KR20110095344A - Device for examining defect of molded sheet - Google Patents

Abstract

The defect inspection apparatus includes an image pickup unit 5 ₁ to 5 _n for imaging a two-dimensional image of a molded sheet a plurality of times to generate a plurality of two-dimensional image data, and a linear light source image in a part of the imaging region in the molded sheet. In order to change the position of the linear light source for illuminating the molded sheet to be projected, and the linear light source image in the molded sheet, the molded sheet is intersected with the longitudinal direction of the linear light source, and in a direction perpendicular to the thickness direction of the molded sheet. conveying the conveying device and the image pickup section (5 ₁ ~ 5 _n) line image analysis unit for the defect (61 ₁ ~ 61 _n) for detecting defects with a plurality two-dimensional line fault detection algorithm from the data generated by and that And point defect image analysis units 62 ₁ to 62 _n for detecting a defect from a plurality of two-dimensional image data generated by the imaging units 5 ₁ to 5 _n by a point defect detection algorithm. Thereby, the defect inspection apparatus of a molded sheet which can detect various defects more reliably can be provided.

Description

DEFECT INSPECTION DEVICE OF MOLDED SHEET {DEVICE FOR EXAMINING DEFECT OF MOLDED SHEET}

This invention relates to the defect inspection apparatus which inspects the defect of a molded sheet | seat, such as an optical film (especially a long optical film wound and stored by web shape), such as a polarizing film and retardation film.

The defect inspection apparatus of the conventional molding sheet uses a 1-dimensional camera called a line sensor, illuminates a molding sheet with linear light sources, such as a fluorescent tube, and lengthens the surface of a molding sheet along the longitudinal direction of the molding sheet. One piece of still image data is acquired by scanning with one-dimensional camera from one end to the other end of the direction, and defects of the molded sheet are inspected based on this one piece of still image data. This still image data usually includes a linear light source image. The linear light source image is an image of light emitted from the linear light source, positively reflected by the molding sheet and reaching the camera when the molded sheet is disposed between the linear light source and the camera and the reflective surface. When the molding sheet is arrange | positioned in between, it is an image of the light radiate | emitted from a linear light source, permeate | transmitted the molding sheet, and reached the camera. In this defect inspection apparatus, when the width | variety of a shaping | molding sheet is wide, a plurality of line sensors are arranged side by side in a width direction so that the whole width direction area | region of a shaping | molding sheet can be examined.

However, in this conventional defect inspection apparatus, the defect in the molded sheet is inspected based on one piece of still image data (hereinafter, simply referred to as "image data") for the entire region of the molded sheet. The positional relationship between the target pixel and the linear light source image becomes one predetermined positional relationship. The defect may appear on the image data only when the positional relationship between the inspection target pixel (the main pixel) and the linear light source image is in a specific positional relationship. For example, bubbles, which are one type of defect, often appear on image data only when they are at or near the periphery of the linear light source image. Therefore, a defect may not be detected depending on the position. Therefore, the said conventional defect inspection apparatus has only the limited defect detection capability.

Therefore, the applicant of this application is applying for the defect inspection apparatus of the molded sheet which improved the defect identification capability with respect to the said conventional defect inspection apparatus (refer patent document 1). In this defect inspection apparatus, moving image data (imaging position on the molding sheet is formed using a two-dimensional camera called an area sensor while illuminating the molded sheet with a linear light source such as a fluorescent tube and continuously conveying the molded sheet in a predetermined direction. A plurality of different image data) is acquired and defects in the molded sheet are inspected based on the moving picture data. In this defect inspection apparatus, it is possible to determine whether there is a defect based on a plurality of pieces of image data in which the positional relationship between the inspection target pixel and the linear light source image is different, so that the defect can be detected more reliably than the conventional defect inspection apparatus. Therefore, this defect inspection apparatus improves a defect detection capability compared with the conventional defect inspection apparatus. Using this moving picture data, it is also possible to see the movement of defects in the illumination image.

Japanese Unexamined Patent Publication No. 2007-218629 (published 30 August 2007)

However, the examination of the inventors of the present application found that the defect inspection apparatus described in Patent Document 1 also has room for improvement in the defect detection capability.

That is, in the defect inspection apparatus described in Patent Document 1, a defect is detected by the following image processing from each of a plurality of (multi-value) image data captured by an area sensor (Patent Document 1 to [0032] to [0032]). [0035]).

First, multivalued image data are binarized, and white areas and black areas are labeled to be detected. Next, from the white region to be detected, a white region having an area (pixel number) exceeding a predetermined value (a relatively large value suitable for the area of the linear light source image; for example, 2500 pixels) is regarded as a linear light source image and excluded. do. Similarly, from the black region to be detected, a black region having an area exceeding a designated value (a relatively large value suitable for the area of the background region) is regarded as a background region (an image of a defect-free region in the molding sheet). Exclude. In addition, from the white region and the black region to be detected, the white region and the black region having an area of less than a specified value (a relatively small value close to one pixel; for example, nine pixels) are excluded as noise. And the area | region which was not excluded from the white area | region and black area | region of a detection object is detected as a defect.

However, since the defect inspection apparatus described in Patent Literature 1 detects a defect on the basis of inversion of contrast, there is a case where a defect with low contrast on image data may not be detected.

Here, the various defects made into a detection object in this invention are demonstrated. This invention mainly intends to detect the defect (exterior defect) accompanying micro unevenness | corrugation (especially unevenness | corrugation of about several micrometers height) of the molded sheet surface. As a defect with micro unevenness | corrugation, For example, the micro unevenness | corrugation which generate | occur | produced on the surface of a molded sheet originating from a bubble or a foreign material; Indentation (indentation caused by pressing); Bent marks (called "kunic"); Indentation by a conveyance roll which generate | occur | produced at the time of conveyance by a conveyance roll at the time of manufacture of a molded sheet (called a "stripe pattern"), etc. are mentioned. Defects with such minute irregularities are very difficult to detect in a defect inspection apparatus using a conventional line sensor. This invention mainly aims at detecting this kind of defect.

In the present specification, for convenience, the micro-concave and convex portions are locally concentrated (the diameter of the convex portion is about 1 mm or less (or about several pixels or less when the resolution of the imaging device is 200 µm / pixel)), for example, Bubbles, foreign matters, scratches, and the like are called point defects, and fine irregularities are linearly connected and are referred to as line defects. Typical line defects, such as stripes, have a width in excess of about 10 mm (a width exceeding several tens of pixels when the resolution of the imaging device is 200 µm / pixel), and typically in the order of tens of centimeters, and in some cases It has a width exceeding several tens of centimeters. The Kunic has a width of about 10 mm or less (or about several tens of pixels or less when the resolution of the imaging device is 200 µm / pixel), typically about several mm, and exhibits intermediate properties between point defects and typical line defects. Have

The example of the defect image (movie) imaged in the imaging part of the defect inspection apparatus of patent document 1 is as showing in FIGS. 14-15 of patent document 1. As shown in FIG. In FIGS. 14-15 of patent document 1, the five consecutive frames which comprise a moving image are shown by (a)-(e) in order of time. If this moving picture is a normal moving picture for television, the time interval (frame rate) between frames is 1/30 second. The time interval between frames depends on the characteristics of the imaging unit. In the defect inspection apparatus of patent document 1, the imaging part is a direction from the imaging part toward the center of the imaging area | region (a rectangle shown by the broken line to the molded sheet surface in FIG. 1 of patent document 1), and the conveyance direction of a molded sheet. In order to achieve this acute angle, the illumination image (reflected image of the linear light source) is included in a part of the imaging area, and an area without the illumination image (background area) is disposed on both sides of the illumination image in the imaging area. . Therefore, the upward direction in the video shown in FIGS. 14-15 of patent document 1 corresponds with the conveyance direction of a molded sheet | seat. In the imaging region in the molded sheet, since the defect moves in the conveying direction of the molded sheet, in the moving picture shown in FIGS. 14 to 15 of Patent Document 1, the defect moves from the bottom up (the light image is a white stripe). Reflected in the area of the shape).

FIG. 14 of Patent Document 1 is an example of a moving picture of an imaging region (size in a direction orthogonal to the conveying direction: 5 mm) including bubbles (point defects) picked up by the imaging unit, wherein bubbles are portions in which contrast is reversed. It is showing. In this video, the bubble is not visible in the first frame (a), but in the second frame (b) where the point defects are close to the illuminated image, the bubble begins to appear and the third bubble is positioned at the edge of the illuminated image. In frame (c) the bubbles are seen relatively well, and in the fourth and fifth frames (d) (e) in which the bubbles enter the illumination image, the bubbles are buried in the illumination light and are not visible.

FIG. 15 of Patent Literature 1 is an example of a moving image of an image capturing region (size in a direction orthogonal to the conveying direction: 5 mm) including a Kunic captured by the image capturing unit. There is no, and as the pass through passes, the illumination image, which should originally be reflected in the rectangular white area, deforms with time.

FIG. 15 of patent document 1 is an example of the moving picture of the imaging area | region (size of the direction orthogonal to a conveyance direction: 200 mm) containing the stripe (line | wire defect) picked up by the imaging part. In the first to third frames (a) to (c), the illumination image is somewhat deformed into a bow shape, but this bow deformation is not caused by a defect but is caused by pulling a molded sheet. will be. On the other hand, in the fourth frame, the illumination image is relatively largely deformed into an S shape. The illumination image is deformed beyond this much because there is a stripe in the deformed position. Therefore, it is desirable to detect a portion where the illumination image is deformed by this much or more as a defect.

However, the defect inspection apparatus of Patent Literature 1 has a threshold value that separates the bright region (the defect region inside the linear light source image and the linear light source image) and the dark region (the defect region within the background region and the background region) in the binarization process. In some cases, it is understood that a defect with low contrast on the image data may not be detected.

That is, in the defect inspection apparatus of Patent Literature 1, when the defect is relatively high in contrast (luminance change due to defect) on the image data, the defect profile is in the vertical direction (molding sheet conveying direction) of FIG. 3. As shown, a change in luminance due to a defect is observed to span the threshold of the binarization process. In more detail, it is larger than the luminance value of the minimum point (valley part in FIG. 3) corresponding to the defect observed in the linear light source image as the dark area, and more than the luminance value of the maximum point on both sides of the minimum point. Becomes smaller. The threshold value of the binarization process is larger than the luminance value of the maximum point (mountain portion in FIG. 3) corresponding to a defect observed as a bright area outside the linear light source image, and the threshold value of the minimum point on both sides of the maximum point. It becomes larger than the luminance value. Therefore, defects observed as dark areas in the linear light source image and defects observed in bright areas outside the linear light source image are also detected. Therefore, the defect inspection apparatus of patent document 1 can reliably detect the defect with comparatively high contrast.

On the other hand, in the defect inspection apparatus of Patent Literature 1, when the defect is low in contrast on the image data, the relationship between the vertical direction luminance profile of the defective portion and the threshold value of the binarization process is the vertical direction of FIG. 4. The relationship shown in the brightness profile, i.e., the change in brightness due to the defect does not exceed the threshold of the binarization processing. As a reference, the following formula

(Luminance change amount due to defect) <｛(luminance level of linear light source image region)-(luminance level of background region)} / 2. (One)

When it satisfy | fills, it may become the relationship shown in FIG. When the relationship shown in FIG. 4 is established, a defect will be missed.

However, even when the contrast of the defect on the image data is low and the equation (1) is satisfied, as shown in the vertical luminance profile of FIG. 5, the luminance variation due to the defect is over the threshold of the binarization processing. In the case of observation, a defect can be detected. Therefore, even when the defect inspection apparatus of patent document 1 satisfy | fills Formula (1), it can detect a defect according to the relationship between the brightness change by a defect and the threshold value of a binarization process.

Moreover, the defect inspection apparatus of patent document 1 uses a moving image, and in a moving image, the movement that a defect image approaches a linear light source image for every coma, passes through a linear light source image, and moves away from a linear light source image is observed. do. Therefore, even when the contrast of a defect in image data is low and Equation (1) is satisfied, the relationship between the luminance change caused by the defect and the threshold value of the binarization processing is at least one of the plurality of images constituting the moving image. If there is an image that becomes the relationship shown in Fig. 5, the defect can be detected.

As described above, in the defect inspection apparatus of Patent Literature 1, the possibility of missing a defect increases as the amount of change in luminance due to the defect decreases around the formula (1). Therefore, the defect inspection apparatus of patent document 1 has room for improvement about the reliability of defect detection with low contrast.

This invention is made | formed in view of said problem, and the objective is to provide the defect inspection apparatus of a molded sheet which can detect various defects more reliably.

In order to solve the said subject, the defect inspection apparatus which concerns on this invention is a defect inspection apparatus which detects the defect of a molded sheet, The imaging which produces | generates several two-dimensional image data by imaging two-dimensional image of the said molding sheet multiple times. The position where the image of the linear light source for illuminating the said molding sheet, and the image of the said linear light source in the said molding sheet is projected so that an image of a linear light source may be projected on a means and a part of the area | region image | photographed in the said molding sheet. Generated by the imaging means and the moving means for moving at least one of the molding sheet and the linear light source to move in a direction crossing the longitudinal direction of the linear light source and orthogonal to the thickness direction of the molding sheet so that is changed. And a line defect detecting means for detecting a line defect from the plurality of pieces of two-dimensional image data. The detection means fits an edge of the linear light source image in the two-dimensional image data with a function curve, and detects a point where the distance between the edge of the linear light source image and the function curve is equal to or greater than a first threshold value as a line defect. Algorithm or a line defect detection algorithm that obtains a curvature in the vicinity of each pixel at the edge of the linear light source image in the two-dimensional image data, and detects a point at which the curvature is equal to or greater than the second threshold value as a line defect. It is characterized by detecting a line defect.

According to the above configuration, the line defect can be detected more reliably.

It is preferable that the defect inspection apparatus further includes point defect detection means for detecting point defects from a plurality of two-dimensional image data generated by the imaging means. Thereby, not only a line defect but a point defect can also be detected.

In the defect inspection apparatus, the point defect detecting means represents a change in luminance depending on a position along the straight line in the two-dimensional image data as a luminance profile, and the movement time between plots is represented by a plot group of the luminance profile. Assuming the quality point moving to be constant, the luminance value of the interest plot is determined from the velocity vector of the quality point between two plots immediately before the plot of interest and the acceleration vector of the quality point between three plots immediately before the plot of interest. A point defect detection algorithm for predicting and detecting a point at which a difference between the predicted luminance value and the actual luminance value is equal to or greater than a third threshold value as a point defect, or smoothing the two-dimensional image data, and smoothing the two-dimensional image data and the original Obtain the difference of two-dimensional image data as difference image data, and calculate the difference in the difference image data. Detecting a point defect by a point defect detection algorithm that detects a point at which the degree value is greater than or equal to the fourth threshold and a point at which the luminance value is less than or equal to the fifth threshold (the fifth threshold is smaller than the fourth threshold) as a point defect. It is preferable. Thereby, compared with the technique of patent document 1, it becomes possible to detect a point defect with low contrast (for example, the defect which shows a small brightness change like FIG. 4) more reliably. Therefore, both a point defect and a line defect can be detected reliably.

In order to solve the said subject, the defect inspection apparatus which concerns on this invention is a defect inspection apparatus which detects the defect of a molded sheet, The imaging which produces | generates several two-dimensional image data by imaging two-dimensional image of the said molding sheet multiple times. The position where the image of the linear light source for illuminating the said molding sheet, and the image of the said linear light source in the said molding sheet is projected so that an image of a linear light source may be projected on a means and a part of the area | region image | photographed in the said molding sheet. Generated by the imaging means and the moving means for moving at least one of the molding sheet and the linear light source to move in a direction crossing the longitudinal direction of the linear light source and orthogonal to the thickness direction of the molding sheet so that is changed. And a point defect detecting means for detecting a point defect from the plurality of pieces of two-dimensional image data. The detection means represents a change in luminance depending on the position along the straight line in the two-dimensional image data as a luminance profile, and assumes a quality point for moving the plot group of the luminance profile so that the movement time between the plots becomes constant. The luminance value of the plot of interest is predicted from the velocity vector of the mass point between two plots immediately before the plot of interest and the acceleration vector of the mass point between three plots immediately before the plot of interest, and the predicted luminance value and actual A point defect detection algorithm that detects a point at which a difference in luminance value is equal to or greater than a third threshold value as a point defect, or smoothes the two-dimensional image data, and differentiates the difference between the smoothed two-dimensional image data and the original two-dimensional image data. Obtained as data, the point where the luminance value in the differential image data is equal to or greater than the fourth threshold value, and the luminance value The point defect is detected by a point defect detection algorithm that detects a point below the fifth threshold (the fifth threshold is smaller than the fourth threshold) as the point defect.

Thereby, compared with the technique of patent document 1, it becomes possible to detect a point defect with low contrast (for example, the defect which shows a small brightness change like FIG. 4) more reliably.

It is preferable that the said defect inspection apparatus is further provided with the line defect detection means which detects a line defect from the some two-dimensional image data produced | generated by the said imaging means. Thereby, not only a point defect but a line defect can also be detected.

As mentioned above, this invention has the effect that the defect inspection apparatus of a molded sheet which can detect various defects more reliably can be provided.

BRIEF DESCRIPTION OF THE DRAWINGS It is a functional block diagram which shows the principal part structure of the defect inspection apparatus which concerns on one Embodiment of this invention.
It is a schematic diagram which shows the overview of the said defect inspection apparatus.
3 is a diagram for explaining a problem of the prior art.
4 is a diagram for explaining a problem of the prior art.
5 is a diagram for explaining a problem of the prior art.
6 is a diagram for explaining an example (edge profile method 1) of a defect detection algorithm.
7 is a diagram for explaining another example (edge profile method 2) of a defect detection algorithm.
8 is a diagram for explaining another example (peak method) of a defect detection algorithm.
9 is a diagram for explaining another example (peak method) of a defect detection algorithm.
10 is a diagram for explaining another example (peak method 2) of a defect detection algorithm.
11 is a diagram for explaining another example (peak method 2) of a defect detection algorithm.
12 is a diagram for explaining another example (edge curve method 1) of a defect detection algorithm.
13A is a diagram for explaining another example (edge curve method 2) of a defect detection algorithm.
13B is a diagram for explaining another example (edge curve method 2) of a defect detection algorithm.
13C is a diagram for explaining another example (edge curve method 2) of a defect detection algorithm.
It is a functional block diagram which shows the principal part structure of the defect inspection apparatus which concerns on other embodiment of this invention.
Fig. 15 is a functional block diagram showing a main part configuration of a defect inspection apparatus according to still another embodiment of the present invention.

[Embodiment 1]

EMBODIMENT OF THE INVENTION One Embodiment of this invention is described below, referring drawings.

The defect inspection apparatus which concerns on this embodiment detects the defect of a molded sheet. The defect inspection apparatus which concerns on this embodiment is suitable for the test | inspection of the molded sheet which consists of resins, such as a light transmissive molding sheet, especially a thermoplastic resin. As a molding sheet which consists of resin, the process which gives a smoothness and gloss to the surface by passing the thermoplastic resin extruded from the extruder through the gap of a roll, is performed, and is taken out while cooling a conveyance roll top with a take-up roll. The molded thing is mentioned. The thermoplastic resin applicable to this embodiment is, for example, methacryl resin, methyl methacrylate-styrene copolymer, polyolefin such as polyethylene or polypropylene, polycarbonate, polyvinyl chloride, polystyrene, polyvinyl alcohol, triacetyl Cellulose resins and the like. The molded sheet may consist of only one of these thermoplastic resins, or may be a laminate of a plurality of kinds of these thermoplastic resins (laminated sheets). Moreover, the defect inspection apparatus which concerns on this embodiment is suitable for the inspection of the optical film, such as a polarizing film and retardation film, especially the long optical film wound and stored and transported in a web shape. Moreover, what kind of thickness may be sufficient as a molded sheet | seat, may have a comparatively thin thickness generally called a film, and may have a comparatively thick thickness generally called a board | plate.

Examples of defects of the molded sheet include point defects such as bubbles (such as those generated during molding), fish eyes, foreign substances, tire traces, scars, and scratches; Slicing, streaks (such as those caused by differences in thickness), and the like.

The structure of the defect inspection apparatus 1 which concerns on this embodiment is demonstrated below based on FIG. 1 and FIG. 1 is a functional block diagram showing the main part configuration of the defect inspection apparatus 1. 2 is a schematic diagram showing an overview of the defect inspection apparatus 1. In addition, in FIG. 2, the contrast of the shaping | molding sheet surface is shown inverted so that the member overlapped with the shaping | molding sheet can be identified easily. Therefore, the black region of the molded sheet surface in FIG. 2 is actually a bright region, and the white region of the molded sheet surface in FIG. 2 is actually a dark region.

The defect inspection apparatus 1 n-forms the shaping | molding sheet 2 illuminated by the linear light source 4, conveying the rectangular shaping | molding sheet 2 by the conveying apparatus (moving means) 3 in a fixed direction. (n is an integer of 2 or more), and the imaging unit (imaging means) 5 ₁ to 5 _{n of} the imaging unit a plurality of times to generate a plurality of two-dimensional image data by each of the imaging unit (5 ₁ to 5 _n ), The analysis device 6 detects a defect of the molding sheet 2 based on the generated two-dimensional image data.

Defect inspection apparatus 1 is picked up by the image pickup area (image pickup section (5 ₁ ~ 5 _n) of the transfer device (transfer means) 3, a shaped sheet (2) for transporting the formed sheet (2) A linear light source 4 for illuminating the molding sheet 2 so that an image of the linear light source 4 is projected onto a part of the rectangular sheet indicated by broken lines on the surface of the molding sheet 2 in FIG. Reflected image of the light source 4 (image of the linear light source 4 formed as a result of the direct light from the linear light source 4 being reflected by the molding sheet 2 and reaching the imaging units 5 ₁ to 5 _n ). ) And a reflection image of the molding sheet 2 (forming sheet 2 formed as a result of the scattered light from the linear light source 4 being reflected by the molding sheet 2 and reaching the imaging units 5 ₁ to 5 _n ). ) imaging section (5 ₁ ~ 5 _n) a plurality of times to the two-dimensional image pick-up an image including the image), and a plurality of secondary And a analyzer (6) for detecting defects of the formed sheet (2) by the basis of the image data, the image processing algorithm (fault detection algorithm).

The conveying apparatus 3 is the direction orthogonal to the thickness direction, especially the longitudinal direction of the shaping | molding sheet 2 so that the position on which the image of the linear light source 4 in the shaping sheet 2 is projected may change. To return. The conveying apparatus 3 is equipped with the sending roller and the receiving roller which convey the shaping | molding sheet 2 to a fixed direction, for example, and measure a conveyance speed with a rotary encoder etc. The conveyance speed is set, for example, at about 2 m to 12 m / minute. The conveyance speed in the conveyance apparatus 3 is set and controlled by the information processing apparatus etc. which are not shown in figure.

The linear light source 4 further has a linear light source 4 such that its longitudinal direction crosses the conveying direction of the molding sheet 2 (for example, a direction orthogonal to the conveying direction of the molding sheet 2). Is reflected so that the reflected image of the molding sheet 2 crosses the imaging region, so that regions (background regions) without the linear light source image exist on both sides of the reflected image of the linear light source 4 in the imaging region. The linear light source 4 is not particularly limited as long as it emits light that does not affect the composition and properties of the molded sheet 2, and is, for example, a fluorescent lamp (especially a high frequency fluorescent lamp), a metal halide lamp, and a halogen transmission. Light and so on. Further, the linear light source 4, from the transmission image (linear light source (4) disposed in position, and the linear light source (4) opposite to the other across a forming sheet (2) the image pickup section (5 ₁ ~ 5 _n) From the linear light source 4 formed by direct light passing through the molding sheet 2 and reaching the imaging units 5 ₁ to 5 _n , and the transmission image of the molding sheet 2 (from the linear light source 4). and the scattered light is transmitted through the shaped sheet (2) the image pickup section (5 ₁ ~ 5 _n) 2-dimensional image is image pickup section (5 ₁ ~ 5 _n) comprises an image of the forming sheet 2) are formed by reaching the You may make it pick up by image.

Each of the imaging units 5 ₁ to 5 _n captures a plurality of two-dimensional images including a reflection image of the linear light source 4 and a reflection image of the molding sheet 2, and generates a plurality of two-dimensional image data. To print. The imaging units 5 ₁ to 5 _n comprise an area sensor composed of an imaging element such as a CCD (Charge Coupled Device) or a CMOS (Complementary Metal-Oxide Semiconductor) for imaging a two-dimensional image. If the defect size is detected by the defect inspection apparatus 1, it depends on the resolution of the imaging section (5 ₁ ~ 5 _n), according to the defect size to be detected selection of the resolution of the imaging section (5 ₁ ~ 5 _n) do. In addition, (the ratio of width to height) of the three-dimensional shape of the defect detected by the defect inspection apparatus (1) is basically because they do not depend on the resolution of the imaging section (5 ₁ ~ 5 _n), the defect to be detected type It is not necessary to select the camera resolution accordingly.

The imaging sections 5 ₁ to 5 _{n are} arranged from the imaging sections 5 ₁ to 5 _n so that the direction toward the center of the imaging region of the molding sheet 2 and the conveying direction of the molding sheet 2 form an acute angle. . The imaging areas 5 ₁ to 5 _n have the entire region in the width direction of the molded sheet 2 (the direction orthogonal to the conveyance direction of the molded sheet 2 and orthogonal to the thickness direction of the molded sheet 2). that picked up by the at least one image pickup section (5 ₁ ~ 5 _n), and is disposed are listed along the width direction of the forming sheet (2). By imaging the entire area in the width direction of the forming sheet 2 by the image pickup section (5 ₁ ~ 5 _n), it is possible to inspect the defects of the entire area of the formed sheet (2).

Image pick-up interval of the imaging unit ₍₅₁ ~ 5 _n) (frame rate) is, may be fixed and, even if the user is to be changed by operating the imaging section (5 ₁ ~ 5 _n) itself, and an image pickup unit ₍₅₁ 5 _n ) may be changed by a user operating an information processing device (not shown; can be omitted) connected thereto. The imaging intervals of the imaging units 5 ₁ to 5 _n may be several seconds, or the like, which are time intervals of continuous shooting of the digital still camera, but may be short time intervals, for example, to improve the efficiency of inspection. It is preferable that it is 1/30 second etc. which is the frame rate of general moving image data.

Here, the distance (transfer distance) in which the molding sheet 2 is conveyed in a period from when each imaging unit 5 ₁ to 5 _n captures one two-dimensional image and then photographs the next two-dimensional image, At least 1 / m (m is 2 or more) of the length of the imaging area | region along the conveyance direction of the shaping | molding sheet 2 is set. Thereby, the two-dimensional image containing the same point of the shaping | molding sheet 2 is imaged m times. It is preferable that m is sufficiently larger than 2. By increasing the number of imaging times of the same point of the molded sheet 2, a defect can be inspected with high precision.

As shown in FIG. 1, the analyzer 6 receives a plurality of two-dimensional image data output from each of the imaging units 5 ₁ to 5 _n , and detects a defect based on the plurality of two-dimensional image data. and the detection result (test result) to the output line image analysis for defect portion (line defect detecting means) (61 ₁ ~ 61 _n) and that the image analysis unit for the defect (point defect detecting _{means) (62 1 ~ 62 n)} , A display unit 64 for displaying a detection result (inspection result) and a control CPU 63 for collectively controlling these units are provided.

Each image pick-up unit (5 ₁ ~ 5 _n) a plurality of two-dimensional image data generated by, the image analysis for each line defect portion (61 ₁ ~ 61 _n) and the image analysis for the point defect portion (62 ₁ ~ 62 _n) Is entered.

Each of the line defect image analyzers 61 ₁ to 61 _n and the point defect image analyzers 62 ₁ to 62 _n are each a linear light source image on the forming sheet 2 by a line defect detection algorithm. Defects are detected from a plurality of (multiple frames) two-dimensional image data having different positions of, and the results are output as inspection results. Each of the point defect image analyzers 62 ₁ to 62 _n has a plurality of two-dimensional (multiple frames) two-dimensional images different in position of the linear light source image on the molding sheet 2 by a point defect detection algorithm. Defects are detected from the data, and the results are output as inspection results. For the line defect image analysis unit (61 ₁ ~ 61 _n) and the image analysis for the point defect portion (62 ₁ ~ 62 _n), the formed sheet (2) two-dimensional image a plurality of one-dimensional position of the light source images different in phase Since it is determined whether there is a defect based on the data, it is possible to reliably detect a defect than a conventional defect inspection apparatus.

The line defect detection algorithm and the point defect detection algorithm will be described later. Although the parameters of the line defect detection algorithm and the point defect detection algorithm may be fixed, the information processing apparatus connected to the line defect image analyzers 61 ₁ to 61 _n and the point defect image analyzers 62 ₁ to 62 _n . (Not shown; can be omitted) may be changed by the user operating.

For the line defect image analysis unit (61 ₁ ~ 61 _n), in a case where from the above line of by a fault detection algorithm m of the plurality of two-dimensional image data, L pieces (L ≤ m) of the line defect is detected, the line defect The result of the inspection may be output as a test result, and in another case, the result of the test may be output as a test result. In the case of detection, the information of a line defect position may be output as an inspection result, and in another case, the inspection result may not be output. When the number m of two-dimensional image data is three or more and L is two or more, when the number of two-dimensional image data in which a line defect is detected by the line defect detection algorithm is less than L, the line defect detection result is returned. It will be regarded as a false (what was originally detected as a defect by mistake) and excluded. As a result, the false information can be reduced. Moreover, when outputting the information of a line defect position as a test result, it is necessary to use the line defect detection algorithm which can obtain a line defect position.

In that the image analysis unit for the defect (62 ₁ ~ 62 _n), the point defect is detected, if by means of an algorithm of the point defect is detected from the above L pieces (L ≤ m) of m number of the plurality of two-dimensional image data, the point defect Outputting the result as a test result, and in other cases, outputting a result of no point defect as a test result, and a point defect is detected from L or more of a plurality of m pieces of two-dimensional image data by a point defect detection algorithm. In the case of detection, the information of a point defect position may be output as an inspection result, and in another case, the inspection result may not be output. When the number m of two-dimensional image data is three or more and L is two or more, when the number of two-dimensional image data in which a point defect was detected by the point defect detection algorithm is less than L, the point defect detection result is It will be regarded as a false (what was originally detected as a defect by mistake) and excluded. As a result, the false information can be reduced. In addition, when outputting the information of a point defect position as a test result, it is necessary to use the point defect detection algorithm which can obtain | require the point defect position.

Control CPU (63), the whole area of the line defect image analysis unit (61 ₁ ~ 61 _n) and the point defect image analysis unit (62 ₁ ~ 62 _n) molded sheet (2) by integrating the outputted test result from for Inspection result information corresponding to the data is created, stored in a storage device (not shown), and displayed on the display unit 64. As inspection result information corresponding to the whole area | region of the molding sheet 2, the information which shows whether or not there is a defect in the whole area | region of the molding sheet 2, the defect map of the whole area | region of the molding sheet 2, etc. are mentioned. . In creating the test result information corresponding to the entire area of the formed sheet (2), the defect in at least one of the line image analysis for defect portion (61 ₁ ~ 61 _n) and the point defect image analysis unit (62 ₁ ~ 62 _n) for In the case where this is detected, inspection result information is created as a defect exists.

When the control CPU 63 creates a defect map of the entire area of the molding sheet 2 as inspection result information corresponding to the entire area of the molding sheet 2, the image analysis units 61 ₁ to 61 for line defects are used. _n ) and the point defect image analysis sections 62 ₁ to 62 _n each convert coordinate positions on the two-dimensional image data into coordinate positions on the molding sheet 2 to generate defect position information. To the control CPU 63. As coordinate change processing of each of the line defect image analyzers 61 ₁ to 61 _n and the point defect image analyzers 62 ₁ to 62 _n , for example, paragraphs [0037] to [0041] of Patent Document 1 ] And [0050] to [0053] can be used. The information of the defect map may be output to a marking apparatus (not shown) and an information processing apparatus (not shown), and the marking apparatus may cause the defect position to be marked on the molded sheet 2 based on the defect map. This marking apparatus has, for example, an arm formed along the width direction of the molding sheet 2, a marker head having a pen, and the like, by which the marker head reciprocates the arm image in the width direction of the molding sheet 2. Marking can be performed at arbitrary positions on the molding sheet 2. The information on the marked defect positions can be used, for example, by cutting the molded sheet 2 into a plurality of sheets of sheet sized products, and then, for example, a process of dividing these sheets into regular products and defective products. Can be.

In addition, in the said embodiment, although the linear light source 4 was fixed and the molding sheet 2 was conveyed, the position where the image of the linear light source 4 in the molding sheet 2 is projected is changed so that it may change. Just do it. Therefore, the shaping sheet 2 may be fixed and the linear light source 4 may be moved, and both the molding sheet 2 and the linear light source 4 may be moved at different directions or at different speeds. In the case where the molded sheet 2 is fixed and the linear light source 4 is moved, it is preferable to move the imaging units 5 ₁ to 5 _n at the same speed in the same direction as the linear light source 4. Thereby, the some two-dimensional image data containing a linear light source image can be obtained. The method of fixing the molding sheet 2 and moving the linear light source 4 can avoid deformation of the linear light source image by pulling the molding sheet 2 with the conveying device 3, Since the length of the shaping | molding sheet 2 which can be reduced is limited to the length corresponding to the movable range of the linear light source 4, in order to test the long shaping | molding sheet 2 efficiently, like the said embodiment, the shaping sheet 2 ) Is preferably returned.

In the above embodiment, the line image analysis for defect portion (61 ₁ ~ 61 _n) and that the image analysis unit (62 ₁ ~ 62 _n) for defects, 2 obtained from the same image sensing unit (5 ₁ ~ 5 _n) D Although detection of defects on the basis of the image data, a line defect image analysis unit for the (61 ₁ ~ 61 _n) and the image analysis unit for the point defect (62 ₁ ~ 62 _n) is, based on the two-dimensional image data obtained from the different image pick-up portion You may detect a defect, respectively. Thus, the image pickup section (5 ₁ ~ 5 _n) image pickup condition (shaped sheet (2) away, or formed sheet (2) carrying direction and the image pickup angle, direction in which the forming of from a) to, conditions suitable for the defect to be detected in It becomes possible to make it. For the line defect image analysis unit (61 ₁ ~ 61 _n) image pickup section for picking up a two-dimensional image data that is used in, that the image analysis for the defect portion (62 ₁ ~ 62 _n) for imaging a two-dimensional image data that are used in Compared with an imaging part, it is preferable to make distance from a molding sheet far, and to narrow the angle which the conveyance direction and imaging direction of the molding sheet 2 make. As a result, both of the point defects and the line defects can be imaged under the optimal imaging conditions, so that both the point defects and the line defects can be detected with better accuracy.

Further, in the above embodiment, each of the image pickup section (5 ₁ to 5 _n) in the two-dimensional image analysis for a line defect image portion (61 ₁ to 61 _n) and that the image analysis for the defect portion (62 picked up by _one or 2 _n images captured by the imaging units 5 ₁ to 5 _n based on the relative positions between the two-dimensional images picked up by the imaging units 5 ₁ to 5 _n , but distributed and processed at 62 _n ). Synthesizes one full width image including the entire area in the width direction of the molding sheet 2 from the image forming unit, and generates a defect in one line defect image analysis unit and one point defect image analysis unit based on the full width image. You may detect. As a method of synthesizing one full width image from n two-dimensional images, for example, the method described in paragraph [0050] of Patent Document 1 can be used.

Next, a description will be given of the line fault detection algorithm and that the defect detection algorithm used in line image analysis for defect portion (61 ₁ ~ 61 _n) and the image analysis for the point defect portion (62 ₁ ~ 62 _n). As the line defect detection algorithm and the point defect detection algorithm, the following seven types of defect detection algorithms A to G can be used. In addition, in the following description, the luminance value (pixel value) of two-dimensional image data shall be represented by a natural number.

[Defect Detection Algorithm A]

The defect detection algorithm A will be described below based on FIG. 6. 6A is an example of multi-value two-dimensional image data (hereinafter referred to as original image data) generated by one of the imaging units 5 ₁ to 5 _n , and the upper side of the image is the downstream side in the conveying direction. The lower side of the image is the transport direction upstream side. In FIG. 6 (a), the band-shaped white region extending in the horizontal transverse direction at the center is a linear light source image, a dark region existing inside the linear light source image, and a small white region present in the vicinity of the linear light source image. It is a fault.

In this defect detection algorithm A, the following processing is performed on each of the plurality of original image data generated by the imaging units 5 ₁ to 5 _n .

First, the original image data is divided into data (luminance value (pixel value) and data indicating luminance position (pixel value) and position of one row along the longitudinal direction (the conveyance direction of the molding sheet 2); luminance profile; one-dimensional image data)). do.

Next, the data of each pixel column is subjected to a first edge determination process of searching for an edge from one end (the upper end of Fig. 6 (a)) to the other end (the lower end of Fig. 6 (a)) as follows. First, the second pixel from one end of the pixel column is the pixel of interest, and it is determined whether the luminance value of the pixel of interest is smaller than the luminance value of the adjacent pixel adjacent to one end of the pixel of interest by the threshold value T1 or more. In the case where it is determined that the luminance value of the pixel of interest is smaller than or equal to the threshold value T1 by the luminance value of the adjacent pixel (that is, La-Lb? T1 when the luminance value of the adjacent pixel is La and the luminance value of the pixel of interest is Lb), It is determined that the adjacent pixel is the first edge, the position of the first edge (position of the adjacent pixel) is recorded, and the data processing of the pixel string to be processed is terminated. Otherwise, the pixel of interest is one pixel facing the other end. In the determination, the determination is repeated until the luminance value of the pixel of interest is determined to be less than or equal to the threshold value T1 in the determination, and the luminance value of the pixel of interest is higher than the luminance value of the adjacent pixel in the determination. When it is determined that the abnormality is small, it is determined that the adjacent pixel is the first edge, the position of the first edge (position of the adjacent pixel) is recorded, and the data processing of the pixel column to be processed is terminated. The threshold value T1 is any natural number and may be the minimum unit of the luminance value. When the threshold value T1 is the minimum unit of the luminance value, the determination simply determines whether the luminance value of the pixel of interest is smaller than the luminance value of the adjacent pixel.

Next, the following 2nd edge determination process which searches for an edge toward the one end from the other end as follows is performed with respect to the data of each pixel column. First, the second pixel from the other end side is the pixel of interest, and it is determined whether the luminance value of the pixel of interest is greater than or equal to the threshold value T2 with respect to the pixel of interest than the luminance value of the adjacent pixel adjacent to the other end side. In the case where it is determined that the luminance value of the pixel of interest is greater than or equal to the threshold value T2 (ie, if the luminance value of the adjacent pixel is La and the luminance value of the pixel of interest is Lb, Lb-La ≥ T2), It is determined that the pixel of interest is the second edge, the position of the second edge (position of the main pixel) is recorded, and the data processing of the pixel string to be processed is finished. Otherwise, the pixel of interest is one pixel facing one end. In the determination, the determination is repeated until the luminance value of the pixel of interest is determined to be greater than or equal to the threshold value T2 in the determination, and the luminance value of the pixel of interest is higher than the luminance value of the adjacent pixel in the determination. When it is determined that the abnormality is large, it is determined that the pixel of interest is the second edge, the position of the second edge (position of the main pixel) is recorded, and the data processing of the pixel string to be processed is terminated. The threshold value T2 is any natural number and may be the minimum unit of the luminance value. When the threshold value T2 is the minimum unit of the luminance value, the determination simply determines whether the luminance value of the pixel of interest is greater than or equal to the threshold value than the luminance value of the adjacent pixel.

An example of the first edge detected by these first edge determination processing is shown as "Δ" in Fig. 6A, and an example of the second edge detected by the second edge determination processing is shown in Fig. 6A. It represents with "(circle)". As can be seen from Fig. 6 (a), since there is no edge other than the edge of the linear light source image in the region without a defect, the first edge and the second edge are the other end edges in the linear light source image ( In the example of Fig. 6 (a), the bottom edge) coincides with each other. On the other hand, as can be seen from Fig. 6 (a), in the defective areas (white area and black area), at least one of the first edge and the second edge coincides with the edge of the defective area, and is applied to the linear light source image. Since the edge search start side deviates from the other end side edge, the first edge and the second edge are separated from each other.

Then, for the data of each pixel column, the distance (pixel number) from the first edge to the second edge is obtained as the edge-to-edge distance. Fig. 6 (b) shows a profile in which the obtained edge-to-edge distance is plotted against the position of the pixel column (coordinate in the horizontal direction). If there is a pixel string whose threshold distance is greater than or equal to the threshold value T3, it is determined that there is a defect. The threshold value T3 is any natural number and may be one pixel. In the case where the threshold value T3 is 1 pixel, the pixel string whose non-edge distance is not 0 is determined as the pixel string containing the defect. What is necessary is just to determine the threshold value T3 suitably according to the allowable defect size, but when multi-dimensional two-dimensional image data is two-dimensional image data of 256 gray levels (luminance values 0-255; 8 bits), it is preferable that it is 3, for example. Do.

In addition, in the example of FIG. 6, the upper end is once (the first edge search start side), but which end of the pixel string is arbitrary, and the lower end may be one end. In that case, in the region without a defect, the first edge and the second edge coincide with the upper edge in the linear light source image.

This defect detection algorithm A can detect various point defects with some certainty. However, the detection certainty of minute point defects, such as bubbles and fisheye, is not very high. On the other hand, this defect detection algorithm A is not suitable for detection of line defects. This defect detection algorithm A is called "edge profile method 1" hereafter.

[Defect Detection Algorithm B]

The defect detection algorithm B fits the edge of the linear light source image in the two-dimensional image data into a function curve, and the defect is a point where the distance between the edge of the linear light source image and the function curve is equal to or larger than the threshold T5 (first threshold). To detect.

The defect detection algorithm B will be described below based on FIG. 7. Figure 7 (a) is the imaging unit (5 ₁ ~ 5 _n) as an example of the original image data produced by one, and the upper side of the image carrying direction downstream side, the lower side of the image carrying direction upstream side. In Fig.7 (a), the strip | belt-shaped white area | region extending in the horizontal direction of the center is a linear light source image, and the locally deformed part (not smooth place) of the lower edge of a linear light source image is a defect.

In this defect detection algorithm B, the following processing is performed on each of the plurality of original image data generated by the imaging units 5 ₁ to 5 _n .

First, at least one of the edges of the linear light source image is obtained from the original image data. An example of the edge of the obtained linear light source image is shown by "(circle)" in FIG. In the example of FIG. 7, although the lower edge of the linear light source image is calculated | required, the upper edge of the linear light source image may be calculated | required, and both the upper edge and the lower edge of the linear light source image may be calculated | required.

As a method for obtaining the edge of the linear light source image, a method of extracting an edge using a known edge extraction filter (e.g., a Sobel filter) and making a strong edge an edge of the linear light source image, and using two-dimensional image data After dividing into data of pixel columns row by row, a strong edge is obtained from the data of each pixel column to be an edge of the linear light source image, and a method of obtaining an area of the linear light source image by the method described in [0057] (Patent Value 2 and Labeling and extracting an area larger than a predetermined value as an area of the linear light source image in the labeled area). Here, as an example, a method of dividing the original image data into data of pixel columns of one row and then obtaining a strong edge from the data of each pixel column to form an edge of the linear light source image will be described. First, original image data is divided into data of pixel columns of one row along the vertical direction (the conveying direction of the molding sheet 2). Subsequently, processing for finding the edge toward the other end (lower end of FIG. 7A) from one end (the upper end of FIG. 7A) to the data of each pixel column is as follows. First, the second pixel from the one end side is the pixel of interest, and is the luminance value of the pixel of interest less than or equal to the threshold value T4 (T4 is a natural number) than the luminance value of the adjacent pixel adjacent to the end of the pixel of interest (i.e., When the luminance value of the adjacent pixel is La and the luminance value of the pixel of interest is Lb, La-Lb? T4) is determined. In order to detect only a strong edge, the threshold value T4 at this time is made into a comparatively large value. When it is determined that the luminance value of the pixel of interest is smaller than the luminance value of the adjacent pixel by the threshold value T4 or more, it is determined that the adjacent pixel is an edge of the linear light source image, and the position (position of the adjacent pixel) of the linear light source image is recorded. When the data processing of the pixel column to be processed is finished, otherwise, the pixel of interest is moved by one pixel toward the other end, in the determination, the luminance value of the pixel of interest is greater than or equal to the threshold T4 in the determination. The above judgment is repeated until it is determined to be small, and when it is determined that the luminance value of the pixel of interest is smaller than or equal to the luminance value of the adjacent pixel by the threshold value T4 or more, it is determined that the adjacent pixel is the edge of the linear light source image, and the edge of the linear light source image The position (the position of the adjacent pixel) of is recorded, and the data processing of the pixel column to be processed ends.

Next, the row of edges of the obtained linear light source image is fitted into a smooth curve represented by a function (fitting with a function curve) to obtain a fitting curve (function curve). Examples of functions used for fitting include n-th order functions (n is 2 or more), Gaussian functions, Lorentz functions, voigt functions, combinations of these functions, and the like. For example, a quadratic function is preferred. Moreover, as the evaluation method of the fitting used at the time of fitting, the least square method can be used, for example.

Next, the two-dimensional image data is divided into data of one row of pixel columns along the vertical direction (the conveying direction of the molding sheet 2), and the distance from the fitting curve to the edge of the linear light source image with respect to the data of each pixel column. (Pixel number) is obtained as a fitting diagram. A profile obtained by plotting the obtained fitting degree with respect to the position of the pixel column (coordinates in the horizontal direction (the direction perpendicular to the conveying direction of the molding sheet 2 and perpendicular to the thickness direction of the molding sheet 2)) is illustrated in FIG. 7B. ). And if this fitting degree exists in the pixel column more than threshold value T5, it is determined that there exists a defect in the position of the edge of the linear light source image in the pixel column. Thereby, presence or absence of a defect can be determined and a defect position can also be calculated | required. As described above, it is possible to detect a line defect that appears as a local deformation of the edge of the linear light source image (deformation of a fine linear light source image near the edge). In addition, the threshold value T5 used for the said determination is arbitrary natural number, and 1 pixel may be sufficient as it. In the case where the threshold value T5 is 1 pixel, if there is a pixel string whose fitting degree is not 0, it is determined that there is a defect. What is necessary is just to determine the threshold value T5 suitably according to the allowable defect size, but when multi-value two-dimensional image data is two-dimensional image data of 256 gray levels, it is preferable that it is four.

In addition, in this defect detection algorithm B, a defect position may be calculated | required in addition to determining the presence or absence of a defect. In that case, the fitting degree may extract a pixel column having a threshold value T5 or more, and determine the position of the pixel between the edge of the linear light source image and the fitting curve in the extracted pixel column as the defect position.

This defect detection algorithm B can detect various line defects with high certainty. On the other hand, this defect detection algorithm B is not suitable for detection of a point defect. This defect detection algorithm B is called "edge profile method 2" hereafter.

[Defect Detection Algorithm C]

The defect detection algorithm C smoothes the two-dimensional image data, obtains the difference between the smoothed two-dimensional image data and the original two-dimensional image data as differential image data, and the luminance value in the differential image data is equal to the threshold value T6B (first 4th threshold value: T6B is above a certain positive number, and the point where luminance value is below threshold value T6D (5th threshold value; T6D is any positive number less than T6B) is detected as a defect.

The defect detection algorithm C will be described in more detail below. This defect detection algorithm C compares the width of the linear light source image with the change in contrast in the horizontal direction (the direction perpendicular to the conveying direction of the molding sheet 2 and perpendicular to the thickness direction of the molding sheet 2). The change in contrast in the direction uses a high spatial frequency to extract a high frequency component of the original image data, and detects as a defect a portion whose luminance value in the high frequency component is greater than or equal to the threshold T6B or less than or equal to the threshold T6D.

(1) First, the original image data is smoothed (smooth) in the horizontal direction using a horizontal smoothing filter (matrix) in one row n columns (n is an integer of 3 or more) to obtain smoothed image data. As a result, the high frequency region of the horizontal luminance change of the original image data is eliminated, and only the low frequency component remains with respect to the horizontal luminance change (the low frequency component of the horizontal luminance change and the vertical luminance change remain). As the horizontal smoothing filter, a weighted averaging filter such as a Gaussian filter, an averaging filter, or the like can be used. In addition, it is preferable that n is three.

(2) Next, the smoothed image data is subtracted from the original image data (the luminance value of each pixel is subtracted). As a result, only the high frequency component of the horizontal luminance change in the original image data remains.

(3) Next, smoothing using a 3x3 pixel smoothing filter (operator) is performed on the image data obtained by subtraction. This smoothing removes the noise and leaves high frequency components other than the noise. As said smoothing filter, it is preferable to use what smooths the edges preserve | saved, such as a bilateral filter and a median filter.

(4) Next, from the original image data, the upper edge (carrying direction downstream edge) and the lower edge (carrying direction upstream edge) in the linear light source image are obtained. Since the method of obtaining the edge of the linear light source image is the same as that described with respect to the defect detection algorithm B, the description is omitted. Next, the conveyance direction of the shaping | molding sheet 2 in original image data is made into the X-axis, the minimum value Min is calculated | required among the X coordinate values of all the pixels which comprise an upper edge, and the X coordinate value of all the pixels which comprise a lower edge. Find the maximum value Max. Then, the value obtained by subtracting the maximum value Max from the minimum value Min is regarded as the width W of the linear light source image, and the area where the X coordinate value is widened outward by the width W from the maximum value Max to the minimum value Min is defined as the inspection area. . That is, an area whose X coordinate value is greater than or equal to Max-(Min-Max) and less than or equal to Min + (Min-Max) is defined as the inspection area. This process is for narrowing the inspection subject region only to the linear light source image and its vicinity region. The inspection area is widened outward from the area from the maximum value Max to the minimum value Min in order to make the inspection area including some deformation of the linear light source image.

(5) Next, from the luminance value of the pixel in the inspection area in the image data (high frequency components other than noise) after smoothing of the above (3), the bright side (high luminance) is expressed by the following equation. Threshold T6B on the dark side and the threshold T6D on the dark side (lower brightness side).

T6B = (average luminance value in inspection region) + (standard deviation of luminance value in inspection region) × k

T6D = (average luminance value in inspection area)-(standard deviation of luminance value in inspection area) × k

(k represents a positive parameter)

In addition, what is necessary is just to determine the value of k suitably according to the allowable defect size, for example, 1.5, 3, 4.5 etc., for example.

(6) Next, a process for determining whether the luminance value is equal to or larger than the threshold T6B or equal to or lower than the threshold T6D for all the pixels in the inspection area in the image data after the smoothing of the above (3). Processing), and the pixel below threshold value T6B or below threshold value T6D is extracted as a defect site | part. Thereby, presence or absence of a defect can be determined and a defect position can also be calculated | required.

On the contrary, if the noise included in the original image data generated by the imaging section (5 ₁ ~ 5 _n) is small, the smoothing process may be omitted in (3). In addition, when it is not necessary to narrow an inspection object area | region to only a linear light source image and its vicinity area | region, the process of defining the inspection area | region of (4) is abbreviate | omitted, and the process of (5) (6) is performed with respect to the whole image data. You may also

This defect detection algorithm C can detect all the point defects including micro defects, such as a bubble and fisheye, with high certainty. On the other hand, this defect detection algorithm C is not suitable for the detection of line defects. However, the processing time is shorter than that of the defect detection algorithm C (the processing time of the defect detection algorithm C is, for example, about 40 ms per frame). This defect detection algorithm C is hereinafter referred to as "high pass filter method".

[Defect Detection Algorithm D]

The defect detection algorithm D will be described below based on FIGS. 8 and 9. 8 is an example of original image data generated by one of the imaging units 5 ₁ to 5 _n . The upper side of the image is the transport direction downstream side, and the lower side of the image is the transport direction upstream side. In FIG. 8, the strip | belt-shaped white area | region extending in the horizontal direction of the center is a linear light source image, the dark area which exists inside the linear light source image, and the small white area which exists in the vicinity of the linear light source image are defects. In FIG. 8, the curves above and below the linear light source image represent the upper limit and the lower limit of the inspection subject region.

In this defect detection algorithm D, the following processing is performed on each of a plurality of original image data generated by the imaging units 5 ₁ to 5 _n .

First, the original image data is divided into data of pixel columns of one row along the vertical direction (the conveying direction of the molding sheet 2), and a plot column showing a change in the luminance value depending on the position in each pixel column is vertical. Obtained as the directional luminance profile. 9 shows an example of the obtained vertical luminance profile. This example is a vertical luminance profile with respect to the pixel column at the position indicated by the arrow in FIG. 8, wherein y represents a downward direction (the direction indicated by the arrow in FIG. 8; a direction opposite to the conveying direction of the molding sheet 2). Y coordinate when

Next, the depth (see FIG. 8) of the valley portion is obtained for the vertical luminance profile of each pixel column. That is, first, all the maximum and minimum points are obtained for the vertical luminance profile of each pixel column, and for all the obtained minimum points, the luminance values (minimum values) of the minimum points and the luminance values of the maximum points closest to the minimum points are obtained. The difference of (maximum value) is calculated | required as depth of a valley part. If the calculated depth of the valley portion is equal to or greater than the threshold value T7 (T7 is positive), it is determined that the valley portion is defective. What is necessary is just to determine the threshold value T7 suitably according to the allowable defect size, but when multi-value two-dimensional image data is two-dimensional image data of 256 gray levels, it is preferable that it is 0.25x255, for example.

This defect detection algorithm D has a relatively short processing time. This defect detection algorithm D can detect various point defects with a certain degree of certainty. In particular, it is suitable for the detection of point defects causing local contrast inversion near the edge of the linear light source image. However, since the image including the point defect and its vicinity needs to be high contrast, the detection certainty of minute point defects such as bubbles, fish eyes, tire traces, etc. is not very high. On the other hand, this defect detection algorithm D is not suitable for detecting line defects. This defect detection algorithm D is called "peak method" hereafter.

[Defect Detection Algorithm E]

The defect detection algorithm E represents the change in luminance depending on the position along the straight line in the two-dimensional image data as the luminance profile, and assumes the quality of moving the plot group of the luminance profile so that the movement time between the plots becomes constant. Predicting the luminance value of the plot of interest from the velocity vector of the material point between the two plots immediately before the plot of interest and the acceleration vector of the material point between the three plots immediately before the plot of interest, The point where the difference in the luminance value of is greater than or equal to the threshold value T8 (third threshold value; T8 is a natural number) is detected as a defect.

The defect detection algorithm E will be described below based on FIGS. 10 and 11. This defect detection algorithm E improves the accuracy of the peak method and detects a defect based on the difference between the measured value and the predicted value instead of the depth of the valley.

In this defect detection algorithm E, the following processing is performed on each of the plurality of original image data generated by the imaging units 5 ₁ to 5 _n .

First, in the same manner as in the peak method, the vertical luminance profile of each pixel column is obtained. An example of the obtained vertical luminance profile is shown in FIG. 10 with the luminance value as the x axis. The circled portion in the vertical luminance profile is a profile corresponding to a defect to be detected by the defect detection algorithm E. FIG.

Next, with respect to the vertical luminance profile of each pixel column, a quality point moving from one end of the plot column to the other end is assumed to be constant regardless of the distance between the plots. And as shown in FIG. 11, it is said that the said quality point moves from the plot c to the plot b adjacent to it, from the plot b to the plot a adjacent to it, and from the plot a to the plot d adjacent to it. In addition, it is assumed that the plot d is a plot corresponding to the pixel of interest.

And the velocity vector and acceleration vector of the quality point in three plots a-c which the quality point passed immediately before the plot d are calculated | required. That is, based on the movement time and the coordinates (x coordinates, y coordinates) of two plots a and b that have passed through the quality point immediately before plot d, the velocity of the quality point in the interval from plot b to plot a Find the vector. Also, based on the movement time and the coordinates (x coordinate, y coordinate) of the plots b and c which have passed the quality point immediately before the plot d, the velocity vector of the quality point in the interval from the plot c to the plot b is obtained. In the interval from plot c to plot a based on the velocity vector of the quality point in the interval from plot b to plot a and the velocity vector of the quality point in the interval from plot c to plot b. Obtain the acceleration vector of the material point of. And the coordinate (position) of the plot d is predicted from the velocity vector of the said quality point in the area | region from the plot b to the plot a, and the acceleration vector of the said quality point in the area | region from the plot c to the plot a.

Thus, the difference between the predicted x coordinate (luminance value) of the plot d and the actual (actually measured) x coordinate (luminance value) of the plot d is obtained. If these differences are equal to or greater than the threshold value T8, the pixel corresponding to the plot d is obtained. Extract as a defect site. Thereby, presence or absence of a defect can be determined and a defect position can also be calculated | required. What is necessary is just to determine the threshold value T8 suitably according to the allowable defect size, but when multi-value two-dimensional image data is 256-dimensional two-dimensional image data, it is preferable that it is 20, for example.

This defect detection algorithm E can detect various point defects with high certainty. This defect detection algorithm E is called "peak method 2" hereafter.

Fault Detection Algorithm F

The defect detection algorithm F will be described below based on FIG. 12. 12 (a) is an example of original image data generated by one of the imaging units 5 ₁ to 5 _n , _wherein the upper side of the image is the downstream side in the conveying direction, and the lower side of the image is the conveyance direction upstream. In Fig. 12 (a), the band-shaped white region extending in the horizontal direction at the center is a linear light source image, and a locally deformed portion (where the inclination with respect to the horizontal line is large) of the lower edge of the linear light source image is It is a fault.

In this defect detection algorithm F, the following processing is performed on each of the plurality of original image data generated by the imaging units 5 ₁ to 5 _n .

First, at least one of the edges of the linear light source image is obtained from the original image data. An example of the edge of the obtained linear light source image is shown by "(circle)" in FIG. In the example of FIG. 12, the lower edge of the linear light source image is obtained, but the upper edge of the linear light source image may be obtained, or both of the upper edge and the lower edge of the linear light source image may be obtained. Since the method of obtaining the edge of the linear light source image is the same as that described with respect to the defect detection algorithm B, the description is omitted.

Next, the horizontal direction is the x-axis and the vertical direction is the y-axis, and the second derivative is obtained by second-order differentiating the curve (edge profile) y = f (x) of the edge of the linear light source image. An example of the obtained second derivative profile is shown in FIG. 12 (b).

Then, for each pixel at the edge of the linear light source image, it is determined whether the second derivative is equal to or greater than the threshold T9 (T9 is positive), and the pixel (where the high frequency) whose second derivative is equal to or greater than the threshold T9 is determined as a defect site. do. As a result, the presence or absence of a defect can be determined, and a defect position can also be obtained. What is necessary is just to determine the threshold value T9 suitably according to the allowable defect size.

This defect detection algorithm F is suitable for the detection of the line defect shown by local curvature of the edge of the linear light source image. The defect detection algorithm F is not so high in defect detection capability. This defect detection algorithm F is called "edge curve method 1" hereafter.

[Defect detection algorithm G]

The defect detection algorithm G calculates a curvature in the vicinity region (range of 2N + 1 pixel) of each pixel with respect to the edge of the linear light source image in the two-dimensional image data, and the curvature is the threshold value T10 (second). Threshold; T10 is a positive point or more).

The defect detection algorithm G will be described below based on FIGS. 13A to 13C.

In this defect detection algorithm G, the following processing is performed on each of the plurality of original image data generated by the imaging units 5 ₁ to 5 _n .

First, at least one of the edges of the linear light source image is obtained from the original image data. Examples of the edges of the obtained linear light source image are shown in Figs. 13A to 13C. Since the method of obtaining the edge of the linear light source image is the same as that described with respect to the defect detection algorithm B, the description is omitted.

Next, the curvature at each point (each pixel) is calculated | required about the curve of the edge of a linear light source image. The method of calculating the curvature is not particularly limited and may be a method of calculating using a mathematically determined formula. However, since the processing time is long in such a method, the curvature is preferably approximated by the following method.

(1) Consists of pixels of interest (white dots in FIGS. 13A to 13C) and pixels of interest (left and right) N pixels (left and right) with respect to the pixels of interest (black dots in FIGS. 13A to 13C) on the edges. The range (the range of 2N + 1 pixels in the vicinity of the main pixel) is taken as the calculation target range (N is a natural number). What is necessary is just to determine N suitably according to the allowable defect size, For example, it is preferable that it is 30. In the example of FIGS. 13A to 13C, N is three.

(2) Next, pixels at both ends of the calculation target range are connected in a straight line.

(3) The predicted luminance value is obtained from the straight line over all the pixels in the calculation target range, the increment of the actual luminance value (the luminance value on the edge curve) with respect to the predicted luminance value is obtained, and the increment or the absolute value of the increment is calculated. Is integrated. By the integrated value obtained here, the curvature in the range of 2N + 1 pixel vicinity of a pixel of interest can fully be approximated. (A curvature value nearly equivalent to the curvature calculated using the mathematical formula determined can be obtained.). Here, in the configuration using the incremental integration value, when a change in the minute luminance value such as going up or down a straight line within the calculation target range occurs as shown in FIG. 13C, these changes are canceled and ignored, and an approximation value of curvature is obtained. Will be obtained. On the other hand, in the configuration using the integrated value of the absolute value of the increment, even when such a change occurs, an approximation of the curvature is obtained including such a change. As shown in Fig. 13C, when a small change in luminance value, such as going up or down a straight line within the calculation target range, is to be detected as a defect, an integrated value of the absolute value of the increment may be used. On the contrary, when such a change is allowed and it is not intended to be detected as a defect, it is sufficient to employ a configuration in which an incremental integration value is used.

(4) The integration value is calculated for all the pixels on the edge while moving the pixel of interest from the edge of the edge in the linear light source image by one pixel. Thereby, the profile (curvature profile) of the approximation of curvature is produced.

Next, for each pixel at the edge of the linear light source image in the curvature profile, it is determined whether the curvature is greater than or equal to the threshold value T10, and the pixel whose obtained curvature is greater than or equal to the threshold value T10 is determined as a defect site (or defect candidate). . As a result, the presence or absence of a defect can be determined, and a defect position can also be obtained. Since the molded sheet 2 is somewhat curved, the edge of the linear light source image is somewhat bent, so that if the curvature of the edge of the linear light source image is to some extent, it should be allowed as not a defect. Therefore, the threshold value T10 should be relatively large. What is necessary is just to determine the threshold value T10 suitably according to the allowable defect size, but when multi-value two-dimensional image data is 256-dimensional two-dimensional image data, it is preferable that it is 110, for example.

This defect detection algorithm G can detect various line defects with high certainty. This defect detection algorithm G is called "edge curve method 2" hereafter.

In this embodiment, the combination of the line defect detection algorithm and the point defect detection algorithm used by the line defect image analyzers 61 ₁ to 61 _n and the point defect image analyzers 62 ₁ to 62 _n , respectively, Any of the followings.

(A) line image analysis unit for the defect (61 ₁ ~ 61 _n) line fault detection algorithm used by the edge, and the profile method 2 or edge curves Method 2, used in image analysis for the point defect portion (62 ₁ ~ 62 _n) The point defect detection algorithm is a high pass filter method or a peak method 2.

(B) line image analysis unit for the defect (61 ₁ ~ 61 _n) line fault detection algorithm used by the edge, and the profile method 2 or edge curves Method 2, used in image analysis for the point defect portion (62 ₁ ~ 62 _n) The point defect detection algorithm is a defect detection algorithm other than the high pass filter method or the peak method 2.

(C) line defect image analysis unit for (61 ₁ to 61 _n) and the line fault detection algorithm the fault detection algorithm other than the edge profiles method 2 or edge curve method 2 used in, that the image analysis unit (62 for the defective _one to The point defect detection algorithm used in 62 _n ) is the high pass filter method or the peak method 2.

The combination of (A) is the most preferable among the combination of (A)-(C). In the case of the combination of (A), both a line defect and a point defect can be detected reliably. In the case of the combination of (B), a line defect can be detected reliably. In the case of the combination of (C), a point defect can be detected reliably.

[Embodiment 2]

Another embodiment of the present invention will be described with reference to FIG. 14 as follows. In addition, for convenience of description, the same code | symbol is attached | subjected to the member which has the same function as each member shown in the said Embodiment 1, and the description is abbreviate | omitted.

The defect inspection apparatus which concerns on this embodiment is the same as the defect inspection apparatus 1 which concerns on Embodiment 1 except having provided the analyzer 6A shown in FIG. 14 instead of the analyzer 6 shown in FIG. It has a configuration. As shown in FIG. 14, the analyzer 6A omits the point defect image analyzers 62 ₁ to 62 _n from the analyzer 6 shown in FIG. 1.

In this embodiment, the line fault detection algorithm used by the image analysis unit for the line defect (61 ₁ ~ 61 _n) is an edge profile or edge curve method 2 method 2. In this embodiment, a line defect can be detected reliably.

In addition, although the defect inspection apparatus of this embodiment may be used independently, it is preferable to use in combination with the defect inspection apparatus which can detect a point defect. As a result, not only line defects but also point defects can be inspected. Although the well-known various defect inspection apparatus may be sufficient as the defect inspection apparatus which can detect the point defect which can be combined with the defect inspection apparatus of this embodiment, it is preferable that it is the defect inspection apparatus of Embodiment 3 mentioned later. Thereby, both a line defect and a point defect can be reliably examined.

[Embodiment 3]

Another embodiment of the present invention will be described with reference to FIG. 15 as follows. In addition, for convenience of description, the same code | symbol is attached | subjected to the member which has the same function as each member shown in the said Embodiment 1, and the description is abbreviate | omitted.

The defect inspection apparatus which concerns on this embodiment is the same as the defect inspection apparatus 1 which concerns on Embodiment 1 except having provided the analyzer 6B shown in FIG. 15 instead of the analyzer 6 shown in FIG. It has a configuration. Analyzing apparatus (6B) is obtained by omitted, analysis device 6 line image analysis unit (61 ₁ ~ 61 _n) for defect from showing in Figure 1, as shown in Fig.

In this embodiment, the point defect detection algorithm used by the point defect image analysis parts 62 ₁ to 62 _n is a high pass filter method or a peak method 2. As shown in FIG. In this embodiment, a point defect can be detected reliably.

In addition, although the defect inspection apparatus of this embodiment may be used independently, it is preferable to use in combination with the defect inspection apparatus which can detect a line defect. As a result, not only the point defect but also the line defect can be inspected. Although the well-known various defect inspection apparatus may be sufficient as the defect inspection apparatus which can detect the line defect which can be combined with the defect inspection apparatus of this embodiment, it is preferable that it is the defect inspection apparatus of Embodiment 2. Thereby, both a line defect and a point defect can be reliably examined.

Experimental Example

Next, in order to confirm the effect of this invention, the experiment result performed using 14 types of experimental defect inspection apparatuses similar to the defect inspection apparatus which concerns on the said embodiment is shown.

Claim 1-7 Laboratory defect inspection apparatus, omitting the imaging section (5 ₂ ~ 5 _n) from the defect inspection apparatus according to the third embodiment, and a transfer device (3), instead of the conveyor roll, the forming sheet (2) The conveyor is used to put the product on the surface and convey it. The 1st-7th experimental defect inspection apparatus is for detecting a point defect from the sample containing a point defect.

The first experimental defect inspection apparatus is an edge provided with a profile law point defect image analysis unit for using a first (62 ₁ ~ 62 _n), and the second experimental defect inspection apparatus is the image for the point defects with edge profiles method 2 analyzer ( 62 ₁ to 62 _n ), the third experimental defect inspection apparatus includes image analysis units 62 ₁ to 62 _n for point defects using a high pass filter method, and the fourth experimental defect inspection apparatus uses a peak method. comprising an image analysis unit for using point defect (62 ₁ ~ 62 _n), and a fifth experimental defect inspection apparatus comprises having the peak method point defect image analysis unit for using _{2 (62 1 ~ 62 n)} , and 6 experiments having a defect inspection apparatus is the edge curve method point defect image analysis unit for using ₁ (62 1 ~ 62 _n), and the seventh experimental defect inspection apparatus using the edge curve method 2 (using the integrated value of the increment) for point defect image analysis unit (62 ₁ ~ 62 _n) And a.

Claim 8-14 guinea defect inspection apparatus, omitting the imaging section (5 ₂ ~ 5 _n) from the defect inspection apparatus according to the second embodiment, and a transfer device (3), instead of the conveyor roll, the forming sheet (2) The conveyor is used to put the product on the surface and convey it. The 8th-14th experimental defect inspection apparatus is for detecting a line defect from the sample containing a line defect.

The eighth experimental defect inspection apparatus includes line analysis image analysis units 61 ₁ to 61 _n using the edge profile method 1, and the ninth experimental defect inspection apparatus includes an image analysis unit for line defects using the edge profile method 2 ( 61 ₁ ~ 61 _n) provided, and a tenth experiment defect inspection apparatus includes an image for line fault with the high-pass filter process analysis unit (61 ₁ ~ 61 _n), and the 11th experimental defect inspection apparatus of the peak method having a with line defect image analysis unit (61 ₁ ~ 61 _n) for the twelfth experimental defect inspection apparatus includes an image analysis for a line fault with the peak method 2 parts (61 ₁ ~ 61 _n), and the 13th experimental defect inspection apparatus includes an edge image analysis for a line fault with curves how part 1 (61 ₁ ~ 61 _n) and 14 guinea defect inspection apparatus using the edge curve method 2 (using the integrated value of the increment) Image analysis unit for line defects 61 ₁ to 61 _n ).

In the present experimental example, as the molding sheet 2, samples of ten kinds of polarizing films containing different kinds of point defects and samples of six kinds of polarizing films containing different kinds of line defects were used. The ten kinds of samples including the point defects include Sample 01 containing bubbles, Sample 02 including fisheye, Sample 03 including first foreign matter, Sample 04 containing second foreign matter different from the first foreign matter, and Sample 06 comprising one tire trail, Sample 07 comprising a second tire trail different from the first tire trail, Sample 08 comprising the first scar, Sample 09 comprising the second scar different from the first scar, and the first Sample 11 comprising a scratch, and sample 12 comprising a second scratch different from the first scratch. The six types of samples including the line defects are along the conveying direction of the sample 10 and the forming sheet 2 including the sample 10 including the Kunic (the line defect), the second Kunic (the line defect) different from the first Kunic. Sample 51 containing streaks, Sample 52 containing strong streaks orthogonal to the conveying direction of the molding sheet 2, Sample 53 containing weak streaks orthogonal to the conveying direction of the molding sheet 2, and Molded sheet ( It is the sample 53 containing the stripe of a diagonal direction with respect to the conveyance direction of 2).

Moreover, in the defect inspection apparatus for 1st-14th experiments, the imaging part 5 ₁ For example, a progressive scan area sensor using a CCD element capable of imaging a two-dimensional image and generating two-dimensional image data of 512 pixels by 480 pixels in length by 256 gray scales was used. In the 1-14 test defect inspection apparatus, the high frequency fluorescent lamp which attached the sharp edge hood (hood which sharpens the edge of a linear light source image) was used as the linear light source 4. In the defect inspection apparatus for experiments 1-14, the conveyance speed of the shaping | molding sheet 2 by a conveyor was set to 20 mm / sec (= 1.2 m / min). In addition, the position at a distance of 145 mm from the end of the conveyor (front end in FIG. 2) is the imaging unit 5 ₁ . It was made to become the center of the imaging area by. Moreover, a metal chuck was used to measure this 145 mm so that the distance from a gold chuck to a defect might be 55 mm.

In the 1-7 test defect inspection apparatus for detecting a point defect, the imaging area | region (field of vision) on the molding sheet 2 is horizontal (orthogonal to the conveyance direction of the molding sheet 2, and also the thickness of the molding sheet 2). Direction orthogonal to the direction) 51.2 mm x vertical (conveying direction of the molding sheet 2) the imaging unit 5 ₁ so as to be 48 mm The position and angle of were adjusted. However, among the 512 pixels x 480 pixels of the progressive scan area sensor, 512 pixels listed in the horizontal direction at the position (center position in the vertical direction) of the 240th pixel from the top are horizontal 51.2 on the surface of the molding sheet 2. Imaging unit 5 ₁ to photograph an area of ㎜ The position and angle of were adjusted. That is, the imaging section 5 ₁ in the molding sheet 2 Imaging part (5 ₁ ) so that distance to the head is 190 mm Adjust the position of the, and the imaging unit (5 ₁ ) Direction of imaging (imaging part (5 ₁ ) From the center of the condensing lens of the imaging unit (5 ₁ ) The imaging unit ₍₅₁₎ such that the center of the area to be imaged toward the direction) and formed sheet (2) surface forms an angle of 40 degrees by The angle of was adjusted. In this case, the imaging unit 5 ₁ The resolution of is 100 µm / pixel. In the first to seventh experimental defect inspection apparatuses for detecting point defects, a C-mount having a focal length of 25 mm, a minimum F value of 1.4, and a lens tip work distance of 270 mm as a progressive scan area sensor. The aperture was adjusted to about 11 by using a lens having a lens mounted on the progressive scan area sensor main body.

Moreover, in the 1-7 test defect inspection apparatus for detecting a point defect, the longitudinal direction of the linear light source 4 is orthogonal to the conveyance direction of the shaping | molding sheet 2, and the linear light source 4 in the shaping sheet 2 is carried out. The linear light source is 240 mm, and the straight line connecting the center of the imaging area on the molded sheet 2 and the center of the linear light source 4 forms an angle of 37 degrees with respect to the surface of the molded sheet 2. (4) was disposed.

In the 8th-14th experimental defect inspection apparatus for detecting a line defect, the imaging part 5 ₁ so that the imaging area | region on the molding sheet 2 may be 204.8 mm x 192 mm in width. The position and angle of were adjusted. However, among the 512 pixels x 480 pixels of the progressive scan area sensor, 512 pixels listed in the horizontal direction at the position (center position in the vertical direction) of the 240th pixel from the top are horizontal 204.8 on the surface of the molding sheet 2. Imaging unit 5 ₁ to photograph an area of ㎜ The position and angle of were adjusted. That is, the imaging section 5 ₁ in the molding sheet 2 Imaging unit (5 ₁ ) so that the distance to is 400 mm Adjust the position of the, and the imaging unit (5 ₁ ) The imaging section 5 ₁ so that the angle formed between the imaging direction of the surface and the surface of the molded sheet 2 is 15 degrees The angle of was adjusted. In this case, the imaging unit 5 ₁ The resolution of is 200 µm / pixel. Moreover, in the 8th-14th experimental defect inspection apparatus for detecting a line defect, as a progressive scan area sensor, C-mount whose focal length is 25 mm, minimum F value is 1.4, and lens tip work distance is 490 mm. The diaphragm was adjusted to about 5.6-8 using what attached the lens of to the progressive scan area sensor main body.

Moreover, in the 8th-14th experimental defect inspection apparatus for detecting a line defect, the linear light source 4 is arrange | positioned so that the longitudinal direction of the linear light source 4 may make an angle of 25 degree with respect to the conveyance direction of the shaping | molding sheet 2. And the work distance of the linear light source 4 was set to 900 mm.

Here, the parameter of the defect detection algorithm was set so that the point defect of 0.5 mm in diameter could be detected reliably. In the defect inspection apparatuses for the first and eighth experiments, the threshold value T3 in the edge profile method 1 was set to three. In the 2nd 9th defect inspection apparatus for experiments, the threshold value T5 in the edge profile method 2 was set to four. In the 3rd and 10th test defect inspection apparatuses, k in the edge profile method 2 was set to 4.5, and the horizontal smoothing filter used for the edge profile method 2 was set as the smoothing filter of 1 row and 3 columns. In the 4th and 11th experimental defect inspection apparatuses, the threshold value T7 in the peak method was set to 25% (255 * 0.25) of the maximum luminance value. In the 5th and 12th experimental defect inspection apparatuses, the threshold value T8 in the peak method 2 was set to 20. In the 6th and 13th experimental defect inspection apparatuses, the distance which is a parameter used in the edge curve method 1 was set to 15, and k was set to 5. In the seventh and fourteenth defect inspection apparatuses for experiments, the calculation target range is set to a range of 30 pixels before and after the pixel of interest (i.e., N is 15) using the method of calculating the curvature approximately as described above as the edge curve method 2. And threshold value T10 was set to 110.

Then, it is checked whether the point defect can be detected from ten kinds of samples including the point defect using the first to seventh test defect inspection apparatuses, and the line defects are determined using the eighth to fourteenth defect inspection apparatuses. It was examined whether line defects could be detected from six kinds of samples included. The obtained results are shown in Table 1.

In the first to seventh defect inspection apparatus for experiment, the total number of frames (total number of captured images) of the moving image of the molded sheet 2 to be imaged is 150, but one defect detection target point or point defect is detected by the imaging unit 5 ₁ . Imaging section of the imaging area (5 ₁ ) 30 frames of moving images are captured by the image capturing unit 5 ₁ within the period until the image exits from the image capturing region. That is, if a defect is always visible, the imaging unit 5 ₁ Of the two-dimensional image data picked up by the above, the defect image is included in 30 pieces (30 frames) of the two-dimensional image data picked up within the period. In each row of the point defects of Table 1, the number of frames (comas) in which defects were detected by each experimental defect inspection apparatus (number of times defects were detected) among the two-dimensional image data of 30 frames photographed within the period described above. Detection coma number ”column.

In the defect inspection apparatus for experiments 8-14, one defect detection target point is an imaging part (5 ₁ ). Imaging section of the imaging area (5 ₁ ) 300 frames of moving images are captured within the period from the image capturing region to the image capturing section 5 ₁ . It is imaged by. In each row of the line defects in Table 1, the number of frames (comas) in which the defects were detected by the respective defect inspection apparatuses among the 300 frames of two-dimensional image data picked up within the above period is displayed in the column of "detection coma numbers". It is described.

In Table 1, the zero detection coma number indicates that a defect could not be detected, and the magnitude of the detection coma number indicates the accuracy or certainty of defect detection. For example, when the detection coma number is a small number such as 1 to 2 coma, it is considered that the accuracy of defect detection is low. In Table 1, "x" (detection coma number is 0 coma), "△" (detection coma number is 1-2 coma), and "(circle)" (detection) according to the determination criteria which the applicant of this application independently determined the precision of defect detection. The coma number is classified into four stages of 3 to 6 coma) and "◎" (detection coma number is 7 coma or more), and is described in the column of "Decision".

In Table 1, the "held" row indicates in which frame the false occurred in 1800 frames (comma).

From the results of Table 1, it can be seen that the defect inspection apparatuses (the third and fifth experimental defect inspection apparatuses) using the high pass filter method and the peak method 2 can detect all kinds of point defects with good accuracy. . In other words, in the defect inspection apparatus according to the embodiment, the point defect image analysis unit for the (62 ₁ ~ 62 _n) If the point defect detection algorithm used in the high pass filter method or the peak method 2 is used, it can be seen that all kinds of point defects can be detected with good accuracy.

Moreover, from the results of Table 1, it was found that the defect inspection apparatuses (second and seventh experimental defect inspection apparatuses) using the edge profile method 2 and the edge curve method 2 can detect all kinds of line defects with good accuracy. Able to know. In other words, in the defect inspection apparatus according to the embodiment, the line defect image analysis unit (61 ₁ ~ 61 _n) line if the fault detection algorithm, the edge profile method 2 or edge curve method 2, all kinds of line defect for use in for It can be seen that can be detected with good accuracy.

In most cases of Table 1, since the detection coma number is smaller than the maximum value (30 in case of point defects, 300 in case of line defects), a plurality of two-dimensional images used for detection of defects in the present experimental example. It can be seen that there is a possibility that a defect cannot be detected in the method of detecting a defect based only on one two-dimensional image data (still image data) of the data (movie data). For example, one piece of two-dimensional image data is randomly selected from among a plurality of two-dimensional image data used for detection of a defect in the present experimental example, and the Kunic included in the sample 13 is detected based only on the one two-dimensional image data. In this case, even if a defect detection algorithm (edge curve method 2) capable of detecting the cubic can be most reliably used, the Kunic can only be detected with a low probability of 8/300.

This invention is not limited to each embodiment mentioned above, Various changes are possible in the range shown in a claim, and also the embodiment obtained by combining suitably the technical means disclosed respectively in the different embodiment is contained in the technical scope of this invention. do.

1 defect inspection device
2 forming sheets
3 conveying device (moving means)
4-linear light source
5 ₁ to 5 _n imaging section (imaging means)
6 analyzer
6A analyzer
6B analyzer
61 ₁ ~ 61 _n for the image analysis unit line fault (line fault detecting means)
62 ₁ ~ 62 _n that the image analysis unit for the defect (point defect detecting means)

Claims (5)

A defect inspection apparatus for detecting a defect of a molded sheet,
Imaging means for imaging a two-dimensional image of the molded sheet a plurality of times to generate a plurality of two-dimensional image data;
A linear light source for illuminating the molded sheet so that an image of the linear light source is projected onto a part of the imaged region in the molded sheet;
At least one of the molding sheet and the linear light source crosses the longitudinal direction of the linear light source, and the thickness direction of the molding sheet so that the position at which the image of the linear light source in the molding sheet is projected changes. Moving means for moving in an orthogonal direction,
A line defect detecting means for detecting a line defect from a plurality of two-dimensional image data generated by the imaging means,
The line defect detection means,
A line defect detection algorithm that fits an edge of an image of the linear light source in the two-dimensional image data into a function curve, and detects a point at which a distance between the edge of the image of the linear light source and the function curve is equal to or greater than a first threshold value; or
A line defect detection algorithm that obtains a curvature in the vicinity of each pixel in the edge of the image of the linear light source in the two-dimensional image data and detects a point at which the curvature is equal to or greater than the second threshold value as a line defect. The defect inspection apparatus which detects a defect. The method of claim 1,
The defect inspection apparatus further provided with the point defect detection means which detects a point defect from the two-dimensional image data produced | generated by the said imaging means. The method of claim 2,
The point defect detection means,
The change in luminance depending on the position along the straight line in the two-dimensional image data is represented as a luminance profile, and the quality of moving the plot group of the luminance profile so that the movement time between the plots is constant is assumed, and immediately before the plot of interest. The luminance value of the plot of interest is predicted from the velocity vector of the mass point between two plots and the acceleration vector of the mass point between three plots immediately before the plot of interest, and the difference between the predicted luminance value and the actual luminance value is predicted. Point detection algorithm for detecting a point at which is equal to or greater than a third threshold value as a point defect, or
The two-dimensional image data is smoothed, and the difference between the smoothed two-dimensional image data and the original two-dimensional image data is obtained as differential image data, and the point and luminance value at which the luminance value in the differential image data is equal to or greater than a fourth threshold value are The defect inspection apparatus which detects a point defect by the point defect detection algorithm which detects the point below a 5th threshold value (5th threshold value is smaller than a 4th threshold value) as a point defect. A defect inspection apparatus for detecting a defect of a molded sheet,
Imaging means for imaging a two-dimensional image of the molded sheet a plurality of times to generate a plurality of two-dimensional image data;
A linear light source for illuminating the molded sheet so that an image of the linear light source is projected onto a part of the imaged region in the molded sheet;
At least one of the molding sheet and the linear light source crosses the longitudinal direction of the linear light source, and the thickness direction of the molding sheet so that the position at which the image of the linear light source in the molding sheet is projected changes. Moving means for moving in an orthogonal direction,
Point defect detecting means for detecting a point defect from a plurality of two-dimensional image data generated by the imaging means,
The point defect detection means,
The change in luminance depending on the position along the straight line in the two-dimensional image data is represented as a luminance profile, and the quality of moving the plot group of the luminance profile so that the movement time between the plots is constant is assumed, and immediately before the plot of interest. The luminance value of the plot of interest is predicted from the velocity vector of the mass point between two plots and the acceleration vector of the mass point between three plots immediately before the plot of interest, and the difference between the predicted luminance value and the actual luminance value is predicted. Point detection algorithm for detecting a point at which is equal to or greater than a third threshold value as a point defect, or
The two-dimensional image data is smoothed, and the difference between the smoothed two-dimensional image data and the original two-dimensional image data is obtained as differential image data, and the point and luminance value at which the luminance value in the differential image data is equal to or greater than a fourth threshold value are The defect inspection apparatus which detects a point defect by the point defect detection algorithm which detects the point below a 5th threshold value (5th threshold value is smaller than a 4th threshold value) as a point defect. The method of claim 4, wherein
The defect inspection apparatus further provided with the line defect detection means which detects a line defect from the two-dimensional image data produced | generated by the said imaging means.

Images