JP2746506B2 - Inspection method of netted glass - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of detecting a defect in a continuously conveyed ribbon-shaped netted glass. What arrange | positioned the line | wire in the stripe form may be called a lined glass, but this invention also includes a lined glass.

[0002]

2. Description of the Related Art As a method of detecting a defect of a screened glass, a pattern of the screen is read in advance by an optical sensor, stored in a computer, and a signal of the screen and a defect read at the time of inspection is used to detect the screen. A pattern in which a pattern is programmatically filtered and cut, and only a signal of a remaining defect is compared with a quality standard level.
No. 22689), and irradiate light from one side of the sheet glass,
An image is captured by a plurality of CCD cameras provided on the other surface, and the amount of transmitted light for each region is changed from large to small in adjacent scanning.
Alternatively, there has been proposed an apparatus that determines an area that changes from small to large as a defect (Japanese Patent Application Laid-Open No. 3-57944).

[0003] However, in the case of the former technology, not only is it complicated to input a net pattern into a computer in advance, but also the operating conditions are changed.
If the stored halftone pattern deviates from the actual halftone pattern due to a change in this condition, the halftone portion erasing process will not be performed accurately, and there is a risk that the halftone portion may become a defect or the defective portion may be overlooked. is there.

[0004] In the latter case, it is possible to detect when the wire constituting the mesh of the sheet glass is arranged in parallel to the traveling direction of the sheet glass. However, when the wire is arranged in the width direction, When the wire is arranged obliquely in the traveling direction, for example, when the net pattern forms a rhombus, it is necessary to rotate the camera by moving the handle and tilt the scanning angle of the camera. There is a drawback that if any deviation occurs, the defect cannot be detected accurately. Further, since the net at the intersection of the nets of netted glass arranged in a lattice or a rhombus is wider than other portions, this portion is erroneously detected as a defect.

[0005] The present invention has been made in view of such a point, and the disadvantages of the netted glass can be solved by using a grid type or diamond type without having to store the net pattern in a computer in advance and changing the position of the camera. It is an object of the present invention to provide a method for inspecting a netted glass which can accurately detect a defect even in the case of a type of netted glass.

[0006]

SUMMARY OF THE INVENTION According to the present invention, a light from a light source is projected onto a ribbon-shaped netted glass which is continuously conveyed, and a plurality of two-sided glass plates are provided on the opposite side of the netted glass. In a method of imaging a netted glass with a three-dimensional camera and inspecting the netted glass based on the obtained grayscale signal ,
When there is a pixel of a dark signal that is continuous in the vertical, horizontal, or oblique direction within the range of the number of pixels , the pixel portion of the dark signal is regarded as a halftone dot.
And regards the dark signal of the pixel center to remove the dark signal to 1 to 4 times the width of the web at regular intervals from the width direction of the remaining dark signal <br/> No. disadvantageous, its The size and the position of the pixel are obtained .

[0007]

[Function] In removing (masking) a signal from the network,
By judging whether there is a dark signal that is continuous not only vertically but also horizontally or diagonally in a predetermined range,
Defects can be detected regardless of the camera position,
In addition, by removing the dark signal at equal intervals from the center in the width direction of the dark signal to a width larger than the mesh, wide parts such as the intersection of meshes such as lattice-type or diamond-shaped netted glass are not regarded as defects. The defect can be detected accurately.

The masking width is the same as that of the net, that is, the masking width is used to check whether or not the nets are continuous as in the embodiment described later. If the width of the net is too close to the width of the net, there is no masking effect. If the width is too large, there is a risk of overlooking the defect. , 1.5 to 4 times, preferably 2 to 3 times the width of the net.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the drawings. FIG. 1 is a schematic view of a main part showing an inspection apparatus of the present invention, and FIG. 2 is a view showing a binary image signal obtained by imaging a part of a netted glass.

As shown in FIG. 1, a light source 2 such as a high-frequency fluorescent lamp housed in a box for detecting a net defect and an edge of the netted glass are provided below the netted glass 1 conveyed in the direction of the arrow. A light source 3 housed in a box for detection is arranged, and a plurality of CCD cameras, etc. for imaging the entire width of the netted glass are provided at a position facing the light source 2 and the light source 3 above the netted glass. And a one-dimensional camera 5 such as a line sensor for detecting the edge of the screened glass.

The output of the two-dimensional camera 4 is input to an image processing device 6 and outputs a processed signal to a personal computer 7 or the like. Further, the one-dimensional camera 5 is connected to a net width measuring controller 8, and even if the edge position of the net changes due to meandering or the like, a signal from the controller 8 is input to the image processing apparatus via the computer 7 and the edge is changed. The range in which an image is captured by a nearby two-dimensional camera is adjusted to be constant.

Next, the procedure of defect detection will be described. The light from the light source 2 is applied to the screened glass, and the output signal of the two-dimensional camera 4 that has captured the screened glass is input to the image processing device 6. It is compared whether the value is larger than a predetermined value (slice value), and a binarized signal from, for example, one two-dimensional camera 4 is used for a glass with a diamond-shaped net as shown in FIG. Is displayed as a dark signal filled in black.

In the image processing device 6, first, it is determined whether or not the dark signal is continuous. Pixels are represented by G (1,
1), G (1, 2)... G (1, m), G (2, 1),
G (2,2), ... G (2, m), ... G (1,
1)... G (m, m), for example, B (1, 1), B (1, 2).
B (1, n), B (2,1)... B (2, n),.
.., B (n, 1)... B (n, n), and B (1,
When it is first checked whether there is a dark signal in the order of 1), B (1, 2)..., A dark signal appears first in B (1, 1). Then, in B (1,1), when the pixel G (1,1) to the pixel G (1,8) are examined, there is a dark signal in G (1,1). It is checked whether the oblique (lower left, lower right) pixels, G (2, 1) and G (2, 2) in this example, are dark signals, and since both are dark signals, G adjacent to these pixels (2,3), G (3,1), G
When (3, 2) and G (3, 3) are examined, G (3, 1), G
Since (3, 2) and G (3, 3) are dark signals, the pixels of dark signals adjacent to these pixels are examined to find G (4, 1),
G (4,2), G (4,3) and G (4,4) correspond. G (8,6), G (8,7),
The pixel of the dark signal that reaches the end of B (1,1) of G (8,8) and then is adjacent to pixels G (8,6), G (8,7), G (8,8) is G (8,8). 8, 9), G (9, 8), G (9,
9), G (9, 10), so B (1, 2), B
When (2,1) and B (2,2) are examined, B (1,
The dark signal is not continuous between 2) and B (2, 1).
For (2, 2), when a block in which the dark signal is continuous to the end and thereafter is sequentially determined, the dark signal is represented by B (3, 3),
B (4, 4)... B (m, m), which is continuous from the end of the screen to the end, is regarded as a net, and the center of the net at both ends, for example, G (6, 3) to Pixel G at the center of G (6,6)
(6, 4), G (6, 5) and G (m-2, m-3) to G
Pixels G (m-2, m-2) to G at the center of (m-2, m)
When (m-2, m-1) is connected by a straight line and masked at equal intervals from the straight line, the mask is masked up to the hatched portion in FIG. 2 and the pixels in the hatched portion including the dark signal portion are removed.

The dark signal that continues from B (1,3) to B (3,1) is also regarded as a net, and is masked and removed in the same manner. After the masking, a dark signal due to the defect remains, and the defect is determined by calculating its magnitude. In practice, the drawbacks are that the drawback tends to be long in the flow direction of the screened glass, and that the drawback tends to be long in the vertical direction on the screen. Therefore, the dark signals that are continuous in the horizontal and vertical directions are counted. Is the minor axis), and if the major axis is 1 mm or less when converted to the actual size, small dust and larger than 1 mm 2
mm or less, fine bubbles (surface bubbles) generated by entraining air when the mesh is inserted, the ratio of the major axis to the minor axis is greater than 2 mm, and the ratio of the major axis to the minor axis is very large. In the case of a large signal of 4 mm or less, if there is a light signal in the dark signal, a relatively large bubble (eyeball bubble) generated by a foreign substance attached to the net or a lip tile or the bubble is pulled in the ribbon flow direction and the surface is broken. Torn bubbles,
If the dark signal is clogged, unmelted gravel, major axis 4
If it is larger than mm, it is judged to be a large gravel.

In this embodiment, the dark signal remaining after masking is in block B (3, 2), and the major axis is 7 pixels in the vertical direction, the minor axis is 5 pixels in the horizontal direction, and the major pixel is 5 pixels. The minute is converted to the actual size, and if it is 1.4 mm, for example, it is determined to be a surface bubble.

When a defect is detected, data such as the type and size of the defect can be output to the personal computer 6 or the like, and can be tabulated and processed as various management information. In addition, the defect can be detected instantaneously, and the location can be specified. Therefore, the position is marked, and the mark is detected in a later process, and the netted glass having the defect is automatically cut and removed. You can also.

Although the preferred embodiments have been described above, the present invention is not limited to these embodiments, and various applications are possible. The netted glass to be inspected includes not only the netted glass in which a rhombic or square net is enclosed in a plate glass, but also the lined glass in which the wires are arranged in a stripe shape and enclosed in the plate glass. In addition to smooth transparent glass, even frosted glass and template glass having fine surface irregularities can be detected.

In the image processing, when it is determined whether or not the dark signal is continuous, the dark signal is counted in the direction of the line (vertical direction in the screen) in the case of a lined glass or the like, and the dark signal is counted above a predetermined count value. For some, the width may be detected by counting pixels in the horizontal direction of the screen, and masking may be performed from the center. In addition, whether or not the nets are continuous is checked in the embodiment. At this time, a portion considered to be continuous may be masked. In this case, the masking width is equal to the net width.

As the two-dimensional camera, in addition to the two-dimensional camera, one using a MOS, an image pickup tube, and the like can be used.

[0020]

As described above, according to the present invention, the center of the width of a net or a line is determined and masked at equal intervals from the center to be larger than the net or the line. It is not necessary to store the mesh pattern in the computer in advance, and without changing the position of the camera, even in the case of a grid-type or rhombic mesh-type glass, as well as a line-type glass in which the lines are arranged in stripes. In this case, the defect can be automatically and accurately detected.

[Brief description of the drawings]

FIG. 1 is a schematic view of a main part showing an inspection apparatus of the present invention.

FIG. 2 is a diagram illustrating a binarized image signal obtained by imaging a part of the netted glass.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Netted glass 2 Light source 4 2D camera 6 Image processing device

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) G01N 21/88-21/89

Claims (1)

(57) [Claims] 1. A method for projecting light from a light source onto a ribbon-shaped netted glass that is continuously conveyed, and imaging the netted glass with a plurality of two-dimensional cameras provided on the opposite side to the netted glass. In the method of inspecting the screened glass using the obtained dark and light signals , when there is a dark signal that is continuous in a vertical, horizontal or oblique direction within a predetermined number of pixels , the dark signal is
The pixel portion of the signal is regarded as a halftone dot, the dark signal is removed from the center in the width direction of the pixel of the dark signal at equal intervals to a width of 1 to 4 times the width of the halftone dot , and the remaining dark signal is regarded as a defect. , Its size
And a method of inspecting a screened glass.