JP3473800B2 - Defect inspection method and device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a defect inspection method and apparatus for an object to be inspected. In particular, the present invention relates to a defect inspection method and apparatus for an object to be inspected, which detects defects such as protrusions, scratches, bubbles, etc. existing on a transparent or semitransparent flat plate-like body such as a glass plate or panel or a plate-like body having a gentle curvature.

[0002]

2. Description of the Related Art Flat glass and panels manufactured by various manufacturing methods are judged to be defective if they have protrusions, scratches, bubbles, deposits, dirt, etc. (hereinafter referred to as defects) on their front, back or inside. Are discarded. Although these defects have been judged by visual inspection, there is a problem in that they cannot be inspected finely by visual inspection.

Therefore, conventionally, a two-dimensional image pickup device images an object to be inspected, and the image data obtained thereby is binarized to extract defect image data, so that a fine defect inspection can be performed. I have to.

[0004]

However, in these conventional defect inspection apparatuses, when detecting a defect in a plurality of inspected objects in a line work, it is determined whether or not the inspected object is in a defect inspectable position. There is a drawback in that the position detecting means for performing the above is additionally required, which complicates the entire apparatus and also complicates the application when the object to be inspected is conveyed in an irregular shape and shifted. Further, since non-defective parts such as edges of the object to be inspected, holes formed in advance, marks indicating the model number of the object to be inspected, etc. are also imaged together with the defects, it is necessary to determine and remove them. In the removal method, the image data of the non-defect portion is stored in advance and is removed from the entire image data in a form of collating with the image data of the non-defect portion. As described above, pre-storing the image data of the non-defect portion is very troublesome to handle, and different image data must be stored for different marks, and the versatility of the apparatus is poor.

The present invention has been made in view of the above circumstances, and provides a defect inspection method and apparatus for an object to be inspected, which is easy to handle and is less likely to misidentify a non-defect portion as a defect. To aim.

[0006]

SUMMARY OF THE INVENTION The present invention provides a defect inspection method for an object to be inspected for detecting defects such as protrusions, scratches, bubbles, adhering substances and dirt existing on the object to be inspected. When the defect is detected from the defect image data by imaging the image data of the inspected object output from the image capturing device by binarizing the image data while continuously conveying In advance, among the individual defect images in the defect image data, those which are within a predetermined distance from each other are linked to form one defect region, and one of the areas of the defect region which is larger than a predetermined area is selected. It is a defect inspection method and apparatus characterized by being discriminated as an edge image of an object to be inspected.

In a preferred aspect of the present invention, when a defect is detected from defect image data, at least one of the area and position of the defect area, the vertical and horizontal dimension ratio of the circumscribing quadrangle, and the area thereof is used in advance. It is characterized by discriminating and removing a non-defect image such as a hole image of an inspection object and a mark image attached to the inspection object. According to the present invention, by separately determining the edge image from the defect image data, it is not necessary to separately provide a means for determining whether or not the inspection object is in a position where defect inspection can be performed. Defects can be detected even when the object to be inspected is transported in an irregular shape with a shift. That is, in the present invention, first, the object to be inspected is imaged by the image pickup means. Then, the image data of the object to be inspected output from the image pickup means is binarized to extract defect image data. Then, among the individual defect images in the defect image data, those that are within a predetermined distance from each other are linked to form one defect area, and the area of the defect area that is larger than a predetermined area is covered. A defect is detected by discriminating and removing it as an edge image of the inspection object, and detecting the presence or absence of the defect image from the defect image from which the edge image has been removed.

Further, according to one aspect of the present invention, when a defect is detected from defect image data, at least one of the area and position of the defect region, the vertical and horizontal dimension ratio of the circumscribing quadrangle, and the area thereof is previously set. By distinguishing and removing non-defective images such as hole images of the inspected object and mark images attached to the inspected object, it is not necessary to store the image data of the non-defective area in advance, and the handling is very easy. To be simplified.

[0009]

BEST MODE FOR CARRYING OUT THE INVENTION Preferred embodiments of a defect inspection method and apparatus for an object to be inspected according to the present invention will be described below in detail with reference to the accompanying drawings. 1 is a perspective view of an essential part of a defect inspection apparatus for an object to be inspected according to the present invention, and FIG. 2 is a side view of an essential part of the defect inspection apparatus shown in FIG. As shown in FIGS. 1 and 2, the defect inspection apparatus 10 includes conveyance rollers 12, 12, ..., Fluorescent lamps 14, and a CCD line sensor 16 which is a one-dimensional image pickup means.

As shown in FIG. 2, the glass plate 20 to be inspected is placed on the conveying rollers 12, 12 ... And these conveying rollers 12, 12 ... Are rotated at a constant speed by a driving device 17. To be driven. As a result, when the conveying rollers 12, 12 ... Are rotated counterclockwise by the driving device 17, the glass plate 20 is conveyed at a constant speed in the arrow A direction in the drawing.

The fluorescent lamp 14 is housed in a light source box 18 and is located below the glass plate 20 in the conveying direction A of the glass plate 20.
Are arranged in a direction orthogonal to the. Further, a slit 22 is formed on the upper surface of the light source box 18 along the longitudinal direction of the light source box 18, and the light from the fluorescent lamp 14 is emitted toward the glass plate 20 through the slit 22.

The CCD line sensor 16 is installed above the glass plate 20, and its pixel row is arranged so as to face the slit 22 of the light source box 18. Accordingly, the glass plate 20 in conveyance is illuminated by the fluorescent lamp 1 4,
Then, the transmitted light that has passed through the glass plate 20 is imaged on the light receiving surface of the CCD line sensor 16 via a photographic lens (not shown). Although transmissive illumination is used in this embodiment, reflective illumination may be used.

The image light formed on the light receiving surface of the CCD line sensor 16 is sequentially output as a voltage signal. The output voltage signal, whose gain is controlled by an analog amplifier, is output to the image processing device 24 and converted into a digital signal by the A / D converter 26 of the image processing device 24. The digital signals are sequentially stored in the image memory 28, and the digital signals are binarized.
Is output to. The binarization processing unit 30 binarizes the digital signal with a predetermined black and white threshold value. Then, the binarized digital signal is stored in the image memory 32 and then output to the image processing unit 34 to detect a defect. The defect detection method by the image processing unit 34 will be described later. The binarization processing unit 30, the image memory 32,
The image processing unit 34 constitutes the defect determination means of the present invention.

The binarization process may be performed by detecting a black image as a defect image with respect to a white image background, or by detecting a white image as a defect image with respect to a black image background. But it's okay. In addition, a black and white image may be detected as a defect against a halftone background. These may be appropriately selected depending on the nature of illumination, the nature of defects, and the like. In addition, the image memory 28
The digital signal of the glass plate 20 for one sheet stored in
It is converted into a video signal and output to the monitor 36. As a result, the image of the glass plate 20 is displayed on the monitor 36.

Next, the operation of the defect inspection device for an object to be inspected constructed as described above will be described. First, the transport rollers 12, 12 are driven to transport the glass plate 20 toward the imaging area at a constant speed. And CCD line sensor 1
The signals imaged and output in 6 are sequentially stored in the image memory 28 of the image processing device 24 as digital signals. That is, while the object to be inspected is conveyed at a constant speed, the object to be inspected is imaged by the one-dimensional imaging means arranged orthogonal to the conveying direction of the object to be inspected.

When a predetermined amount of information is stored in the image memory, the CPU 42 controls the image processing device 24 to output the digital signal stored in the image memory 28 to the binarization processing unit 30. . Then, the binarization processing unit 30
Then, the digital signal is binarized by a predetermined black and white threshold value to obtain an edge of the glass plate 20, a protrusion, a scratch, a bubble, a hole,
A digital signal indicating a mark or the like is extracted. The extracted digital signal is stored in the image memory 32 and output to the image processing unit 34.

Next, a method of discriminating a defect by the image processing unit 34 will be described. First, the image processing unit 34 two-dimensionally develops one glass plate 20 based on the digital signal, and then, in advance, from the two-dimensionally developed image, the individual defect images in the defect image data are mutually pre-registered. One defect area is linked by linking objects having a predetermined distance or less, and an edge image of an object to be inspected that is larger than a predetermined area out of the area of the defect area (image of edge 20A (see FIG. 3)). It is determined and removed as. Since this edge image has a one-to-one correspondence with each glass plate to be conveyed, by discriminating the edge image in advance, there is a drawback that it will be detected later without separately providing a position detecting means for the glass plate. The position of which glass plate is located can be determined.

The area of the edge 20A is obtained by multiplying the long side X and the short side Y of a rectangle surrounding the edge portion of the glass plate 20 as shown in FIG. In this case, the area of the defect is stored in advance as the area of the quadrangle, and the long side X0 and the short side Y0 of the quadrangle are compared with the long side X and the short side Y of the edge 20A to make the determination. Also,
Since the normal edge image forms the largest defect area, the largest defect area detected in a certain time interval can be discriminated as an edge image.

After the edge image is discriminated and removed, the glass plate 2 is preliminarily determined according to at least one of the area and position of the defect area, the aspect ratio of the circumscribing quadrangle, and the area thereof.
The image of the hole 20B (see FIG. 4) bored in 0 (see FIG. 4) and the image of the mark 20C (see FIG. 5) are discriminated and removed. As shown in FIG. 4, the hole 20B is discriminated by obtaining diameter dimensions X and Y in two orthogonal directions and comparing these X and Y with the long side X0 and the short side Y0 of the defect area. . Further, the mark 20C composed of a plurality of characters or the like is discriminated as shown in FIG. First, each character or the like forming the mark portion is recognized as a defect area. The long side X2 and the short side Y2 of this defect area are often indistinguishable from actual defects such as scratches. Therefore, the defect areas within the predetermined sizes X1 and Y1 are further connected in advance to make the entire mark portion a new defect area.

Then, the total number of pixels of the entire mark portion is counted, and the area is calculated by multiplying the long side X and the short side Y of the rectangle surrounding the edge portion of the total number of pixels. Then, the determination is made by comparing the long side X and the short side Y of the mark 20C with the long side X0 and the short side Y0 of the defect area. Next, the image processing unit 34 uses the edge 20A and the hole 2
The presence or absence of an actual defect portion 20D (see FIG. 2) such as protrusions, scratches, bubbles, etc. is detected depending on whether or not the image exists after removing the image of the non-defect portion indicating 0B and the mark 20C. That is, when the image does not exist, the glass plate 20 is determined as a good product, and when the image exists, it is determined as a defective product. Then, the discrimination result is displayed on the monitor 36.

As described above, according to the present embodiment, since the CCD line sensor 16 can perform the image capturing while conveying the plurality of glass plates 20, the cost can be reduced and the flexibility of the image capturing range can be widened. You can Further, in advance, the individual defect images in the defect image data, which are within a predetermined distance from each other, are linked to form one defect area, and the area of the defect area larger than a predetermined area is selected. Since the edge image of the glass plate 20 is discriminated, the defect of the glass plate 20 can be detected by the flow work without separately providing the position detecting means of the glass plate 20. In particular, even if the glass plate 20 has an irregular shape and the glass plate 20 is misaligned and conveyed, the defect can be identified. Further, in the image processing unit 34, an image of a hole and an image of a mark which are preliminarily drilled in the glass plate according to at least one of the area and the position of the defect area, the aspect ratio of the circumscribed quadrangle, and the area thereof Since the non-defect image of was identified and removed,
Very easy to handle.

In the present embodiment, the digital signal output from the A / D converter 26 is binarized in the binarization process for the defect extraction, but the present invention is not limited to this. For example, an image of the glass plate 20 is displayed on the monitor 3
CCD line sensor 16 if there is no need to display 6
The voltage signal output from may be binarized. Furthermore, when the size of the non-inspection body is wide in the width direction, it is possible to take an image using two or more cameras.

[0023]

As described above, according to the method and apparatus for inspecting a defect of an object to be inspected according to the present invention, the plurality of objects to be inspected are conveyed at a constant speed while being imaged by the image pickup means.
Obtaining defect image data and linking, in advance, individual defect images in the defect image data that are within a predetermined distance to each other to form one defect region, and a predetermined area of the area of the defect region is obtained. Since a larger image is discriminated as the edge image of the object to be inspected, the defect of the object to be inspected can be detected by the flow work without separately providing the position detecting means of the object to be inspected.

Further, non-defects such as images of holes and images of marks preliminarily drilled in the glass plate are selected depending on at least one of the area and position of the defect area, the vertical and horizontal dimension ratio of the circumscribing quadrangle, and the area thereof. Since the image is discriminated and removed, the inspection device is extremely easy to handle, and the defective image is rarely recognized as a defect. Furthermore, the cost can be reduced and the degree of freedom of the imaging range can be increased by performing the imaging by the one-dimensional imaging means while conveying the inspection object at a constant speed.

[Brief description of drawings]

FIG. 1 is a perspective view of a defect inspection device for an inspection object according to the present invention.

Side view of a defect inspection apparatus of the device under test according to the present invention; FIG

FIG. 3 is an explanatory diagram for calculating an edge area of a glass plate.

FIG. 4 is an explanatory diagram for calculating an area of a hole portion of a glass plate.

FIG. 5 is an explanatory diagram for calculating the area of the mark portion of the glass plate.

[Explanation of symbols]

10 ... Defect inspection device 12 ... Conveyor roller 14 ... Fluorescent lamp 16 ... CCD line sensor 20 ... Glass plate 24 ... Image processing device 30 ... Binarization processing unit 34 ... Image processing unit 42 ... CPU

─────────────────────────────────────────────────── ─── Continuation of front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G01N 21/84-21/958 G01B 11/00-11/30

Claims (4)

(57) [Claims] 1. A defect inspection method for an object to be inspected for detecting defects such as protrusions, scratches, bubbles, adhering substances, and stains existing on the object to be inspected, while imaging a plurality of objects to be inspected continuously. Imaged by means, the image data of the object to be inspected output from the image pickup means is binarized to extract defect image data, and when the defect is detected from the defect image data, in advance,
Of the individual defect images in the defect image data, those which are within a predetermined distance from each other are linked to form one defect region, and the defect region having a larger area than a predetermined area is inspected. Defect inspection method characterized by discriminating as an edge image of. 2. The defect inspection method according to claim 1, wherein at the time of detecting the defect from the defect image data, at least one of the area and position of the defect region, the vertical and horizontal dimension ratio of the circumscribing quadrangle, and the area thereof. A defect inspection method for an object to be inspected, which comprises discriminating and removing a non-defect image such as a hole image of the object to be inspected and a mark image attached to the object to be inspected. 3. A defect inspecting apparatus for an object to be inspected, which detects defects such as protrusions, scratches, bubbles, adhering substances, stains, etc. existing on the object to be inspected, and a conveying means for continuously conveying a plurality of objects to be inspected. An image pickup means for picking up an image of the conveyed inspection object; a binarization processing means for binarizing the image data of the inspection object outputted from the image pickup means to extract defect image data; Of the individual defect images in the defect image data of the object to be inspected extracted by the binarization processing means, those which are within a predetermined distance from each other are linked to form one defect area, and the area of the defect area Among them, the one having a larger area than a predetermined area is discriminated and removed as an edge image of the object to be inspected, and the defect is detected by detecting the presence or absence of the defect image from the defect image data from which the edge image has been removed. Defect discriminator An apparatus for inspecting defects of an object to be inspected, which comprises steps and. 4. The defect inspecting apparatus according to claim 3, wherein the defect discriminating means determines the object to be inspected based on at least one of the area and position of the defect region, the aspect ratio of the circumscribed quadrangle, and the area thereof. A defect inspection device for an object to be inspected, characterized by discriminating and removing a non-defect image such as a hole image and a mark image attached to the object.