JP2020118572A - Surface defect inspection device and surface defect inspection method - Google Patents

Abstract

To provide a surface defect inspection device and a surface defect inspection method that can accurately detect a defect even when a movement speed of a surface to be inspected is not constant and a movement locus is not straight.SOLUTION: A surface defect inspection device 1 includes defect detection means 40 for detecting a defect from a plurality of images imaged by imaging means 20 at given times during relative movement of the imaging means 20 and a surface to be inspected. The defect detection means 40 comprises defect candidate extraction means 42 for extracting a defect candidate from the plurality of images with different imaging times, and determination means 43 for determining that there is a defect when a movement distance and a movement angle between at least two sheets of images with different imaging times for the extracted defect candidate are within a range of a reference movement distance and a reference movement angle that are movement values estimated when there is a defect on the surface to be inspected.SELECTED DRAWING: Figure 2

Description

The present invention relates to a surface defect inspecting apparatus and a surface defect inspecting method, and more particularly to a surface defect inspecting apparatus and a surface defect inspecting method suitable for inspecting a condition such as coating on a body surface of an automobile.

2. Description of the Related Art Conventionally, in the process of painting an automobile body, an inspector visually inspects the painted surface. However, the visual inspection by the inspector is a labor-intensive task, and there is a possibility that an inspection error or an inspection omission may occur because it varies depending on individuals. Further, in the visual inspection by the inspector, the time required for the inspection increases, and the labor cost is one of the factors that increase the production cost of the product. Therefore, automation of appearance inspection is desired, and in recent years, development of a surface defect inspection apparatus capable of optically and automatically inspecting is underway.

For example, in Patent Document 1, a surface to be inspected is irradiated with a bright and dark pattern of stripes to move the surface to be inspected over time, and an image is taken by an imaging means at arbitrary time points. There is described a defect inspection method and a defect inspection apparatus for extracting a defect candidate from the image and determining that the defect candidate is a defect when the defect candidate moves in proportion to the movement amount of the surface to be inspected. According to this, it is possible to avoid erroneous detection due to noise and erroneous recognition of the boundary line, and it is possible to accurately detect the defect.

JP-A-8-145906

However, when inspecting the body of an automobile, if the vehicle body is transported by a conveyor, the traveling speed of the surface to be inspected is not constant due to uneven transportation speed of the conveyor, distortion of the conveyor, or shaking of the vehicle body on the conveyor, Alternatively, since the movement locus does not become a straight line, there is a problem in that it is not possible to make an accurate determination by determining whether or not there is a defect based on only the amount of movement of the surface to be inspected as in Patent Document 1. In addition, when irradiating a bright and dark pattern of stripes, there is also a problem that so-called flesh-like skin that appears at the boundary line of the bright and dark is judged to be a convex defect.

The present invention has been made based on such a problem, and even if the moving speed of the surface to be inspected is not constant or the movement locus is not a straight line, a surface defect inspection apparatus capable of accurately detecting a defect, and An object is to provide a surface defect inspection method.

The surface defect inspection apparatus of the present invention includes an illuminating unit that irradiates a surface to be inspected with light, an imaging unit that captures an image of the surface to be inspected illuminated by the illuminating unit, and an image of the surface to be inspected with respect to the imaging unit. Moving means for relatively moving the position, and defect detecting means for detecting a defect from a plurality of images taken by the imaging means at arbitrary times while relatively moving the imaging means and the surface to be inspected by the moving means. The defect detection means comprises: a defect candidate extraction means for extracting a defect candidate from each of a plurality of images at different imaging times; and at least two or more defect candidates extracted by the defect candidate extraction means at different imaging times. When the moving distance and the moving angle between the images are within the range of the reference moving distance and the reference moving angle that are assumed to move when a defect is present on the surface to be inspected, the determining means determines the defect. Is to have.

According to the surface defect inspection method of the present invention, the surface to be inspected irradiated with the light from the illumination means is imaged by the imaging means, and the position of the surface to be inspected with respect to the imaging means is moved relatively by the moving means, so that the imaging time is different. An imaging procedure for acquiring a plurality of images, a defect candidate extraction procedure for extracting defect candidates from a plurality of images obtained at the imaging procedure at different imaging times, and an imaging time for the defect candidate extracted in the defect candidate extraction procedure Is a defect when the movement distance and the movement angle between at least two or more different images are within the range of the reference movement distance and the reference movement angle which are assumed to move when a defect exists on the surface to be inspected. And a determination procedure for determining.

According to the present invention, the extracted defect candidate is within the range of the reference movement distance and the reference movement angle assumed to move when a defect exists on the surface to be inspected between at least two images at different imaging times. Since it is determined that the defect is a defect, it is possible to see the continuity of the defect over time, and even if the moving speed of the surface to be inspected is not constant or the moving trajectory is not a straight line, noise etc. The erroneous detection of can be eliminated with high accuracy, and the omission of defect detection can be reduced. Therefore, the defect can be detected with high accuracy.

In particular, if the reference movement angle is set based on the mark movement locus obtained by capturing a plurality of reference images at different image capturing times while moving the mark on the surface to be inspected in advance, it is high. The defect can be easily detected with high accuracy.

If an approximate straight line of the mark movement locus is drawn, and if the approximate straight lines are separated by a predetermined value or more in the width direction, a dividing line is drawn between them, and the approximate straight line and the angle of each reference line with the dividing line as the reference line. If the reference movement angle is set based on, the defect can be detected more easily and in a short time.

It is a figure showing the whole surface defect inspection device composition concerning one embodiment of the present invention. FIG. 3 is a block diagram showing the configuration of a defect detection unit shown in FIG. 1. It is an example of an image obtained by the preprocessing means. It is an example of an image obtained by the defect candidate extraction means in the reflector image. It is an example of an image obtained by the defect candidate extraction means outside the reflecting mirror. It is an example of a gradient image obtained by the gradient image generating means. It shows an example of a mark movement locus. 6 illustrates an example in which the mark moving locus is divided in the mark moving direction. It is an example of an approximate straight line of a mark movement locus. It is an example of an angle basic straight line. It is a figure showing the procedure of the surface defect inspection method which concerns on one embodiment of this invention. It is a figure for demonstrating the determination procedure shown in FIG.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 shows the overall configuration of a surface defect inspection apparatus 1 according to an embodiment of the present invention. The surface defect inspection apparatus 1 detects a defect existing on the surface of the surface M to be inspected, for example, by setting the painted surface of the body of the automobile as the surface M to be inspected.

The surface defect inspection apparatus 1 includes, for example, an illumination unit 10 that irradiates a surface M to be inspected with light, an image capturing unit 20 that captures an image of the surface M to be inspected illuminated by the illumination unit 10, and an image capturing unit. 20, a moving unit 30 that relatively moves the position of the surface M to be inspected with respect to 20, and the moving unit 30 relatively moves the imaging unit 20 and the surface M to be inspected, and the imaging unit 20 takes images at arbitrary time intervals. The defect detecting means 40 for detecting a defect from the plurality of images and the display means 50 for displaying the detection result by the defect detecting means 40 are provided.

The illumination unit 10 has an irradiation light source 11, and it is preferable to use a fluorescent lamp for the illumination light source 11. This is because a white light source is preferable because the color of the body of the car is various, and the price is low. Moreover, it is preferable to arrange a plurality of irradiation light sources 11 with respect to the surface M to be inspected so that the surface M to be inspected can be observed from multiple directions. The image pickup means 20 has a camera 21 such as a CCD camera, for example, and can obtain a digital image. The camera 21 is arranged, for example, so as to face the irradiation light source 11, and is configured to capture an image of the reflecting mirror of the fluorescent light, which is the illumination light source 11, and its peripheral region.

The moving means 30 moves at least one of the image pickup means 20 and the surface M to be inspected, thereby moving the position of the surface M to be inspected relative to the image pickup means 20. For example, it is preferable that the surface M to be inspected is conveyed in one direction at a constant speed by a conveyer such as a conveyor. The defect detecting means 40 is composed of, for example, a computer and is configured to detect a defect by image processing. The display means 50 is composed of, for example, a display or the like, and is arranged to display, for example, a circle mark or the like at the center of gravity of the defect.

(Defect detection means 40)
FIG. 2 shows the configuration of the defect detecting means 40 shown in FIG. The defect detection means 40 is, for example, an image storage means 41 such as a memory that stores a plurality of images captured by the imaging means 20 at different imaging times while relatively moving the imaging means 20 and the surface M to be inspected. Defect candidate extraction means 42 for extracting a defect candidate from each of a plurality of images stored in the image storage means 41 at different imaging times, and whether the defect candidate is extracted from the movement status of the defect candidate extracted by the defect candidate extraction means 42. It has a determination means 43 for determining whether or not.

(Defect candidate extraction means 42)
The defect candidate extraction means 42 is, for example, a preprocessing means 421 for preprocessing the image obtained by the image pickup means 20, and a reflection mirror image area detection means 422 for detecting the reflection mirror image area of the image obtained by the preprocessing means 421. A defect-in-reflective-mirror-defect candidate extracting unit 423 for extracting a defect candidate in the detected reflective-mirror-image region; Gradient image generation means 425 for generating a gradient image from the image obtained by the means 421, defect candidate determination means 426 for determining defect candidates, and defect candidate storage means for storing the defect candidates determined by the defect candidate determination means 426. And 427.

The preprocessing unit 421 converts the image obtained by the image pickup unit 20 into a grayscale image or other grayscale image, cuts out the region of the surface M to be inspected, reduces noise, and binarizes it. Is. Examples of noise reduction include a Gaussian filter and a median filter. An example of the image obtained by the preprocessing means 421 is shown in FIG. The reflection mirror image area detection unit 422 detects the reflection mirror image area of the fluorescent lamp, which is the illumination light source 11, in the image obtained by the preprocessing unit 421. Specifically, for example, the white portion in FIG. 3 is the reflecting mirror image area of the fluorescent lamp, and such a white portion is detected.

The defect candidate extracting unit 423 in the reflector image, for example, inverts the pixels in the reflector image region detected by the reflector image region detecting unit 422 (that is, black pixels are white pixels, white pixels are black pixels), and after inversion. Is recorded, and the white pixel portion in the reflection mirror image area is extracted as a defect candidate in the reflection mirror image. FIG. 4 shows an example of an image obtained by the defect candidate extracting means 423 in the reflecting mirror image. In FIG. 4, a portion surrounded by a white frame is in the reflection mirror image, and a white pixel portion in a circle is a defect candidate in the reflection mirror image.

The non-reflecting mirror image defect candidate extracting means 424 expands the region of interest from the reflecting mirror image region detected by the reflecting mirror image region detecting means 422 so as to include a region between the respective reflecting mirror image regions, and the differential filter is applied to this region of interest. A differential image is generated by multiplying the differential image by threshold processing on the area of the white pixel block for this differential image to remove the block with a large area, and the remaining white pixel part is extracted as a defect outside the reflection mirror image. is there. FIG. 5 shows an example of an image obtained by the defect candidate extraction means 424 outside the reflecting mirror image. In FIG. 5, the white pixel portion surrounded by ◯ is a defect candidate outside the reflection mirror image.

The gradient image generation means 425, for example, performs difference calculation on the image preprocessed by the preprocessing means 421 in two directions orthogonal to each other from both positive and negative directions, and generates a gradient image from the calculation result. Is. The difference calculation is executed as an approximate differentiation process in the digital image, and the gradient image generation means 425 performs difference calculation from both the positive direction and the negative direction to emphasize the defect in the positive difference direction. The defect can be treated two-dimensionally by representing it by the positive emphasis part and the negative emphasis part emphasized in the negative difference direction.

Specifically, for example, the gradient image generating means 425 uses Formula 1 and Formula 2 to perform difference calculation from both positive and negative directions in two orthogonal directions, and obtains using Formula 3 and Formula 4, respectively. The gradient image is preferably generated from absolute values of the positive and negative gradient vectors of the first direction difference and the second direction difference in the two orthogonal directions.

(In Expressions 1 and 2, (x, y) represents the coordinates of the pixel of interest, and I(x, y) is the luminance value of the pixel (x, y). G _xn (x, y) and G _yn. (X, y) are a first direction difference and a second direction difference in two orthogonal directions, n is a variable that controls the direction in which the difference is taken, and 0 indicates a positive direction and 1 indicates a positive direction. In the negative direction, k and s are variables that control the mask that takes the difference, ω is a variable that controls the orthogonal axis that takes the difference, and when ω=1, the difference is the vertical and horizontal axes, and ω=0. In the case of, the difference is on the diagonal orthogonal axis. γ and θ are not 0 at the same time, but are variables that control the positional relationship between two pixels that take the difference with (x, y) as the pixel of interest.)

For example, when ω=1 and γ=θ=1, Formula 1 and Formula 2 become Formula 5 and Formula 6 that take the difference in the horizontal direction and the vertical direction.

(In Expressions 5 and 6, the meanings of symbols are the same as those in Expressions 1 and 2.)

In the difference results in the two orthogonal directions calculated by Equations 1 and 2, a positive value indicates that the brightness changes from “bright” to “dark”, and a negative value conversely changes from “dark” to “bright”. Represents the change of. Therefore, by using only positive values from the calculation result, it is possible to represent the change from “bright” to “dark” with the difference in the positive direction and the change from “dark” to “bright” with the difference in the negative direction. .. Furthermore, when a gradient image is generated using Equations 3 and 4, for example, when a convex defect is present, half of the convex defect is emphasized in the positive difference direction and the opposite half is emphasized in the negative difference direction. Is represented by Therefore, for example, the defect can be easily recognized by displaying the positive emphasis portion emphasized in the positive difference direction and the negative emphasis portion emphasized in the negative difference direction in different colors.

FIG. 6 shows an example of the gradient image. In FIG. 6A, the positive emphasis portion emphasized in the positive difference direction is shown in green, and the negative emphasis portion emphasized in the negative difference direction is shown in red. In FIG. 6A, two lines extending in the left-right direction are boundary areas of the reflection mirror image of the fluorescent lamp, the sandwiched portion is the reflection mirror image of the fluorescent lamp, and the portion surrounded by the white broken line is a convex defect. .. 6B is an enlarged view of a portion surrounded by a white broken line in FIG. 6A, and a positive emphasis portion (green portion) emphasized in the positive difference direction is represented by hatching of a lower left diagonal line, and the negative difference direction is indicated. The negative emphasis part (red part) emphasized by is shown by hatching of the lower right diagonal line. Although it is similarly expressed in the area outside the reflecting mirror image, in the case of a defect, since the background is dark and the defect portion is bright, it is emphasized in the positive difference direction and in the negative difference direction. The position of the negative emphasis part is reversed.

On the other hand, when there is not a defect but so-called “Yuzu skin”, half as shown in FIG. 6 is emphasized in the positive difference direction, and the other half is emphasized in the negative difference direction. The part is hard to appear. Therefore, it is preferable to use the gradient image generated by the gradient image generation means 425, because it is possible to easily distinguish the defect and the orange peel skin.

Further, the gradient image generation means 425 does not perform a difference calculation in both the positive and negative directions in the two orthogonal directions and generates a gradient image from the calculation result, but performs a difference calculation in the two orthogonal directions. The gradient image is generated from the absolute value of the gradient vector calculated using only the positive value from the calculation result and the absolute value of the gradient vector calculated using only the negative value from the calculation result. You may

Specifically, for example, the gradient image generation means 425 uses Equation 7 and Equation 8 to perform difference calculation in two orthogonal directions, and Equations 9 and 10 are used to obtain only positive values from the difference calculation result. It is preferable that the gradient image is generated from the absolute value of the gradient vector calculated by using the absolute value of the gradient vector calculated by using only the negative value.

(In Expressions 7 and 8, (x, y) represents the coordinates of the pixel of interest, and I(x, y) is the luminance value of the pixel (x, y). G _x (x, y) and G _y. (X, y) are a first direction difference and a second direction difference in two orthogonal directions, k and s are variables that control the mask that takes the difference, and ω is a variable that controls the orthogonal axis that takes the difference. When ω=1, the difference is on the vertical and horizontal axes, and when ω=0, it is the difference on the diagonal orthogonal axes.γ and θ are not 0 at the same time, and (x, y) is set as the pixel of interest. It is a variable that controls the positional relationship between two pixels that take the difference.)

For example, when ω=1 and γ=θ=1, Equations 7 and 8 become Equations 11 and 12 that take the difference in the horizontal direction and the vertical direction.

(In Expression 11 and Expression 12, the meanings of symbols are the same as those in Expression 7 and Expression 8.)

In the difference results in the two orthogonal directions calculated by Equations 7 and 8, a positive value indicates that the brightness changes from “bright” to “dark”, and a negative value conversely changes from “dark” to “bright”. Since only the positive value is used from the calculation result, the change from “bright” to “dark” is shown, and the negative value is shown from “dark” to “bright”. be able to. Therefore, when a gradient image is generated using Equations 9 and 10, for example, when a convex defect is present, half of the convex defect is emphasized with a positive value and the other half is emphasized with a negative value. To be done. Therefore, for example, the defect can be easily recognized by displaying the positive emphasis part emphasized with a positive value and the negative emphasis part emphasized with a negative value in different colors.

The defect candidate determination unit 426, for example, the defect candidate inside the reflection mirror image extracted by the defect candidate extraction unit 423 inside the reflection mirror image, the defect candidate outside the reflection mirror image extracted by the defect candidate outside the reflection mirror image extraction unit 424, and the gradient image generation unit. The defect candidate is determined from the gradient image generated by 425. Specifically, in the generated gradient image, a positive emphasis portion and a negative emphasis portion appear in the area of the defect candidate inside the reflection mirror image or the area of the defect candidate outside the reflection mirror image, and the positive emphasis portion. And, when the number of pixels of the negative emphasis portion is equal to or larger than a predetermined threshold value, the area is determined as a defect candidate, and in other cases, it is preferable to determine that the area is not a defect.

The defect candidate storage unit 427 is composed of, for example, a memory or the like, and is configured to store the barycentric coordinates of the defect candidate determined by the defect candidate determination unit 426.

(Determination means 43)
The determining unit 43 determines, for example, when a defect exists on the surface M to be inspected in the defect candidates extracted by the defect candidate extracting unit 42 when the moving distance and the moving angle between at least two images at different imaging times are different. Continuity determining means 431 for determining a defect and a reference value storage means 432 for storing the reference moving distance and the reference moving angle when they are within the range of the reference moving distance and the reference moving angle assumed to move. Have The moving distance and the moving angle of the defect candidate are preferably viewed, for example, between three or more images consecutive in the order of the imaging time, but the number of images can be arbitrarily determined according to the surface M to be inspected. Further, depending on the surface M to be inspected, it may be seen between two images. The moving distance of the defect candidate is, for example, the length of a straight line connecting the defect candidates between the images at different imaging times, and the moving angle of the defect candidate is, for example, the angle of the straight line connecting the defect candidates between the images at different imaging times. Is.

Since the camera lens has a distortion aberration, even if the moving speed by the moving means 30 is constant, the movement locus of the defect existing on the surface M to be inspected is not a straight line. By making a determination based on this, it is possible to make a highly accurate determination. The reference movement distance and the reference movement angle which are supposed to move when a defect exists on the surface M to be inspected are, for example, marked on the surface M to be inspected in advance, and moved by the moving means 30 at a constant speed while imaging means. It is preferable that a plurality of reference images having different image capturing times are acquired by capturing images at a constant time interval with 20, and the reference images are set based on the mark movement locus obtained by combining the plurality of reference images. FIG. 7 shows an example of the mark movement locus.

As shown in FIG. 7, the moving distance of the mark depends on the position of the defect in the image. The reference movement distance may be set within a predetermined range according to the position (that is, coordinates) of the defect in the image based on the mark movement locus, but in the continuity determination means 431, in addition to the reference movement distance, Since the determination is made based on the reference movement angle, the range of the reference movement distance may be relatively wide. For example, if the moving distance is negative (the moving direction is opposite) and the moving distance is larger than the predetermined distance threshold, it is necessary to exclude it. It is preferable. It is preferable that the distance threshold is determined corresponding to the mark moving time based on the mark moving locus, for example.

Further, as shown in FIG. 7, the moving angle of the mark also differs depending on the position of the defect in the image. The reference movement angle is preferably set within a predetermined range according to the position (that is, coordinates) of the defect in the image based on the mark movement locus. For example, it is preferable to draw an approximate straight line of the mark movement locus and use this approximate straight line as a reference line to set the reference movement angle based on the angle of each reference line. Since the mark movement locus is curved as shown in FIG. 7, the mark movement locus is divided into a plurality of pieces in the mark movement direction as shown in FIG. Then, it is preferable to draw an approximate straight line for each. Although FIG. 8 shows the case where the mark is divided into three parts in the moving direction, the mark may be divided into two parts or four or more parts depending on the mark moving path. FIG. 9 shows an example of the approximate straight line. In FIG. 9, the white broken line is the division position of the mark movement locus.

Further, when the approximate straight lines of the moving loci of the adjacent marks are separated from each other by a predetermined value or more in the width direction with respect to the moving direction of the marks, a dividing line is drawn to divide them between them in the width direction, and the approximate straight line and the dividing line are drawn. The reference line is preferable. That is, when adjacent approximate straight lines are separated from each other, it is preferable to add a dividing line that divides the approximate straight lines to the reference line.
This is to increase the accuracy of defect determination. The dividing line can be drawn by projective transformation, for example.

Specifically, it is preferable that the range of the reference movement angle be the range of the angle between two adjacent reference lines according to the position (that is, the coordinate) of the defect. This is because the defect can be detected more easily in a short time. The moving angle, the reference moving angle, and the angle of the reference line in the determining unit 43 are angles with respect to the angle basic straight line extended in the width direction with respect to the moving direction of the mark. FIG. 10 shows an example of the angle basic straight line. In FIG. 10, the thick white solid line is the angle basic straight line, and the white broken line is the division position of the mark movement locus. That is, if the movement angle of the defect candidate extracted by the defect candidate extraction means 42 is between the angles of two reference lines adjacent to each other, the continuity determining means 431 determines the reference movement angle. It is preferably configured so as to be judged to be within the range.

The surface defect inspection apparatus 1 is used, for example, as follows. FIG. 11 shows a procedure of a surface defect inspection method using the surface defect inspection apparatus 1. In this surface defect inspection method, first, the surface M to be inspected irradiated with light from the illumination means 10 is imaged by the imaging means 20, and the position of the surface M to be inspected with respect to the imaging means 20 is moved relatively by the moving means 30. A plurality of images having different imaging times are acquired (step S110; imaging procedure). The image captured by the image capturing unit 20 is stored in the image storage unit 41.

Next, for example, a defect candidate is extracted from each of a plurality of images obtained by the image capturing procedure (step S110) at different image capturing times (step S120; defect candidate extracting procedure). In the defect candidate extraction procedure (step S120), for example, first, the preprocessing unit 421 performs preprocessing on the image obtained by the imaging unit 20 as described above (Step S121; preprocessing procedure).

Then, for example, the reflection mirror image area detection unit 422 detects the reflection mirror image area of the fluorescent lamp which is the illumination light source 11 from the preprocessed image (step S122; reflection mirror image area detection procedure). When the reflection mirror image area cannot be detected (step S122:N), the process is ended, and when the reflection mirror image area is detected (step S122:Y), the process proceeds to the next procedure. Next, for example, the gradient image generation means 425 generates a gradient image as described above from the preprocessed image (step S123; gradient image generation procedure).

Then, for example, the defect candidate in the reflection mirror image is extracted by the defect candidate extraction unit 423 in the reflection mirror image in the reflection mirror image detected by the reflection mirror image area detection unit 422 as described above (step S124; In the reflection mirror image). Defect candidate extraction procedure). Further, for example, the outside-reflector-mirror-image defect candidates are extracted by the outside-reflector-mirror-image defect candidate extracting means 424 as described above with respect to outside the reflecting-mirror image detected by the reflecting-mirror-image area detecting means 422 (step S125; outside the reflecting-mirror image). Defect candidate extraction procedure).

Then, for example, by the defect candidate determination unit 426, the gradient image generated in the gradient image generation unit (step S123), the defect candidate in the reflection mirror image extracted in the defect candidate extraction procedure in the reflection mirror image (step S124), and the reflection mirror image. Defect candidates are determined from the defect candidates outside the reflection mirror image extracted in the external defect candidate extraction procedure (step S125) (step pS126; defect candidate determination procedure). The defect candidates determined by the defect candidate determination means 426 are stored in the defect candidate storage means 427.

After the defect candidates are extracted by the defect candidate extraction procedure (step S120), for example, in the determination unit 43, the continuity determination unit 431 determines whether or not the defect candidates extracted as described above are defects ( Step S130; determination procedure). That is, in the extracted defect candidate, the moving distance and the moving angle between at least two images at different imaging times are assumed to move when the surface M to be inspected has a defect and the reference movement. If it is within the angle range, it is determined to be a defect.

Specifically, first, as shown in FIG. 12, among the four images that are consecutive in the order of the imaging time, those in which the movement distance of the defect candidate is within the range of the reference movement distance are extracted and grouped. (Step S131; moving distance determination procedure). Specifically, for example, the image with the latest imaging time is set as the main frame, one of the defect candidates extracted in this frame is extracted, and the main frame is extracted from the defect candidates extracted in other images. A defect whose distance to the defect candidate extracted in step is within the range of the reference distance is extracted. In FIG. 12, ◯ is one defect candidate extracted in this frame, Δ is a defect candidate extracted from the frame one frame before this frame, □ is a defect candidate extracted from a frame two frames before this frame, ▽ is the defect candidate extracted from the frame three frames before this frame, and △, □, ▽ are the reference movement distances in other frames for one defect candidate extracted in this frame indicated by ◯. It is a defect candidate within the range of.

The reference movement distance is, for example, as described above, larger than 0 and not larger than a predetermined distance threshold. As the predetermined distance threshold, for example, the maximum value of the movement amount of the center of the defect candidate in all frames can be taken. For example, the defect candidate in the frame before the main frame indicated by X in FIG. 12 has a movement direction opposite to that of the defect candidate extracted in the main frame indicated by ◯ (the movement distance is negative), or , O are excluded because they are farther than the predetermined distance threshold with respect to the defect candidates of this frame indicated by ◯. Note that, with respect to one defect candidate extracted in this frame, there may be no defect candidate within the range of the reference movement distance in all other frames. For example, when a defect candidate is extracted between four images that are consecutive in the order of imaging time, even if the second and third frames are not detected, the defect candidates detected in the first and fourth frames should be within the range of the reference movement distance. For example, the defect candidates extracted in the first and fourth frame images may be grouped. This is because some defect candidates may not be detected. This is sequentially performed for each defect candidate extracted in this frame.

Next, for example, with respect to the defect candidates grouped in the movement distance determination procedure (step S131), if the movement angle between adjacent defect candidates is within the range of the reference movement angle, it is determined to be a defect (step S132; Moving angle judgment procedure). Specifically, for example, as described above, the reference line is based on the mark movement locus obtained by marking the surface M to be inspected in advance and moving the surface M at a constant speed while capturing an image at a constant time interval. , And a reference movement angle is set based on the angles of the respective reference lines. In the grouped defect candidates, the movement angle between the adjacent defect candidates is the difference between the two reference lines to which the defect candidate is adjacent. If it is between the angles, it is determined to be within the range of the reference movement angle. In addition, in FIG. 12, a straight line connecting adjacent defect candidates is indicated by a broken line. This is sequentially performed for each group of defect candidates that are grouped.

After that, the display unit 50 displays the determination result obtained in the determination procedure (step S130) (step S140; display procedure). The display unit 50 displays, for example, a circle or the like on the center of gravity of the defect on a display or the like. Thereby, the surface defect can be inspected.

As described above, according to the present embodiment, the extracted defect candidate has a reference movement distance that is assumed to move when there is a defect on the surface M to be inspected between at least two images at different imaging times, and Since it is determined that the defect is a defect when it is within the range of the reference movement angle, it is possible to see the continuity of the defect over time, the movement speed of the surface M to be inspected is not constant, or the movement locus is a straight line. Even if it does not occur, erroneous detection of noise or the like can be eliminated with high accuracy, and the omission of defect detection can be reduced. Therefore, the defect can be detected with high accuracy.

In particular, if the reference movement angle is set based on the mark movement loci obtained by picking up a plurality of reference images at different image pickup times while moving the mark M on the surface M to be inspected in advance, The defect can be easily detected with high accuracy.

In addition, draw an approximate straight line of the mark movement locus, and if the approximate straight line is separated by a predetermined value or more in the width direction, draw a dividing line between them, and the approximate straight line and the angle of each reference line with the dividing line If the reference movement angle is set based on, the defect can be detected more easily and in a short time.

Although the present invention has been described with reference to the embodiments, the present invention is not limited to the above embodiments, and various modifications can be made. For example, in the above-described embodiment, each constituent element has been specifically described, but not all constituent elements may be included, or other constituent elements may be included.

Further, in the above embodiment, the case of inspecting the painted surface of the automobile body was specifically described, but the present invention is not limited to the body of the automobile and is also applied to the case of inspecting the surface of other coated products. can do. Further, the present invention can be applied not only to a painted surface but also to an inspection of a surface having a reflective property.

Furthermore, although the defect candidate extraction means 42 and the defect candidate extraction procedure (step S120) are specifically described in the above embodiment, the defect candidates may be extracted by other configurations and procedures.

DESCRIPTION OF SYMBOLS 1... Surface defect inspection apparatus, 10... Illuminating means, 11... Illuminating light source, 20... Imaging means, 21... Camera, 30... Moving means, 40... Defect detecting means, 41 Image storing means, 42... Defect candidate extracting means, 421 ... preprocessing means, 422... reflecting mirror image area detecting means, 423... reflecting mirror image inside defect candidate extracting means, 424... outside reflecting mirror image defect candidate extracting means, 425... gradient image generating means, 426...
Defect candidate determination means, 427... Defect candidate storage means, 43... Judgment means, 431... Continuity judgment means, 432... Reference value storage means 432, 50... Display means, M... Inspected surface

Claims (4)

An imaging procedure for capturing a plurality of images at different imaging times by moving the position of the surface to be inspected relative to the imaging means while imaging the surface to be inspected illuminated with light by the imaging means,
A defect candidate extraction procedure for extracting defect candidates from a plurality of images at different imaging times obtained by the imaging procedure,
With respect to the defect candidate extracted in the defect candidate extraction procedure, a reference movement in which a movement distance and a movement angle between at least two images having different imaging times are assumed to move when a defect exists on the surface to be inspected. A surface defect inspection method, comprising: a determination procedure of determining a defect if the distance and the reference movement angle are within a range. Illuminating means for irradiating the surface to be inspected with light,
An image pickup means for obtaining an image by picking up the surface to be inspected illuminated by the illuminating means;
Moving means for moving the position of the surface to be inspected relative to the imaging means,
Defect detecting means for detecting a defect from a plurality of images picked up by the image pickup means at arbitrary time intervals while relatively moving the image pickup means and the surface to be inspected by the moving means,
The defect detection means,
Defect candidate extraction means for extracting defect candidates from each of a plurality of images at different imaging times,
In the defect candidates extracted by the defect candidate extraction means, the reference distance is assumed to be such that the movement distance and the movement angle between at least two images at different imaging times are supposed to move when there is a defect on the surface to be inspected. A surface defect inspection apparatus comprising: a determining unit that determines a defect if the distance and the reference movement angle are within a range. The determining means preliminarily adds a mark to the surface to be inspected, and based on a mark movement locus obtained by capturing a plurality of reference images at different image capturing times by the image capturing means while moving by the moving means, The surface defect inspection apparatus according to claim 2, wherein the reference movement angle is set. The determination means draws an approximate straight line of the mark moving locus, and when adjacent approximate straight lines in the width direction with respect to the moving direction of the mark are apart from each other by a predetermined value or more, a dividing line that divides them in the width direction. And the reference movement angle is set based on the angle of each of the reference lines with the approximation line and the dividing line as the reference line, and the movement angle of the defect candidate is two of the two adjacent to the defect candidate. The surface defect inspection apparatus according to claim 3, wherein when the angle is between the angles of the reference line, it is determined that the angle is within the range of the reference movement angle.