JP2019100937A - Defect inspection device and defect inspection method - Google Patents

Abstract

To provide a defect inspection device and a defect inspection method capable of improving a defect inspection accuracy.SOLUTION: A defect inspection device prepares a plurality of low resolution images each obtained by lowering a resolution of a picked-up image imaging an inspection target, and extracts a defect candidate according to an image synthesizing the plurality of low resolution images in a non-binarized state.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a defect inspection apparatus that finds a defect of an inspection target using an image obtained by imaging the inspection target, and a defect inspection method.

  Conventionally, while changing the phase by a division means for dividing a divided area of a predetermined size from a captured image obtained by imaging an inspection object, a phase changing means for changing a phase (position) for dividing the divided area, and the phase changing means For each of the plurality of divided areas divided by the dividing means, an average value of the luminances of a plurality of pixels contained in the divided area is determined, and the average of the resolution is reduced by replacing the luminances of the plurality of pixels with the average value. There is known an image processing apparatus having an image processing unit (for example, see Patent Document 1 and Patent Document 2).

  Further, in the image processing apparatus, for each of the images reduced in resolution by the averaging means (hereinafter referred to as a low resolution image), after creating quantized data obtained by binarizing each pixel included in the low resolution image, A plurality of pieces of the quantized data are added to create an inspection image, and a region having relatively high or low luminance in the inspection image can be recognized as a defect candidate.

JP 2017-13371 A Patent No. 5821708

By the way, in the above-mentioned conventional defect inspection apparatus, although the low resolution image includes low values and high values of the luminance level, the above luminance level is used to generate the quantization data from the low resolution image. Is converted into a binary value, information on luminance indicating a defect may not be reflected in the quantized data. As a result, the defect candidate is not correctly reflected on the inspection image created by adding the quantized data, and the defect inspection accuracy is lowered.
In addition, when an image whose phase is shifted for each resolution is added, since the number of times the phase is shifted differs depending on the resolution, the weight of defects with a small number of times the phase is shifted less when added as it is. Is also a problem.

  Then, this invention makes it a subject to provide the defect inspection apparatus which can raise the inspection accuracy of a defect, and the defect inspection method in view of this situation.

The defect inspection apparatus of the present invention is
Resolution changing means for creating a plurality of low resolution images having a lower resolution than the captured image based on the captured image obtained by capturing an inspection object;
Defect candidate extracting means for generating a composite image obtained by combining the plurality of low resolution images created by the resolution changing means and extracting a defect candidate from the composite image;
The resolution changing means is
Division area setting means for setting the division area in the captured image while shifting the phase in the captured image;
An average image generation unit that generates an average image in which the average value of the luminance of each pixel in the divided area is set to the luminance value of one pixel when the divided area is set by the divided area setting unit;
Luminance value setting means for setting a luminance value to the average image created by the average image creation means to create a luminance image;
And phase image combining means for combining the plurality of brightness images created by the brightness value setting means to create the low resolution image.
The average image creating unit, the luminance value setting unit, and the phase image combining unit create a plurality of low resolution images having different resolutions while changing the size of the divided region set by the divided region setting unit. Configured as
The defect candidate extraction unit is configured to generate the combined image by combining the plurality of low resolution images created by the phase image combining unit.

  According to the defect inspection apparatus of the above configuration, when creating a low resolution image, the divided images are combined without being binarized, and when creating a combined image, the low resolution image is not binarized. Therefore, the contrast information of the defect remains in the composite image used to extract the defect candidate.

  Therefore, in the defect inspection apparatus, the defect inspection accuracy can be enhanced by making it possible to judge (extract) the presence or absence of the defect candidate in accordance with the contrast information of the composite image.

In the defect inspection apparatus of the present invention,
The luminance value setting means
For the average image,
Extracting a pixel of interest and a pair of adjacent pixels adjacent to both sides of the pixel of interest;
With respect to the luminance value of the target pixel, the luminance value is set based on the luminance difference only when both of the pair of adjacent pixels are both bright or dark, and the luminance value is set to 0 for the other pixels.
It may be configured as follows.

  According to the defect inspection apparatus of the above configuration, when the pixel on one side of the adjacent pixels on both sides is bright with respect to the pixel of interest (point of interest) and the other pixel is dark, the luminance change in the average image is the luminance change of the normal part Therefore, it is possible to prevent the normal part from being excessively detected as a defect.

  In addition, when a pattern is included in the normal part, in the boundary part, a pixel in which one side of both adjacent pixels has the same luminance as the target pixel and the other is bright or dark comes in, but this is also normal. It can be determined as a part.

Furthermore, in the defect inspection apparatus of the present invention,
The phase image synthesizing unit synthesizes and normalizes the luminance image.
The defect candidate extraction unit combines the plurality of low resolution images normalized by the phase image combining unit to create the combined image.

  According to this configuration, it is possible to eliminate the difference in weight due to the difference in the number of times of phase shift in each resolution, and to indicate a defect when combining multiple low resolution images or multiple luminance images. Loss of information can be suppressed.

The defect inspection method of the present invention is
Creating a plurality of low resolution images having a lower resolution than the captured image based on the captured image obtained by capturing the inspection object;
Creating a composite image combining the plurality of low resolution images created in the step of creating a plurality of low resolution images, and extracting a defect candidate from the composite image;
In the step of creating a plurality of low resolution images,
Average image creation processing for creating an average image in which divided areas are set in the captured image while shifting the phase in the captured image, and the average value of the luminance of each pixel in the divided area is the luminance value of one pixel And a luminance value setting process for generating a luminance image having a luminance value set for the average image generated in the average image generation process and each luminance image generated in the luminance value setting process to obtain the low resolution image A plurality of low resolution images having different resolutions are generated by repeatedly executing phase image synthesis processing to be created while changing the size of the divided area when setting the divided area in the captured image. ,
The step of extracting the defect candidate is configured to create the combined image by combining a plurality of the low resolution images.

  According to the defect inspection apparatus of the above configuration, when creating a low resolution image, the divided images are combined without being binarized, and when creating a combined image, the low resolution image is not binarized. Therefore, the contrast information of the defect remains in the composite image used to extract the defect candidate.

  Therefore, in the defect inspection apparatus, the defect inspection accuracy can be enhanced by making it possible to judge (extract) the presence or absence of the defect candidate in accordance with the contrast information of the composite image.

In the defect inspection method of the present invention,
In the luminance value setting process,
Extracting a pixel of interest and a pair of adjacent pixels adjacent to both sides of the pixel of interest with respect to the average image;
The luminance value is set based on the luminance difference only when both of the pair of adjacent pixels are both bright or dark with respect to the pixel value of the pixel of interest, otherwise the luminance value is set to 0. It may be configured.

  According to the defect inspection apparatus of the above configuration, when the pixel on one side of the adjacent pixels on both sides is bright with respect to the pixel of interest (point of interest) and the other pixel is dark, the luminance change in the average image is the luminance change of the normal part Therefore, it is possible to prevent the normal part from being excessively detected as a defect.

  In addition, when a pattern is included in the normal part, in the boundary part, a pixel in which one side of both adjacent pixels has the same luminance as the target pixel and the other is bright or dark comes in, but this is also normal. It can be determined as a part.

Furthermore, in the defect inspection method of the present invention,
The phase image synthesis process synthesizes and normalizes the luminance image.
In the step of extracting the defect candidate, the plurality of low resolution images normalized by the phase image combining process are combined to create the combined image.

  According to this configuration, it is possible to eliminate the difference in weight due to the difference in the number of times of phase shift in each resolution, and to indicate a defect when combining multiple low resolution images or multiple luminance images. Loss of information can be suppressed.

  As described above, according to the defect inspection apparatus and the defect inspection method of the present invention, it is possible to achieve an excellent effect that a defect to be inspected can be found accurately.

FIG. 1 is a block diagram of a defect inspection apparatus according to an embodiment of the present invention. FIG. 2 is a flowchart of the defect inspection apparatus according to the embodiment. (A)-(c) in FIG. 3 is explanatory drawing of the division size of a captured image. (A)-(i) in FIG. 4 is explanatory drawing of how to set the phase of a division | segmentation area | region. FIG. 5 is an explanatory view of a process of creating an average image of the defect inspection apparatus according to the embodiment. In FIG. 6, (a) to (d) are explanatory diagrams of processing for creating a luminance image. (A)-(c) in FIG. 7 is explanatory drawing of a blob process.

  Hereinafter, a defect inspection apparatus according to an embodiment of the present invention will be described with reference to the attached drawings. The defect inspection apparatus is configured to discover a result such as a scratch based on an image obtained by imaging the appearance of the article. Examples of the article include metals, plastic films, cloths, and printed materials.

  The defect inspection apparatus may be provided with, for example, a processing apparatus that processes information or controls an apparatus, a storage apparatus that stores information, an output apparatus such as a display, or the like, such as a personal computer.

  As shown in FIG. 1, the defect inspection apparatus 1 has a resolution higher than that of the captured image based on the image acquisition unit 2 that acquires a captured image obtained by imaging the inspection target and the captured image acquired by the image acquisition unit 2. Resolution changing means 3 for generating a plurality of lowered low resolution images, and defect candidate extracting means for generating a composite image combining the plurality of low resolution images generated by the resolution changing means 3 and extracting defect candidates from the composite image 4 and.

  The image acquisition unit 2 may directly acquire from the imaging device an image of an article to be inspected which is imaged by the imaging device, and is stored in the storage device of the defect inspection apparatus 1 or an external storage device. It may be configured to acquire an image of an inspection target.

  The resolution changing unit 3 sets divided regions in the captured image while shifting the phase in the captured image, and each region in the divided regions when the divided regions are set by the divided region setting unit 30. A brightness image is created by setting the brightness value for the average image created by the average image creation unit 31 that creates an average image using the average value of the pixel brightness as the brightness value of one pixel, and the average image created by the average image creation unit 31 Brightness value setting means 32; and phase image combining means 33 for combining the respective brightness images created by the brightness value setting means 32 to create the low resolution image.

  In the resolution changing unit 3 according to the present embodiment, after the divided area setting unit 30 shifts the phase (phase of the divided area) in the captured image, the average image creating unit 31 shifts the phase in the divided area. By creating an average image with the average value of the luminance of each pixel as the luminance value of one pixel, the luminance value setting means 32 repeatedly performs processing of setting the luminance value to the average image to create a luminance image. It is configured to create a luminance image.

  Further, the resolution changing unit 3 changes the size of the divided area (the size of the average image generated by the average image generating unit 31) by the divided area setting unit 30, and sets the average image generating unit 31 and the luminance value setting. The means 32 and the phase image combining means 33 are configured to create a plurality of low resolution images of different resolutions.

  The division area setting means 30 is configured to be able to set the size of the division area and the phase (position to be set in the division area).

  The divided area setting means 30 can be configured to be able to change the size of the divided area, and furthermore, the size (that is, the initial value of the size of the divided area) set in advance for creating the low resolution image is It may be defined.

  The divided area setting unit 30 has, for example, a size of 3 × 3 pixels (3 vertical pixels × 3 horizontal pixels) 4 × 4 pixels (4 vertical pixels × 4 horizontal pixels) 5 × 5 pixels (5 vertical pixels) The size of the divided area may be changed by increasing the number of pixels in order of the size of 5 pixels horizontally and the number of pixels in order to obtain the same size as the largest defect (the defect predicted to be the largest) shown in the inspection image .

  Furthermore, the divided area setting unit 30 is configured to shift the phase (the position in the captured image) of the divided area by one pixel without changing the size of the divided area (FIG. 4 (a) to FIG. 4 (i)). The phases are shifted in the horizontal direction and in the vertical direction. In FIG. 4A to FIG. 4I, the captured image is indicated by “P1” and the divided areas are indicated by “R1”.

  The average image creation unit 31 creates an average image with the average value of the luminance of each pixel in the divided area set by the divided area setting unit 30 as the luminance value of one pixel (see FIG. 5). In FIG. 5, the captured image is indicated by “P1”, the divided area is indicated by “R1”, and the average image is indicated by “P2”.

  The luminance value setting unit 32 compares the luminance of the pixel serving as a reference with the pixels around the pixel with respect to the average image generated by the average image generation unit 31, and calculates the luminance value calculated by comparison. Convert the luminance value of the corresponding pixel.

  In the phase image synthesizing unit 33, after setting the luminance value to all the average images for each size of the divided area (each resolution), a luminance image of the same size (a luminance image created based on the divided areas of the same size) A low resolution image is created by creating an image combining all of the above and normalizing the pixel values of the image.

  The luminance value setting means 32 will be described in detail. The luminance value setting means 32 extracts a pixel of interest and a pair of adjacent pixels adjacent to both sides of the pixel of interest with respect to the average image, and calculates an average image based on the relationship between the luminance of the pixel of interest and each adjacent pixel. Set the brightness value to

  The pixels extracted by the luminance value setting means 32 as a pair of adjacent pixels may be pixels adjacent to the left and right of the pixel of interest, or may be pixels adjacent to the upper and lower sides of the pixel of interest, It may be pixels adjacent to diagonally above and diagonally below.

  The luminance value setting means 32 obtains the pixel value of the pixel of interest and the pixel values of the adjacent pixels, and the pair of adjacent pixels is brighter or darker than the pixel of interest (FIGS. 6A and 6B) Reference) only, ie, when the brightness (brightness value) of each pair of adjacent pixels is higher or lower than the brightness (brightness value) of the pixel of interest, the average is based on the brightness difference between the pixel of interest and the adjacent pixels. Set the brightness value to the image.

In the present embodiment, assuming that the difference between the pixel value of the target pixel and the pixel value of one adjacent pixel is L1 and the difference between the pixel value of the target pixel and the other adjacent pixel value is L2, the pixel value L of the target pixel Is the calculated value of equation 1.

  This is apparent when the difference L1 between the pixel value of the pixel of interest and one of the adjacent pixels, and the difference L2 between the pixel value of the pixel of interest and the other adjacent pixel value are equal (see FIG. 6A). When the difference L1 between the pixel value of the pixel of interest and one adjacent pixel and the difference L2 between the pixel value of the pixel of interest and the other adjacent pixel value are large because a defect is output (FIG. 6 (b)) Since the reference) may not be a defect, the sensitivity (detection rate of defect candidate) is lowered by setting the luminance value of the pixel of interest to a small value. In addition, when the luminance value calculated by Formula (1) turns into a negative number, the luminance value of a view pixel is set to zero.

  The pixel value L is a value determined by the pixel values of pixels adjacent to the left and right of the pixel of interest, a value determined by pixels adjacent to the upper and lower sides of the pixel of interest, and pixels adjacent to the upper right and lower left of the pixel of interest. It can obtain | require based on four kinds of values of the value calculated | required by, and the value calculated | required by the pixel which adjoins the diagonally upper left and the diagonal lower right. Also, the maximum value of the values may be set as the luminance value, or the total value may be set as the luminance value, or it may be set from two values of only up and down and left and right, or 4 All three values may be used to set the value.

  If one of the pixel values of the pair of adjacent pixels is brighter than the pixel of interest and the other is darker than the pixel of interest (see FIG. 6C), 0 is set as the average image luminance value. Thus, by setting a value that allows the defect candidate extraction means 4 to determine that the defect candidate extraction unit 4 is a normal part to the value of the average image luminance value, it is possible to regard the change in luminance in the average image as the change in luminance of the normal part. It will be.

  Also, when one of the pixel values of the pair of adjacent pixels has the same brightness as the focused pixel and the other is brighter or darker than the focused pixel (see FIG. 6D), the brightness value of the average image is obtained. Set 0. Even in this case, by setting a value that allows the defect candidate extraction means 4 to determine that the defect candidate extraction means 4 is a normal part to the value of the average image luminance value, a pattern is included in the normal part. If one of the adjacent pixels on both sides has the same luminance as the target pixel and another pixel is bright or dark, it can be determined as a normal part. In FIGS. 6A to 6D, the pixel of interest is indicated by “Px1” and the adjacent pixel is indicated by “Px2”.

  As described above, the phase image synthesis unit 33 synthesizes the luminance image to create a low resolution image, but the number of luminance images to be synthesized differs depending on the size of division of the captured image, so the low resolution image is normalized. It is configured to

  For example, when the size of dividing the captured image is 3 × 3 pixels, nine average images (luminance images) are created, and when the size of dividing the captured image is 4 × 4 pixels, 16 average images (luminance Since the image is created, if the size of dividing the captured image becomes large, the average number of images to be combined increases. Therefore, when the average image (luminance image) is added as it is and synthesized, the luminance value after synthesis tends to increase as the size of dividing the captured image increases (FIG. 3A to FIG. c) see In FIGS. 3A to 3C, the captured image is indicated by “P1” and the divided areas are indicated by “R1”.

  In normalization by the phase image synthesizing unit 33, there are a plurality of luminance images having different phases for each size of divided areas, and pixel values of the same coordinates of the luminance images (regions where the luminance images overlap) are added. Thus, the luminance images are combined.

  Then, binarization is performed on the image obtained by combining the luminance image, and defect candidate points are extracted by blob processing. The threshold value for binarization is a large value in proportion to the number of combined images. For the defect candidate point, the brightness is recalculated using the brightness value of the captured image obtained by the image acquisition unit 2.

  Subsequently, an example of the method of recalculating the luminance value will be described. Binarization processing is performed on the image synthesized by adding the pixel values of the luminance image, and the shape of the defect candidate point by the blob processing and the defect candidate Calculate the average brightness of the points.

  At this time, the pixel value of the image before resolution reduction obtained by the image acquisition unit 2 is used as the image for calculating the luminance of the defect candidate point. Furthermore, the shape of the defect candidate point is expanded by dilation processing, and the area before the enlargement is subtracted from the expanded region to obtain the shape of the peripheral portion of the defect candidate point. The average luminance of the peripheral portion of the defect candidate point is calculated using the pixel value of the image before resolution reduction obtained by the image acquiring unit 2 (see FIGS. 7A to 7C). In FIGS. 7A to 7C, the defect candidate point is indicated by “R2”, and the region obtained by enlarging the defect candidate point is indicated by “R3”.

  The difference between the average luminance of the defect candidate point and the average luminance of the peripheral portion of the defect candidate point is taken as the luminance after combination. In the method using this blob processing, if the original low resolution image which is the input is 8 bits, the output image after the synthesis which is the output is also 8 bits, and it is possible to output the defect contrast independent of the number of additions. it can.

  In addition, in the normalization at the time of the synthesis | combination of the luminance image by the phase image synthetic | combination means 33, when synthesize | combining a luminance image by adding the pixel value of the same coordinate (region which a luminance image overlaps) of a luminance image, Since the number of images for each size to be divided (size of divided area) is different, the luminance value after addition may exceed the bit number of the input image, but the luminance of the image after combining the maximum values of pixel values at the same coordinates If it is a value, the maximum luminance of the input luminance image and the maximum luminance of the output composite image can be made the same. Also in this method, the number of bits of the input and output images can be made uniform.

  The phase image synthesizing unit 33 is configured to make the sizes of the respective low resolution images uniform by enlarging the low resolution image according to the size of the captured image obtained by the image acquisition unit 2. In the phase image synthesizing means 33, while the luminances of the low resolution images are normalized, the sizes of the low resolution images are different from each other, but the sizes of the low resolution images are equalized by the processing of the phase image synthesizing means 33. .

  By adding pixel values at the same coordinates to the phase synthesized image having the same image size, it is possible to obtain an image after synthesis of the low resolution image. Note that the maximum value of the pixel values at the same coordinates may be used as the pixel value after synthesis for the phase synthesized image in which the image sizes are aligned.

  The defect candidate extraction unit 4 is configured to determine the presence or absence of a defect (determination as to whether it is a defect or a normal part) based on the luminance value on the image after low resolution image synthesis. The defect candidate extraction unit 4 according to the present embodiment is configured to extract a defective portion by a general binarization process such as a discriminant analysis method, and to determine whether it is a defect or a normal portion by its area and shape. ing.

  The configuration of the defect inspection apparatus 1 according to the present embodiment is as described above. Subsequently, a defect inspection method by the defect inspection apparatus 1 will be described.

  The defect inspection method includes the steps of acquiring a captured image by the image acquisition unit 2, creating a plurality of low resolution images having a lower resolution than the captured image based on the captured image obtained by imaging the inspection target, and the resolution And a step of creating a composite image obtained by combining the plurality of low resolution images created by the change means 3 and extracting a defect candidate from the composite image.

  As shown in FIG. 2, the captured image is acquired by the image acquisition means 2 (S1), and the divided area setting means 30 determines the size of the divided area by the divided size setting means (S2), and further divided into the imaged image Are determined (S3), and division areas are set in the captured image.

  Then, the average image creation unit 31 creates an average image by setting the average value of the luminance of each pixel in the divided area as the luminance value of one pixel (S4).

  Subsequently, the luminance value setting means 32 sets a luminance value for the average image (S5). More specifically, the luminance value setting unit 32 extracts the pixel of interest and a pair of adjacent pixels adjacent to both sides of the pixel of interest from within the average image, and the brightness of the pixel of interest and each adjacent pixel (luminance The brightness value of the average image is set based on the value relationship.

  When each pixel value of a pair of adjacent pixels is brighter or darker than the pixel of interest, that is, the brightness (luminance value) of each pair of adjacent pixels is both higher than the brightness (luminance value) of the pixel of interest or If it is low, the luminance value is set to the average image based on the luminance difference between the pixel of interest and the adjacent pixel. The luminance value of the average image in this case is derived by Equation (1).

  Then, when one pixel value of the pair of adjacent pixels is brighter than the focused pixel and the other is darker than the focused pixel, 0 is set as the luminance value of the average image. The luminance value of the average image is set to 0 also when one of the pair of adjacent pixels has the same brightness as the focused pixel and the other is brighter or darker than the focused pixel.

  Thus, a luminance image in which the luminance value is set to the average image is created.

  In the case of creating a new luminance image from the captured image (No in S6), the phase of the divided area is changed, and processing (S2 to S5) of creating a luminance image is executed again.

  When the luminance image has been created (Yes in S6), the phase image combining means 33 combines the luminance image (S7) to create a low resolution image.

  Furthermore, when creating a new low resolution image with another resolution (No in S8), the size of the divided area is newly set (S2), average image creation means 31, luminance value setting means 32, phase image synthesis The low resolution image creating process by means 33 is executed again (S3 to S7)

  When the low resolution image has been created (Yes in S8), a composite image combining the plurality of low resolution images created by the resolution changing unit 3 is created (S9), and the defect candidate extracting unit 4 uses the composite image Defect candidates are extracted (detected) (S10).

  As described above, according to the defect inspection apparatus 1 and the defect inspection method according to the present embodiment, when creating a low resolution image, division images are synthesized without being binarized, and a synthesized image is generated. In this case, since the low resolution image is synthesized without being binarized, the contrast information of the defect remains in the synthesized image used to extract the defect candidate.

  Therefore, by being able to judge (extract) the presence or absence of a defect candidate in accordance with the contrast information of the composite image, it is possible to achieve an excellent effect that the inspection accuracy of the defect can be enhanced. In the present embodiment, in the output image of the low resolution image combining means, the area serving as a defect candidate point is extracted, so that it is possible to judge the defect by a simple process in the defect detection process.

  In addition, when the pixel value on one side of adjacent pixels on both sides with respect to the pixel of interest (focused point) is bright or dark, the brightness change in the average image is regarded as the brightness change of the normal part, and the normal part is excessive as a defect. It can be prevented from being detected, and when the normal part has a pattern, in the boundary part, a pixel in which one side of the adjacent pixels on both sides has the same luminance as the target pixel and the other is bright or dark However, this can also be determined as a normal part.

  In the present embodiment, by changing the resolution (that is, the size of the divided image), it is possible to create a low resolution image in which the size of the defect and the size of the pixel are the same. When the size of the defect and the size of the pixel (divided image) of the low resolution image are the same by shifting the division position of the divided image from the captured image, the defect is reflected at the center of the pixel of the low resolution image You can get an image. Therefore, a defect can be imaged at the center of one pixel, and thereby it is possible to obtain the maximum defect contrast by comparing the brightness of the pixel where the defect appears with the pixels on both sides thereof.

  Furthermore, it is possible to eliminate the difference in weight due to the difference in the number of times the phase is shifted for each resolution, and when combining multiple low resolution images and multiple luminance images, information on luminance values indicating defects is lost. It is possible to suppress being broken.

  The defect inspection apparatus 1 and the defect inspection method of the present invention are not limited to the above embodiment, and it goes without saying that various changes can be made without departing from the scope of the present invention.

  In the above embodiment, when there is a region in which the luminance images to be combined overlap, the luminance value is obtained by adding the luminance values of the overlapping regions of the luminance images, but the present invention is not limited to this configuration. For example, the maximum value of the luminance values of overlapping regions where the luminance images to be synthesized overlap may be set.

  In the above embodiment, the defect candidate extraction unit 4 is configured to binarize a combined image obtained by combining a plurality of low resolution images, but the present invention is not limited to this configuration. For example, the defect candidate extraction unit 4 may be configured to use an image on which information on luminance values of one low resolution image is reflected as a composite image and to binarize the composite image.

In the above embodiment, although the value of the luminance difference to be set to the average image is calculated by equation (1), the present invention is not limited to this configuration. For example, the value of the luminance difference to be set to the average image by the following equation (2) May be calculated.

  DESCRIPTION OF SYMBOLS 1 ... Defect inspection apparatus, 2 ... Image acquisition means, 3 ... Resolution change means, 4 ... Defect candidate extraction means, 22 ... Image acquisition means, 30 ... Division area setting means, 31 ... Average image creation means, 32 ... Brightness value setting Means, 33 ... phase image synthesizing means

Claims (6)

Resolution changing means for creating a plurality of low resolution images having a lower resolution than the captured image based on the captured image obtained by capturing an inspection object;
Defect candidate extracting means for generating a composite image obtained by combining the plurality of low resolution images created by the resolution changing means and extracting a defect candidate from the composite image;
The resolution changing means is
Division area setting means for setting the division area in the captured image while shifting the phase in the captured image;
An average image generation unit that generates an average image in which the average value of the luminance of each pixel in the divided area is set to the luminance value of one pixel when the divided area is set by the divided area setting unit;
Luminance value setting means for setting a luminance value to the average image created by the average image creation means to create a luminance image;
And phase image combining means for combining the plurality of brightness images created by the brightness value setting means to create the low resolution image.
The average image creating unit, the luminance value setting unit, and the phase image combining unit create a plurality of low resolution images having different resolutions while changing the size of the divided region set by the divided region setting unit. Configured as
The defect inspection apparatus, wherein the defect candidate extraction unit is configured to create the combined image by combining the plurality of low resolution images created by the phase image combining unit. The luminance value setting means
For the average image,
Extracting a pixel of interest and a pair of adjacent pixels adjacent to both sides of the pixel of interest;
The luminance value is set based on the luminance difference only when both of the pair of adjacent pixels are both bright or dark with respect to the luminance value of the pixel of interest, and the other pixels have the luminance value set to 0. The defect inspection device according to 1. The phase image synthesizing unit synthesizes and normalizes the luminance image.
The defect inspection apparatus according to claim 1, wherein the defect candidate extraction unit synthesizes the plurality of low resolution images normalized by the phase image synthesis unit to create the synthesized image. Creating a plurality of low resolution images having a lower resolution than the captured image based on the captured image obtained by capturing the inspection object;
Creating a composite image combining the plurality of low resolution images created in the step of creating a plurality of low resolution images, and extracting a defect candidate from the composite image;
In the step of creating a plurality of low resolution images,
Average image creation processing for creating an average image in which divided areas are set in the captured image while shifting the phase in the captured image, and the average value of the luminance of each pixel in the divided area is the luminance value of one pixel And a luminance value setting process for generating a luminance image having a luminance value set for the average image generated in the average image generation process and each luminance image generated in the luminance value setting process to obtain the low resolution image A plurality of low resolution images having different resolutions are generated by repeatedly executing phase image synthesis processing to be created while changing the size of the divided area when setting the divided area in the captured image. ,
The defect inspection method is configured to create the combined image by combining the plurality of low resolution images in the step of extracting the defect candidate. In the luminance value setting process,
Extracting a pixel of interest and a pair of adjacent pixels adjacent to both sides of the pixel of interest with respect to the average image;
The luminance value is set based on the luminance difference only when both of the pair of adjacent pixels are both bright or dark with respect to the pixel value of the pixel of interest, otherwise the luminance value is set to 0. The defect inspection method according to claim 4 configured. The phase image synthesis process synthesizes and normalizes the luminance image.
The defect inspection method according to claim 4 or 5, wherein, in the step of extracting the defect candidate, the plurality of low resolution images normalized by the phase image synthesis process are synthesized to create the synthesized image.

Images