JPH08101916A - Defect inspection method and device therefor - Google Patents

Abstract

PURPOSE: To correctly discriminate whether or not a defect is present in an object and to stably inspect the defect of the object regardless of the kind of the object and observation conditions, etc. CONSTITUTION: A rectangular inspection area 4 is set in the image part 1A of the object of a gradation image 1. First, a rectangular small area 20 is set at an upper end position inside the inspection area 4 and the feature amount of an average density or the like is measured for the image inside the small area 20 at the position. Then, the small area 20 is moved for prescribed pitches and the feature amount is measured for the image inside the small area 20 at the position. Thereafter, after repeatedly executing the movement of the small area 20 and the measurement of the feature amount, a prescribed arithmetic operation is executed by respective measured values and the defect of the object is discriminated from the arithmetic result.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a defect inspection method and apparatus for inspecting a defect of an object by using an image obtained by picking up an image of the object. The present invention relates to a defect inspection method and apparatus for inspecting a defect of an object by determining whether the object has a defect such as a scratch, a protrusion or a chip, regardless of the observation conditions.

[0002]

2. Description of the Related Art Generally, for inspecting a defect of an object, an image of the object is picked up by an image pickup device under appropriate illumination, an inspection area is set in the obtained grayscale image or its binary image, and the inspection is performed. A predetermined characteristic amount is measured for the image in the region, and whether the object is good or bad, that is, whether or not the object has a defect is determined from the measured value.

FIG. 12 shows an example of a conventional defect inspection method. FIG. 12 (1) shows a grayscale image 1 obtained by imaging the object. In the figure, 1A is the image portion of the object, and 1B is the background image portion. A defect 2 such as a scratch appears in the image portion 1A of the object.

FIG. 12B shows a binary image 3 obtained by binarizing the grayscale image 1 with a predetermined binarization threshold value. In FIG. 12, the shaded area is a black pixel area. The white background areas are white pixels. The background image portion 1B and the defect 2 are binarized into black pixels, and the image portion 1A of the object is binarized into white pixels. A rectangular inspection area 4 is set in the image portion 1A of the object, and the area of the black pixel area, that is, the number of black pixels is measured as a feature amount in the image in the inspection area 4, and the measured value is a predetermined judgment value. If it is above, it is determined that the object has a defect, and if it is smaller than the determination value, it is determined that the object has no defect.

FIG. 13 shows another example of the conventional defect inspection method. FIG. 13 (1) shows a grayscale image 1 obtained by imaging the object. In the figure, 1A is an image portion of the object, 1B is a background image portion, and 2 is a defect 2. FIG.
In (4), the brightness distribution M1 along the line AA of the grayscale image 1 is shown. In this brightness distribution M1, the background image portion 1B is dark and the object image portion 1A is bright, but the brightness is lowered at the position of the defect 2.

FIG. 13B shows an edge image 5 obtained by subjecting the grayscale image 1 to contour extraction processing. In the figure,
1a is an edge that is the contour of the image portion 1A of the object,
a represents an edge that is the contour of the defect 2. FIG.
3 (5) shows the brightness distribution M2 along the line A'-A 'of the edge image 5. In the brightness distribution M2, the brightness at the positions of the edges 1a and 2a is rapidly increased.

FIG. 13C shows a binary image 6 obtained by binarizing the edge image 5 with a predetermined binarizing threshold value TH. In the figure, 1a 'is a black pixel area obtained by binarizing the edge 1a of the image portion 1A of the object in the edge image 5, and 2a' is binarizing the edge 2a of the defect 2. The obtained black pixel areas are shown respectively. FIG.
3 (6) shows a data distribution D along the line A'-A 'of the binary image 6. The data is "1" at the positions of the edges 1a and 2a, and "0" at the other positions.

A rectangular inspection area 4 is set in the image portion 1A of the object of the binary image 6, and the area of the black pixel area of the image in the inspection area 4, that is, the number of black pixels is measured as a feature amount. If the measured value is greater than or equal to a predetermined determination value, it is determined that the target object has a defect, and if it is smaller than the determination value, it is determined that the target object has no defect.

In each of the above conventional examples, the inspection area 4 is set in the image portion 1A of the object of the binary images 3 and 6 obtained by binarizing the grayscale image 1 and the edge image 5. The feature amount is measured and the presence / absence of a defect is determined. In addition to these methods, the inspection area 4 is set in the image portion 1A of the object of the grayscale image 1, and the average of the images in the inspection area 4 is set. There is also a method in which a density or a variance value is measured as a feature amount and the measured value is compared with a determination value to inspect a target object for defects.

[0010]

In the above defect inspection method, depending on the type of the object and the observation conditions, it is possible to correctly determine whether or not there is a defect in the object, as will be described in detail in the following specific examples. However, there is a problem that stable inspection cannot be performed.

FIG. 14 shows a problem when a cylindrical body is an object. 14 (1) is a grayscale image 1 obtained by imaging the object, and FIG. 14 (2) is a grayscale image 1 obtained by binarizing the grayscale image 1 with a predetermined binarization threshold TH. Binary image 3 is shown. In FIG. 14 (1), 1A is the image portion of the target object, 1B is the background image portion, and 2 is the defect 2 appearing in the image portion 1A of the target object. Shadows 7 called “shading” appear on both sides of the shadow 2. FIG. 14C shows an image brightness distribution M1 along the line AA of the grayscale image 1. In this brightness distribution M1, the background image portion 1B is dark and the object image portion 1A is bright, but the brightness is lowered at the position of the defect 2 and the position of the shadow 7. In the binary image 3 shown in FIG. 14 (2), the shaded area is a black pixel area and the white background area is a white pixel area, but not only the defect 2 but also the shadow 7 area is a black pixel. Has been converted. In FIG. 14 (4), the grayscale image 1 is represented by the binarization threshold TH shown in FIG. 14 (3).
The data distribution D when binarized, that is, the data distribution D along the A′-A ′ line of the binary image 3 is shown.

When the inspection area 4 is set in the image portion 1A of the object of the binary image 3 and the area of the black pixel area of the image in the inspection area 4 is measured, the measured value is that of the defect 2. It is the sum of the area and the area of the shadow 7, and only the area of the defect 2 cannot be detected. Therefore, even if the defect 2 does not exist, the object may be determined to be defective due to the presence of the shadow 7.

FIG. 15 shows a defect when an object having a satin-finished surface is an object. FIG. 15 (1) is a grayscale image 1, FIG. 15 (2) is a binary image 3, 1A is an image portion of the object, 1B is a background image portion, and 2 is a defect. In the binary image 3, the image portion 1A of the target object is binarized into white pixels, but the image portion 1A of the target object has a defect 2 and a satin pattern binarized into black pixels. Countless scratches 8 have appeared.

When the inspection area 4 is set in such a binary image 3 and the area of the black pixel area is measured, the measured value is the sum of the area of the defect 2 and the area of all the scratches 8, Only the area of the defect 2 portion cannot be detected.

FIG. 16 shows a defect example in the case of inspecting a defect such as a protrusion or a chip appearing on the edge of an object. FIG. 16 (1) shows the grayscale image 1, FIG. 16 (2) shows the binary image 3, 1A shows the image portion of the object, and 1B shows the background image portion. A projection 9 and a chip 10 appear on one edge of the image portion 1A of the object. FIG. 16 (2)
In the binary image 3 shown in FIG. 3, the shaded area is a black pixel area and the white background area is a white pixel area, but the protrusion 9 is binarized and the chipped portion 10 is black pixel. .

In the image portion 1A of the object of the binary image 3 as described above, a rectangular inspection area 11 is set on the inside and a belt-shaped inspection area 12 is set on the outside. The area of the black pixel area is measured for the image, and the area of the white pixel area is measured for the image in the strip-shaped inspection area 12. If the area of the black pixel area in the inspection area 11 is equal to or larger than the determination value, it is determined that the chip 10 exists. If the area of the white pixel area in the inspection area 12 is equal to or larger than the determination value, it is determined that the protrusion 9 is present.

In such a method, in order to detect the small projection 9 and the chip 10, it is necessary to set the inspection areas 11 and 12 close to the edge of the image portion 1A of the object. Such setting is not easy, and if the positional deviation of the image is not corrected with high accuracy, it is difficult to inspect the protrusion 9 and the chip 10.

FIG. 17 and FIG. 18 show problems when the illumination conditions for picking up an object change. FIGS. 17 (1) and 17 (2) are images before the illumination intensity is changed. FIG. 17 (1) is the grayscale image 1 and FIG. 17 (2) is the binary image 3. 1A is the image part of the object, 1B
In FIG. 17 (3), a brightness distribution M1 along the line AA of the grayscale image 1 is shown. When the grayscale image 1 is binarized with a predetermined binarization threshold value TH, a binary image 3 shown in FIG. 17B is obtained. In this binary image 3, the shaded area is a black pixel area and the white background area is a white pixel area.

FIGS. 18 (1) and 18 (2) are images when the illumination intensity is lowered. FIG. 18 (1) shows the grayscale image 1 and FIG.
8 (2) is the binary image 3. In FIG. 18 (3),
The brightness distribution M2 along the line A'-A 'of the grayscale image 1 is shown, and the brightness level is lower than the brightness distribution M1 before the illumination change. When this grayscale image 1 is binarized by the binarization threshold TH, as shown in FIG. 18B, the image portion of the object disappears, and the whole image becomes a black pixel of one color.
Value image 3 is obtained, and therefore a defect inspection is impossible.

FIG. 19 shows a defect in the case where the defect is dirt having a small density difference from the background. FIG.
(1) shows a grayscale image 1 obtained by imaging such a dirty object. In the figure, 1A is the image portion of the object, 1B is the background image portion, and the image portion 1 of the object is
Defect 2 due to thin dirt appears in A. FIG.
In (2), the brightness distribution M1 along the line AA of the grayscale image 1 is shown. In this brightness distribution M1, the background image portion 1B is dark and the target image portion 1A is bright, but the brightness is slightly reduced at the position of the defect 2. FIG. 19C shows the brightness distribution M2 of the edge image (not shown) generated from the grayscale image 1, but the edge of the defect 2 is not extracted in this edge image, and therefore the defect 2 is not extracted. The brightness does not increase at the position. For such a grayscale image 1, a binary image including a defect cannot be stably obtained, and a defect inspection is difficult.

FIG. 20 shows an example of a problem when a net-like body is the object. The figure shows a binary image 3 obtained by binarizing the grayscale image of the object. 13A is a mesh image portion and 13B is a background image portion, and defect 2 due to breakage appears in the mesh image portion 13A. In the inspection of this kind of defect 2, the entire binary image 3 is used as the inspection area, and the area of the black pixel area, that is, the number of black pixels is measured for this binary image 3, and the measured value is determined in a predetermined manner. If the value is greater than or equal to the value, it is determined that the target object has no defect, and if it is smaller than the determination value, it is determined that the target object has a defect. However, in the case of the mesh body, even if a part of the image portion 13A of the object is missing due to the presence of the defect 2, it is absorbed by the variation of the image portion 13A of the object. is there.

FIG. 21 shows a problem when a ring-shaped body such as a rubber band is used as an object. FIG. 21 (1)
(2) is a grayscale image of an object having a defect such as a cutout.
The binary image 3 obtained by the binarization process is shown. 1A in the figure
Is an image part of the object, 1B is an image part of the background, and a defect 2 due to a cut appears in the image part 1A of the object. In this binary image 3, the shaded area is a black pixel area and the white background area is a white pixel area, but the defect 2 area is binarized into a black pixel. A circular linear inspection area 14 is set in the image portion 1A of the object of the binary image 3 as described above, and the number of black pixels on the inspection area 14 is measured. If so, it is determined that the defect 2 exists in the object, and if the number of black pixels is smaller than the determination value, it is determined that the defect 2 does not exist in the object. In this method, it is necessary to correctly set the inspection region 14 on the image portion 1A of the narrow object, but such a setting is not easy, and as shown in FIG. Is not performed properly or the object is deformed,
The set position of the inspection area 14 with respect to the image portion 1A of the object is displaced, and it becomes difficult to detect the defect 2.

The present invention was made by paying attention to the above-mentioned various problems, regardless of the type of object or the observation conditions.
It is an object of the present invention to provide a defect inspection method and an apparatus therefor capable of accurately determining whether or not a defect exists in an object and stably inspecting the object for defects.

[0024]

According to a first aspect of the present invention, there is provided a method for inspecting a defect of an object by an image obtained by imaging the object, wherein an inspection area of the image is divided into a plurality of small areas. Then, the feature amount is measured for each image in each small area, and then the measured value is processed to perform the defect determination of the object.

According to the present invention, the image is a grayscale image, and the average density, the variance value, the position of the center of gravity, and the grayscale correlation value with the model image of the image in each small area are measured as the feature amount. I am trying.

According to the sixth and seventh aspects of the invention, the image is a binary image, and the area of the black pixel region or the white pixel region and the position of the center of gravity of the image in each small region are measured as the characteristic amount. There is.

According to an eighth aspect of the present invention, in a method of inspecting a defect of the object by an image obtained by picking up an image of the object, the inspection area of the image is divided into a plurality of small areas, and each small area is divided. After measuring multiple types of feature values for each of the images in the image, a defect determination of the object was performed by processing each measurement value for each type of feature amount, and defect determination was performed for all types of feature amounts. At this time, it is characterized in that the object is judged to be defective.

According to a ninth aspect of the present invention, in a method of inspecting a defect of the object by an image obtained by picking up an image of the object, the inspection area of the image is divided into a plurality of small areas, and each small area is divided. After measuring a plurality of types of feature values for each image in the image, the measured value is processed for each type of feature amount to determine the defect of the target object, and the defect determination is performed for any type of feature amount. It is characterized in that the object is judged to be defective when it is turned on.

According to a tenth aspect of the present invention, in an apparatus for inspecting a defect of the object by an image obtained by picking up an image of the object, an area dividing means for dividing an inspection area of the image into a plurality of small areas, A feature amount measuring unit that measures a feature amount for each image in each small region divided by the region dividing unit, and a defect determination of an object by processing each feature amount measured by the feature amount measuring unit. And a discriminating means for performing.

[0030]

In the case of an object having a defect, since the inspection area of the image is divided into a plurality of small areas, the defective portion appearing in the image is included in any one of the small areas. A feature amount measured from the image in the small area including the defective part,
There is a difference between the feature amount measured from the image in the small area that does not include the defective portion, and the presence or absence of the defect can be determined from this difference.

According to the second to fifth aspects of the present invention, the average density, the variance value, the position of the center of gravity, and the density correlation value with the model image are measured as the feature quantity, and are measured from the image in the small area including the defective portion. The average density, variance, centroid position, and grayscale correlation value with the model image, and the average density, variance, centroid position, and grayscale correlation value with the model image measured from the image in the small area that does not include the defect part. There is a difference between them, and the presence or absence of a defect can be determined from this difference.

According to the sixth and seventh aspects of the present invention, the area and barycentric position of the black pixel area or the white pixel area are measured as the characteristic amount, and the area and barycenter measured from the image in the small area including the defective portion. There is a difference between the position and the area and the center of gravity measured from the image in the small area that does not include the defective portion,
The presence or absence of defects can be determined from this difference.

According to the invention of claim 8, when the measured values of the characteristic amount are processed for each type of the characteristic amount and the defect determination of the object is performed, the defect determination is performed for all types of the characteristic amounts. , The object is judged to be defective.

According to the ninth aspect of the present invention, as a result of processing the measured values of the feature amount for each type of feature amount and performing defect determination of the object, defect determination is performed for any type of feature amount. At this time, it is determined that the object is defective.

In the defect inspection apparatus according to the tenth aspect of the present invention, the inspection area of the image is divided into a plurality of small areas by the area dividing means, and the characteristic amount is measured for each image in each small area. To measure. The discriminating means discriminates the defect of the object by processing each measured characteristic amount.

[0036]

1 to 7 specifically show a defect inspection method according to the present invention. FIG. 1 shows a specific example in which a cylindrical body is used as an object, and FIG. 14 (1) shows a grayscale image 1 obtained by imaging the object. In the figure, 1A is an image part of the object, 1B is an image part of the background, 2 is a defect 2 appearing in the image part 1A of the object, and the image part 1A of the object is located on both sides of the defect 2. To "shading"
There is a shadow 7 called.

A rectangular inspection area 4 is set in the image portion 1A of the object of the grayscale image 1. The inspection area 4 is divided into a plurality of small areas 20, and the images in the respective small areas 20 are respectively divided. The average density is measured as a feature amount. In this embodiment, first, a rectangular small area 20 having a horizontal width H and a vertical width W is set at the upper end position in the inspection area 4, and the average density of the image in the small area 20 is measured at this position. Then, the small area 20 is moved downward by a predetermined pitch d (where d ≦ W), and the average density of the image in the small area 20 is measured at that position in the same manner. Hereinafter, the movement of the small area 20 and the measurement of the average density are repeatedly executed until the small area 20 reaches the lower end position of the inspection area 4.

FIG. 1B is a plot of average densities measured at each moving position of the small area 20.
The average density is low at the position of the presence of and in the vicinity. Now, the average density at the i-th moving position is R (i), the average density at each moving position immediately before and immediately after that is R (i-1),
If R (i + 1), the average density R (i) and the average density R
The absolute value of the difference between the average value of (i-1) and R (i + 1) is calculated by the following equation.

[0039]

[Equation 1]

FIG. 1C is a plot of the calculated value a at each moving position of the small area 20, and when the defect 2 exists, a point P exceeding the predetermined judgment value S appears. Therefore, it is determined whether or not the object is defective depending on whether or not there is a point that exceeds the determination value S.

In the above embodiment, the average density of the grayscale image in the small area 20 is measured as the feature amount.
Not limited to this, the variance value may be measured as a feature amount,
Further, a grayscale correlation value with the position of the center of gravity or the model may be calculated as the feature amount. Further, in the above embodiment, the grayscale image is targeted, but the area or the center of gravity of the region of the black pixel or the white pixel may be measured or calculated for the binary image. Further, in the above-described embodiment, one type of feature amount (average density) is measured and the calculated values are sequentially calculated, and when any of the calculated values exceeds the predetermined judgment value S, a defect exists in the object. However, it is determined that a plurality of types of feature amounts (such as average density, variance value, barycentric position for grayscale image, and grayscale correlation value with model, area, barycentric position for binary image, etc.) are measured. The calculated values may be sequentially calculated, and for any type of feature amount, it may be determined that the object has a defect when any of the calculated values exceeds a predetermined determination value. It may be determined that a defect exists in the object when any of the calculated values exceeds a predetermined determination value.

FIG. 2 shows a specific example in which an object having a satin-finished surface is used as an object, and FIG. 2 (1) shows a grayscale image 1 obtained by imaging the object. In the figure, 1A is an image portion of the object, 1B is an image portion of the background, 2 is a defect 2 appearing in the image portion 1A of the object, and the image portion 1A of the object is a myriad of satin-finished patterns. The scratches 8 have appeared.

A rectangular inspection area 4 is set in the image portion 1A of the object of the grayscale image 1. First, the inspection area 4 is set.
A rectangular small area 2 similar to that of the embodiment of FIG.
0 is set, and the variance value of the image in the small region 20 is measured at this position as a feature amount. Then, the small area 20 is moved downward by a predetermined pitch, and the variance value of the image in the small area 20 is similarly measured at that position. Less than,
The movement of the small area 20 and the measurement of the variance value are repeatedly executed until the small area 20 reaches the lower end position of the inspection area 4.

FIG. 2 (2) is a plot of the dispersion value measured at each movement position of the small area 20, and the dispersion value is lowered at the position where the defect 2 exists and its vicinity. Now i
The variance value at the th moving position is σ (i), and the variance value at each moving position immediately before and immediately after is σ (i−1), σ (i +
1), the slope b of the variance value is σ (i), σ (i-
It is calculated by the following equation using the difference between 1) and the difference between σ (i) and σ (i + 1).

[0045]

[Equation 2]

FIG. 2 (3) is a plot of the calculated value b at each moving position of the small area 20, and when the defect 2 exists, points P1 and P2 that exceed the predetermined judgment value S appear. . Therefore, it is determined whether or not the object is defective depending on whether or not there is a point that exceeds the determination value S.

FIG. 3 shows a specific example of the case of inspecting for defects such as protrusions and chips appearing on the edge of the object.
(1) shows a grayscale image 1 obtained by imaging the target object. In the figure, 1A is an image portion of the object, and 1B is an image portion of the background, and a projection 9 and a chip 10 appear on one edge of the image portion 1A of the object.

A rectangular inspection area 4 is set along one edge of the image portion 1A of the object. First, a rectangular small area 20 is set at the upper end position in the inspection area 4,
At this position, the average density of the image in the small area 20 is measured as a feature amount. Then, the small area 20 is moved downward by a predetermined pitch, and the average density of the image in the small area 20 is measured at that position in the same manner. Below, small area 20
The movement of the small area 20 and the measurement of the average density are repeatedly executed until reaches the lower end position of the inspection area 4.

FIG. 3B is a plot of the average densities measured at the respective moving positions of the small area 20. The average densities decrease at the positions where the projections 9 are present and in the vicinity thereof, and the chipping 1
The average density increases at the position where 0 exists and its vicinity.
The average density at the i-th moving position is R (i), and the average density at each moving position immediately before and immediately after that is R (i-).
1) and R (i + 1), the absolute value a of the difference between the average concentration R (i) and the average value of the average concentrations R (i-1) and R (i + 1) is calculated by the above-mentioned formula. FIG. 3 (3) is a plot of the calculated value a at each movement position of the small area 20, and when the projection 9 or the chip 10 is present, points P1 and P2 that exceed the predetermined determination value S are appear. Therefore, it is determined whether or not the object is defective depending on whether or not there is a point that exceeds the determination value S.

FIG. 4 shows a specific example in the case where the illumination condition (illumination intensity) for imaging the object changes.
(1) shows a grayscale image 1 obtained by imaging the target object. In the figure, 1A is an image portion of the object, 1B is a background image portion, 2 is a defect 2 appearing in the image portion 1A of the object, and FIG. The brightness distribution M1 along the line AA of the image 1 and the brightness distribution M2 along the line AA of the grayscale image 1 when the illumination intensity is reduced.
And are shown.

A rectangular inspection area 4 is set in the image portion 1A of the object of the grayscale image 1. First, the inspection area 4 is set.
A rectangular small area 2 similar to that of the embodiment of FIG.
0 is set, and the variance value of the image in the small region 20 is measured at this position as a feature amount. Then, the small area 20 is moved downward by a predetermined pitch, and the variance value of the image in the small area 20 is similarly measured at that position. Less than,
The movement of the small area 20 and the measurement of the variance value are repeatedly executed until the small area 20 reaches the lower end position of the inspection area 4.

FIG. 4 (3) is a plot of the variance value measured at each movement position of the small area 20, and the variance value increases at the position where the defect 2 exists and its vicinity, and the predetermined judgment value is obtained. A point P that exceeds S appears. Therefore, it can be determined whether or not the object is defective depending on whether or not there is a point that exceeds the determination value S.

FIG. 5 shows a specific example of the case where the defect is dirt having a small density difference from the background, and FIG. 5 (1) shows a grayscale image 1 obtained by imaging the object. 1A in the figure
Is an image portion of the object, 1B is an image portion of the background, and a defect 2 due to a light stain appears in the image portion 1A of the object. FIG. 5B shows the brightness distribution M1 along the line AA of the grayscale image 1. In this brightness distribution M1, the background image portion 1B is dark and the image portion 1A of the object is
Is bright, but the brightness is slightly reduced at the position of defect 2.

A rectangular inspection area 4 is set in the image portion 1A of the object of the grayscale image 1. First, the inspection area 4 is set.
A rectangular small area 20 is set at the upper end position in the inside, and the variance value of the image in the small area 20 is measured at this position as a feature amount. Then, the small area 20 is moved downward by a predetermined pitch, and the variance value of the image in the small area 20 is similarly measured at that position. Hereinafter, the movement of the small area 20 and the measurement of the variance value are repeatedly executed until the small area 20 reaches the lower end position of the inspection area 4.

FIG. 5 (3) is a plot of the dispersion value measured at each moving position of the small area 20, and the dispersion value increases at the existence position of the defect 2 and its vicinity. The variance value at the i-th moving position is σ (i), and the variance value at each moving position immediately before and after that is σ (i−1), σ (i + 1).
Then, the slope b of the variance value is calculated by the above formula using the difference between σ (i) and σ (i−1) and the difference between σ (i) and σ (i + 1). FIG. 5 (4) is a plot of the calculated value b at each moving position of the small area 20, and the defect 2
Exists, points P1 to P4 that exceed a predetermined judgment value S
Appears. Therefore, it is determined whether or not the object is defective depending on whether or not there is a point that exceeds the determination value S.

FIG. 6 shows a concrete example of the case where a net-like body is an object, and FIG. 6 (1) shows a grayscale image 1 obtained by imaging the object. In the figure, 1A is an image portion of a mesh body, 1
B is an image portion of the background, and a defect 2 due to breakage appears in the image portion 1A of the mesh body.

The gray image 1 has a rectangular inspection area 4a.
4d are set in four columns, and the inspection area 4a in the first column is first set.
A rectangular small area 20 is set at the upper end position in the inside, and the variance value of the image in the small area 20 is measured at this position as a feature amount. Then, the small area 20 is moved downward by a predetermined pitch, and the variance value of the image in the small area 20 is similarly measured at that position. Hereinafter, the movement of the small area 20 and the measurement of the variance value are repeatedly executed until the small area 20 reaches the lower end position of the inspection area 4a. Next, the rectangular small area 20 is set at the upper end position in the inspection area 4b in the second column, and the variance value of the image in the small area 20 is measured as the feature amount at this position. Then, the small area 20 is moved downward by a predetermined pitch, and the small area 20 is similarly moved at that position.
Measure the variance of the images inside. Below, small area 20
The movement of the small area 20 and the measurement of the variance value are repeatedly executed until the point reaches the lower end position of the inspection area 4b. The same applies to the inspection areas 4c and 4d in the third and fourth columns.

FIG. 6B is a plot of the dispersion value measured at each moving position of the small area 20, and the dispersion value is lowered at the existence position of the defect 2 and its vicinity. Now i
The variance value at the th movement position is σ (i), i-
The first and the i−2nd variance values are σ (i−1), σ
(I-2), assuming that the i + 1-th and i + 2-th variance values immediately after are (σ (i + 1), σ (i + 2)), the variance value σ (i)
And variance values σ (i-1), σ (i-2), σ (i + 1),
The absolute value of the difference between σ (i + 2) and the average value is calculated by the following formula.

[0059]

(Equation 3)

FIG. 6C is a plot of the calculated value c at each moving position of the small area 20, and when the defect 2 exists, a point P exceeding the predetermined judgment value S appears. Therefore, it is determined whether or not the object is defective depending on whether or not there is a point that exceeds the determination value S.

FIG. 7 and FIG. 8 are concrete examples in which a ring-shaped body such as a rubber band is used as an object, and FIG. 7 (1) is a grayscale image obtained by imaging the undeformed object. 1
FIG. 8A shows a grayscale image 1 obtained by imaging the deformed object. In the figure, 1A is an image portion of the object, 1B is an image portion of the background, and a defect 2 due to a break appears in the image portion 1A of the object.

A circular linear inspection area 14 is set in the grayscale image 1, a small area 20 is set at an appropriate angular position on the inspection area 14, and the small area 20 is located at this angular position. The variance value of the image is measured as a feature amount.
Then, the small area 20 is moved by a predetermined angle pitch θ,
Similarly, at the angular position, the variance value of the image in the small area 20 is measured. Hereinafter, the movement of the small area 20 and the measurement of the variance value are repeatedly executed until the small area 20 reaches the first angular position.

FIG. 7B and FIG. 8B show the small area 2
It is a plot of the dispersion value measured at each movement position of 0, and the dispersion value is lowered at the position where the defect 2 exists and its vicinity. The variance value at the i-th movement position is σ
(I), i−1th immediately before, i−2nd, i
-3rd variance value is σ (i-1), σ (i-2), σ
(I-3), the i + 1th, i + 2th, and i + 3th variance values immediately after that are σ (i + 1), σ (i +
2) and σ (i + 3), the variance value σ (i) and the variance values σ (i-1), σ (i-2), σ (i-3), σ (i
The absolute value of the difference between the average value of +1), σ (i + 2), and σ (i + 3) is calculated by the following formula.

[0064]

[Equation 4]

FIG. 7C and FIG. 8C show the small area 2
It is a plot of the calculated value d at each moving position of 0. When the defect 2 exists, points P1 to P3 that exceed a predetermined judgment value S appear. Therefore, it can be determined whether or not the object is defective depending on whether or not there is a point exceeding the determination value S.

FIG. 9 shows an example of the configuration of a defect inspection apparatus for carrying out the above defect inspection method. The defect inspection apparatus of the illustrated example includes a television camera 31, an image processing unit 32, and a monitor 3.
3, the image processing unit 32 includes an A / D conversion unit 34,
Image memory 35, affine transformation unit 36, feature amount measurement unit 3
7, display memory 38, CPU 39, ROM 40, RAM
41, a D / A converter 42, an external interface 43, and the like.

The television camera 31 picks up an object and generates a grayscale image, and an analog image signal from the television camera 31 is converted into digital image data by the A / D converter 34. . Image data of a digital amount for one frame is stored in the image memory 35, and an image representing the small area 20 is written in the display memory 35 as display data to be displayed on the monitor 3. The D / A converter 42 inputs the image data from the A / D converter 34 and the image memory 35 and the display data from the display memory 35, converts the image data into an analog image signal, and outputs it to the monitor 33. The external interface unit 43 is connected to a strobe light source, peripheral devices, and the like, which are not shown.

The affine transformation unit 36 uses the image memory 3
The grayscale image stored in 5 is rotated by affine transformation to generate a proper grayscale image with no rotation deviation. The feature amount measuring unit 37 sets an inspection region in the gray image after affine transformation, divides the inspection region into a plurality of small regions, and measures the feature amount of the gray image in each small region. Specifically, as in the specific examples shown in FIGS. 1 to 8, the small area 20 is moved along the inspection area by a predetermined pitch or an angular pitch, and the grayscale image in the small area 20 is moved at each moving position. Measures characteristic quantities such as average density and variance. CPU
Reference numeral 39 captures each measured value by the characteristic amount measuring unit 37, performs a process such as a predetermined calculation, and determines a defect of the target object from the process result. The ROM 40 stores programs and fixed data for controlling the operation of the defect inspection apparatus, and the RAM 41 is used as a work area during operation of the apparatus.

FIG. 10 shows an example of the configuration of a defect inspection apparatus that measures the density correlation value with a model image as a feature amount. In this embodiment, a model memory 44 is provided for storing a model image for obtaining a grayscale correlation with the grayscale image. Note that the other configurations are similar to those shown in FIG. 9, and therefore description thereof will not be repeated here.

FIG. 11 shows a control procedure of the above-described defect inspection apparatus from step 1 (hereinafter referred to as "ST1") to ST12.
It shows with.

First, when a grayscale image of the object is obtained at the start point, in ST1, the first set position of the small area 20 with respect to the grayscale image 1, that is, the starting point coordinates (X1, X2) and the final set position, that is, the end point coordinate. (X2, Y2), the horizontal width H and vertical width W of the small area 20, and the pitch d or the angular pitch θ for moving the small area 20 are fetched from the RAM 41 into the CPU 39. In the next step ST2, the feature amount measuring unit 37 sets the small area 20 at the initial position of the grayscale image 1 based on the data.

In the next step ST3, the feature amount measuring unit 37 measures the average density, the variance value, etc. of the image in the small area 20 as the feature amount. This measured value is fetched by the CPU 39 and stored in the RAM 41. In the next ST4,
It is determined whether or not the small area 20 has moved to the final position. If the determination is “NO”, the process proceeds to ST5, where the small area 20 is moved by a predetermined pitch or angular pitch, and then at that position. Similarly, the feature amount is measured for the image in the small area 20 (ST3). Thereafter, the movement of the small area 20 in ST5 and ST until the small area 20 reaches the final position.
The measurement of the feature amount of 3 and the measurement are repeatedly performed.

Thus, when the measurement of the characteristic amount at the final position is completed, the determination in ST4 becomes "YES", and the CPU 39
Reads out the measurement value obtained at the first set position of the small area 20 and the measurement values obtained at the front and rear positions thereof from the RAM 40 and executes a predetermined calculation (ST6, 7), and the calculated value is the judgment value S. It is determined whether or not (ST8). If S
If the determination in T8 is “YES”, the process proceeds to ST12 and C
The PU 39 outputs an abnormal signal indicating that the object is defective.

If the determination in ST8 is "NO", S
After T9, the process proceeds to ST10, the measured value obtained at the moving position of the next small area 20 and the measured values obtained at the front and rear positions thereof are read out from the RAM 40, and a predetermined calculation is executed. Determine if S is exceeded (ST
8). Thus, a predetermined calculation is executed from the measurement value obtained at the final movement position and the measurement values obtained at the front and rear positions thereof,
When the calculated value does not exceed the determination value S, the determination in ST9 becomes "YES", the process proceeds to ST11, and the CPU 39 outputs a normal signal indicating that the object is not defective.

[0075]

As described above, the present invention divides the inspection area of the image obtained by imaging the object into a plurality of small areas,
After measuring the feature amount for each image in each small area, the measured value is processed to determine the defect of the target object. It is possible to accurately determine whether or not there is a defect and stably inspect for defects in the object.

According to the invention of claim 8, as a result of measuring a plurality of types of characteristic amounts, processing each measured value of the characteristic amount for each type of characteristic amount, and performing defect determination of the object, all types of characteristic Since the object is determined to be defective when the quantity is determined to be defective, a highly accurate defect inspection can be stably performed.

According to the invention of claim 9, as a result of processing the measured values of the feature amount for each type of feature amount and performing defect determination of the object, defect determination is performed for any type of feature amount. At this time, since it is determined that the object is defective, there is an effect that many kinds of defects can be detected.

[Brief description of drawings]

FIG. 1 is an explanatory diagram showing an example in which a defect inspection method of the present invention is applied to a defect inspection of a cylindrical body.

FIG. 2 is an explanatory diagram showing an example in which the defect inspection method of the present invention is applied to a defect inspection of an object having a satin pattern.

FIG. 3 is an explanatory diagram showing an example in which the defect inspection method of the present invention is applied to inspection of protrusions and chips.

FIG. 4 is an explanatory diagram showing an example in which the defect inspection method of the present invention is applied to a defect inspection when illumination conditions change.

FIG. 5 is an explanatory diagram showing an example in which the defect inspection method of the present invention is applied to an inspection of dirt having a small density difference from the background.

FIG. 6 is an explanatory diagram showing an example in which the defect inspection method of the present invention is applied to a defect inspection of a mesh body.

FIG. 7 is an explanatory diagram showing an example in which the defect inspection method of the present invention is applied to a defect inspection of a ring-shaped body.

FIG. 8 is an explanatory diagram showing an example in which the defect inspection method of the present invention is applied to a defect inspection of a ring-shaped body.

FIG. 9 is a block diagram showing a configuration example of a defect inspection apparatus according to an embodiment of the present invention.

FIG. 10 is a block diagram showing a configuration example of a defect inspection device according to another embodiment of the present invention.

11 is a flowchart showing a control flow of the defect inspection apparatus of FIG.

FIG. 12 is an explanatory diagram showing a conventional defect inspection method.

FIG. 13 is an explanatory diagram showing another conventional defect inspection method.

FIG. 14 is an explanatory diagram showing a defect of a conventional example when a cylindrical body is an object.

FIG. 15 is an explanatory diagram showing a defect of a conventional example when an object having a satin pattern is used as an object.

FIG. 16 is an explanatory diagram showing a defect of a conventional example when inspecting a protrusion and a chip.

FIG. 17 is an explanatory diagram showing a defect of the conventional example when the illumination changes.

FIG. 18 is an explanatory diagram showing a defect of the conventional example when the illumination changes.

FIG. 19 is an explanatory diagram showing a defect of a conventional example when inspecting a stain having a small density difference from the background.

FIG. 20 is an explanatory diagram showing a defect of a conventional example when a mesh body is an object.

FIG. 21 is an explanatory diagram showing a defect of a conventional example when a ring-shaped body is an object.

[Explanation of symbols]

 1 Grayscale Image 2 Defects 4, 4a to 4d, 14 Inspection Area 20 Small Area 31 TV Camera 35 Image Memory 37 Characteristic Measurement Unit 39 CPU 40 ROM 41 RAM

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location 9061-5H G06F 15/70 320 9061-5H 355 9061-5H 360 9061-5H 460 C

Claims (10)

[Claims] 1. A method for inspecting a defect of an object using an image obtained by picking up an image of an object, wherein an inspection area of the image is divided into a plurality of small areas, and an image in each of the small areas is individually divided. A defect inspection method characterized by performing defect determination of an object by processing each measured value after measuring a characteristic amount. 2. The defect inspection method according to claim 1, wherein the image is a grayscale image, and the average density of each image in each small area is measured as a feature amount. 3. The defect inspection method according to claim 1, wherein the image is a grayscale image, and the variance value of each image in each small region is measured as a feature amount. 4. The defect inspection method according to claim 1, wherein the image is a grayscale image, and the barycentric position of each image in each small region is measured as a feature amount. 5. The defect inspection method according to claim 1, wherein the image is a grayscale image, and the grayscale correlation value of each of the images in each small region with the model image is measured as a feature amount. 6. The defect inspection method according to claim 1, wherein the image is a binary image, and an area of a black pixel region or a white pixel region is measured as a feature amount for each image in each small region. . 7. The defect inspection method according to claim 1, wherein the image is a binary image, and the barycentric position of each image in each small area is measured as a feature amount. 8. A method for inspecting a defect of an object using an image obtained by capturing an image of the object, wherein an inspection area of the image is divided into a plurality of small areas, and an image in each of the small areas is individually divided. After measuring multiple types of feature quantities, the measured values are processed for each type of feature quantity to determine whether the target is defective, and when the defect determination is performed for all types of feature quantities, the target is A defect inspection method characterized by determining that the defect is defective. 9. A method for inspecting a defect of an object using an image obtained by capturing an image of an object, wherein an inspection area of the image is divided into a plurality of small areas, and an image in each of the small areas is individually divided. After measuring a plurality of types of feature quantities, the measured values are processed for each type of feature quantity to determine whether the target object is defective. A method for inspecting defects, which comprises determining that the defect is defective. 10. An apparatus for inspecting a defect of the object by an image obtained by picking up an image of the object, comprising: an area dividing means for dividing an inspection area of the image into a plurality of small areas; and the area dividing means. A feature amount measuring unit that measures a feature amount of each image in each of the divided small areas, and a determination unit that performs a defect determination of an object by processing each feature amount measured by the feature amount measuring unit. A defect inspection device equipped.