symmetric with respect to the center point of the defect candidate area of the maximum luminance pixel, and a minimum luminance symmetric pixel that is point symmetric with respect to the center point of the defect candidate area of the minimum luminance pixel. Calculating a luminance difference between the maximum luminance pixel and the maximum luminance symmetric pixel, a luminance difference between the minimum luminance pixel and the minimum luminance symmetric pixel, and a pixel difference between the maximum luminance pixel and the maximum luminance symmetric pixel; and A direction connecting any one of the pixels between the minimum luminance pixel and the minimum luminance symmetric pixel where the value of the luminance difference is small is set as the scanning direction. A defect detection device. The defect detection apparatus according to claim 5,
The defect detection means includes a pixel having a larger value of the luminance difference among pixels between the maximum luminance pixel and the maximum luminance symmetric pixel and between the minimum luminance pixel and the minimum luminance symmetric pixel. One of the corresponding maximum luminance pixel and the minimum luminance pixel is detected as a defective pixel. A defect detection method for detecting a defect to be inspected,
Image the inspection object to obtain a captured image,
This captured image is divided into a plurality of inspection areas,
Calculate the standard deviation of the brightness of all the pixels to be inspected in these inspection areas,
A region where the calculated standard deviation is equal to or greater than a predetermined region standard deviation threshold is detected as a defect candidate region,
For the detected defect candidate area, set the scanning direction, recognize the change in luminance of each pixel to be inspected along the scanning direction, calculate the amount of luminance change,
A defect detection method characterized by detecting, as a defective pixel, an inspection target pixel in which the calculated luminance change amount is equal to or greater than a predetermined defect threshold.

Images