JPS5899738A - Apparatus for detecting defect on transparent body - Google Patents

Abstract

PURPOSE:To effect a defect detection with a desired resolving power, by reducing the distance between the adjacent unit elements in a diode array in order to narrow the dead zone width, as well as carrying out an analog operation between the outputs from the adjacent unit elements. CONSTITUTION:The light from a diffusion plate 2 is applied to a glass bottle 1, and the transmitted light is condensed by a lens system 3 to form a real image 5 of the bottle 1 on an image-forming surface 4. A diode array 6 as a photoelectric transducer is disposed on the image-forming surface 4. The diode array 6 is arranged so that the dead zone 7 between the adjacent unit elements A1, A2,... has a narrowed width W. The outputs EA1, EA2,... EAn obtained from the respective unit elements are subjected to an analog operation of¦EAn-EAn+1¦ (n=1, 2, 3...) for the adjacent unit elements thereby to detect a possible defect on the glass bottle 1.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a defect detection device for transparent bodies including semi-transparent bodies such as glass bottles.

Various inspection machines are in practical use for inspecting transparent containers such as bottles, but they are difficult to detect for so-called appearance defects such as dirt on the bottle surface, bubbles and other impurities mixed into the material base, and wrinkles on the product surface. Visual inspection is still the most commonly performed inspection. Conventional defect detection devices have been developed that use laser light, television cameras, diode arrays, etc. However,
For those that use laser light, both the irradiated light from the projector and the reflected light from defects are highly directional, making it difficult to control the resolution according to the size of the defect desired to be detected. There were major technical and economic problems such as price.

In addition, when using a television camera, the resolution is determined by the size of the scanning aperture, and if higher resolution control is required, logic control is required. In addition, in conventional diode arrays, the resolution is determined by the area of the unit element, but the unit length of a normal self-scanning diode array is about one-half of one pitch, and the effect of the fuwazai is low. In order to avoid this, a fine pitch must be used as the minimum resolution, and control by logic is required, which is economically disadvantageous.

That is, FIGS. 1 to 4 are front views showing the arrangement of a conventional self-scanning diode array. (A) indicates the unit element, and (B) indicates the dead zone. This dead zone (b) is about one pitch P of the unit element (a). Therefore, there is a large difference in sensitivity between when the real image of the defect (c) comes above the unit element as shown in Figure 2 and when the real image of the defect (c) comes to the dead zone (b) as shown in Figure 3. It makes a difference. Therefore, in order to avoid this, the size of the minimum defect to be detected is 2 to 3 as shown in Figure 184.
The size of the image (c) and unit element (b) had to be chosen so that they would be exactly the same. The present invention aims to provide a device that eliminates the drawbacks of the prior art, and for this purpose, it employs a distinctive diode array, performs analog calculations on the output amounts of two or more neighboring unit elements, and Inspection is performed at the desired resolution without using logic control or large-capacity memory, and malfunctions caused by the pattern of the object to be inspected, unavoidable mold seams, and other satin finishes, in other words, non-defective products, are eliminated as defective products. This prevents the occurrence of defects and enables effective inspection of appearance defects. This will be explained in detail below.

FIGS. 5 and 6 show an example in which the apparatus of the present invention is applied to defect inspection of the body of a glass bottle 1, which is a typical example of a transparent object to be inspected including translucent objects. A glass bottle l is irradiated with light from a diffused light source or a diffuser plate 2, and the transmitted light is focused by a lens system 3 to form a real image 5 of the bottle l on an imaging plane 4. A diode array 6, which is a photoelectric converter, is arranged on this image forming plane 4 (see Figs. 8 to 14), and this diode array 6 converts the optical signal into an electrical signal, and the output electrical signal is subjected to analog calculation processing. to detect defects in 1 part of glass bottle 1t. In this case, the glass bottle 1 is rotated in the inspection position. FIG. 7 shows an example of inspecting the bottom of a glass bottle 1 for defects. In this case, diffused light is irradiated from near the bottom of the glass bottle 1. Note that during the inspection, instead of rotating the bottle 1, the detection system may be rotated.

Next, the diode array 6 to be arranged on the imaging plane 4 will be explained. Before that, it should be stated that in the present invention, the shape, arrangement, and analog calculation formula of the diode array 6 are considered in order to prevent the magnitude of the defect detection signal from changing depending on the detection position. By doing that,
The second purpose is to reduce the signal of the parts other than the defect as much as possible, and the third purpose is to reduce the number of necessary equipment such as the number of processing amplifiers.

First, in the present invention, as shown in FIG. 8, the width W of the dead zone 7 between each unit element A4*A1... of the diode array 6 is narrowed. For example, the pitch P is 1.51EI and the radius W is 0.
3111. Diode array 6 shown in FIG.
is placed on the imaging plane 4. In the case of inspection of the 1st part of the bottle, the arrangement should be ``the longitudinal direction of the diode array 6 is
In the case of inspections 1 and 5 of the bottom of the bottle 1, the longitudinal direction of the diode array 6 is aligned in the radial direction of the bottom of the bottle 1. The arrangement of this diode array 6 is the same for other diode arrays 6 shown below.

The unit elements Al are lined up in one row.

In the diode array 6 consisting of A,... now,
Output from each unit element AI + A4 y Al-... 1lltm 1°KA! ylm3. ...and 1 as an analog/four calculation formula! cAn”An+11 (n −
If 1t 2 e 3 ”・) is adopted, the seam line etc. in which the image passes through the diode array 6 in the vertical direction will be eliminated. That is, lI!im□Hj z l =Q
+ l m A , I A s l 0Q +…
This is because it is not detected because 1]1Alln+5l=0. On the other hand, in the diode array 6 as shown in FIG. 8, even if the defect image H2 has the same size, the defect image H2 that passes through the dead zone 7 has an output of 11cA4-"AS l t l"Al.
l -x,, l l 11cA6 薯- and the output of M 71 does not pass through the dead zone 7. Output of defect image H l KA, -m, 31 * l IAI ''A41
As a result, the defective image processing is judged to be due to a defect that is smaller than its actual size. In other words, in defect imaging, unit element A1
If the image area occupied by 1 is S''6, and the image area occupied by unit element A is S, then the output IKA4-E^,l is the area S,
corresponding to the output 1mA, -IIliA, l is 186
Corresponding to the area of ssl, the output l K A @ K
A? Since l corresponds to the plane WIss, it is not possible to obtain an output that accurately corresponds to the defect image.

Next, in the present invention, the rectangular unit elements A and IA are used.

Two means were considered in order to eliminate the five drawbacks of the diode array 6 shown in FIG. 8, in which . One means for this is to make each unit element A1+AI... into a parallelogram, as shown in FIG. By arranging the corresponding sides of each parallelogram on the same straight line, the dead zone 7 forms an angle with respect to the passing direction X of the defective image (11), and the influence of the dead zone 7 can be reduced. It is from. That is, to explain the case in which the defect image J passes through the diode array 6 in the direction of the X arrow in FIG. J is located! When , the output l E a 4 '' A s l becomes a value that approximately corresponds to the area of the defect image J, and the influence of the dead zone 7 can be reduced.

Another method is to use a diode array 6 arranged in two rows as shown in FIG. 1O. In this case, it is preferable to arrange the A column and the third column so that they are shifted by, for example, half a pitch. Now, in this example, the defect image K is the diode array 6.
When the defect image X passes in the direction of the X arrow, the output in column A is I E A @-KA@ l t lI! ′＾E
A41'l"^4"Aal has a much lower value than the output corresponding to the area of the defect image X, but on the other hand, the output in the third column 1 ]lin, Hm, I + lE+s.

Since -1nl is a value corresponding to the area of the defect image, the influence of the dead zone 7 can be eliminated. However, the defect image indicated by the symbol is
When passing in the direction of the arrow, output 111, 'H1,
l*lEA, -zm, l and l"i"s
Since both Isl 1C) and -ICBy + are less than the output corresponding to the area of the defect image, they will be affected to some extent by the dead zone 7. As an example of preventing the influence of the dead zone 7 by forming two rows, the diode array 6 shown in FIGS. 11 to 13 can be considered.

Figure 11 shows the A and 3 rows shifted by 1 pitch.0 Figure 12 shows the parallelogram unit element AI+A,...
・B,,B,... are arranged symmetrically in two rows.

Unit elements A1+A1... in row A and unit element B in row 3,
, B, . . . are arranged symmetrically in order to prevent signal variations due to the direction of inclination of the diagonal defect. Figure 13 shows a parallelogram unit element AI +Ag...B 1
+ B @... is shifted by half a pitch between column A and column 3. Similarly, in FIG. 14, row A and row 3 are shifted by one pitch.

As mentioned above, the means to prevent the influence of the dead zone 7 is to narrow the width W of the dead zone 7 itself, and to reduce the width W of the dead zone 7 itself. +
A! . . are made into parallelograms and arranged in two rows, but the present invention also considers means for reducing the number of amplifiers necessary for inspection. In other words, if only defects are to be detected, the analog calculation formula is as follows between the upper and lower unit elements for each column: l za, -'x*, I f l
IcAl -X Al l + I'HA@ -H^
, 1゜-”'11cm-n mAn+s l + 1
1! +11. -IC), I*l11cm, Km
, 1°°°1chi In 1! ! mml+I I, but in this case, processing systems 0 are required for the number of unit elements in the processing steps of the device described later, each processing system 0 requires multiple processing devices, and each If amplification or the like is required for each processing device, the number of amplifiers will be equal to the number of processing devices x the number of unit elements, making the device large, occupying space, and becoming expensive. Therefore, in the present invention, A of the diode 6 is used.

Analog calculation formula 1 (Pin + IC + n) (IC6z + s + Kmn
+IN (n-1, :L51+0Ll(IC*H+l
mn-1-t) (EA!1+I-v--N l (
,,l engineering 1 '1st generation-844a FR, J-'!
Now, as an example, we will use the formula l(] for diode 6 shown in Figure @14.
1CAn+Kin) (Ixn+x +IC+n+t)
To explain the case where the defect image M passes through the diode array 6 in the direction of the X arrow by applying l, the output l (1CA1+
Em, )(-]Ca, +Mx, ) l is zea, output 1 (ICA, to +!+, )-(Em4-""4
) I is a value whose waveform peak corresponds to the area of the defect image M. Furthermore, output 1 (IC II""II)
−Cytm, +I, m, )I also has a value corresponding to the area of the defect image M. Further, even when the defect @M passes through another position -, a value corresponding to the area of the defect image M can be obtained from any of the outputs. In this case, the number of operations required may be the number of unit votes. As an example, the diode array 6 shown in Figure 10 has the formula I (11!*U+Kmm++)
A case where (KAn+t +Imn)l(nsml, :L5...) is applied is shown. First, when the defect image passes through the diode array 6 in the direction of the X arrow, the output 1 (IM1 + I
C), )-(E*, +ICm, ) I is zero, output 1 (ICA, +'E), )-(E*4 +]ri
m, ) The waveform peak of I corresponds to the area of the defect image X. Also, when the defect image moves in the direction of the X arrow, the output l
(]Ii*, +1!:), ) (IC,, +1
8ml1l) l and output + (KA, +!+,)
(mA.

+IC+, )l corresponds to the area of the defect image. The same applies when L passes through other positions in the defect image.

Even in these cases, the number of operations required is only % of the number of unit elements. Of course, analog calculations between these four unit elements will cause some error from the output corresponding to the area of the defect image, but it will be quite large. Accuracy can be ensured. For example, in the diode array 6 shown in FIG.
If (n=1 e 3 t 5”) is adopted,
Theoretically, it is about ±20%, and the data shows that the P-wave device 1 described later
Due to the electrical characteristics factors No. 2 and 13, the error was within ±10% or less. In the case of the diode array 6 shown in FIGS. 11 and 12, the accuracy is not very good. The reason why the required number of operations is 2, which is the number of unit elements, is 4 for analog operations.
By using unit elements that are at different positions in the A and B rows, you can obtain an output that corresponds to the area of the defective image by performing duplicate operations. 〇In addition, as another usage method, the number of unit elements is 2
In the diode array 6 arranged in a row, it is also possible to perform subtraction between the left and right unit probe outputs.

This method is effective, for example, when one of the adjacent unit elements is used as a dummy in order to prevent the influence of changes in wall thickness in the circumferential direction of the bottle l.

Figures 1 and 14 are circuit diagrams showing the mechanism of this device, and 11 is an arithmetic circuit that receives signals from four predetermined neighboring unit elements of the diode array 6 and performs arithmetic operations using the above-mentioned analog/four arithmetic formula. Let's go. Thereby, four output signals from four unit elements are made into one output signal.

Next, this output signal is input to the high frequency converter 12.

The high-pass filter 12 can eliminate the effects of, for example, changes in the wall thickness of the bottle 11i and long line patterns in the circumferential direction. In other words, the wall thickness change itself constitutes a low frequency wave, and images due to long line patterns or joints in the circumferential direction enter the diode array 6 for a long time as if crossing the ζ. , that waveform constitutes a low frequency. 'Therefore, by cutting out low frequencies below a certain level in advance, the effects of inner thickness changes, lines, and streaks that are not defects can be eliminated.

Next, the output from the high-frequency F wave device 12 is input to the low-frequency T wave device 13. This low-frequency ν wave generator 13 can eliminate the influence of the satin pattern on the bottle surface. This is because in the case of a satin pattern, the images are small and generated at short intervals, so the output waveform of that portion constitutes a high frequency. The signal from the low frequency transducer 13 further passes through the absolute value circuit 14 and enters the comparator 15. This comparator 15
It compares it with the reference voltage from the reference voltage generator 16 and outputs an exclusion signal. By making the reference voltage from the reference voltage generator 16 variable, the displacement can be adjusted. The output from comparator 15 is input to OR circuit 17. Then, a final elimination signal is sent from the OR circuit 17 to the elimination device 18 for the glass bottle 1. In the above, the arithmetic circuit 11, high frequency
wave generator 12, low-frequency ν wave generator 13, absolute value circuit 14, comparator 1
5. Since one processing system 0 constituted by the reference voltage generator 16 is constructed for every four unit elements, the required processing system d is equal to η, where n is the number of unit elements.

This means that the equipment 11.12゜13 that constitutes the processing system O
, 14, 15, and 1'6 each.

Furthermore, for example, if the total number of amplifiers used in each device in each processing system 0 is m, in the present invention, Hm-rv,
In other words, the number of amplifiers can be reduced by 3'' compared to the conventional 4. This value of 3ny is extremely large. After all, the present invention can significantly reduce the number of instruments. By using switching elements, it is also possible to process the path of the high-frequency wave device 12 in a single row, for example in Figure 15.However, in this case, it is necessary to use a high-sensitivity device with a quick response. There is.

The present invention has the above-described structure, and uses a diode array in which a plurality of unit elements with a narrow dead band width are arranged, and an analog arithmetic processing circuit between adjacent unit element outputs is provided, so that the number of required unit elements is small. I'll try it. In this case, the influence of the dead zone can be more effectively prevented by forming each unit element into a parallelogram or by forming it in double rows. Furthermore, by performing analog arithmetic processing between the outputs of adjacent unit elements, it is possible to prevent the effects of dead zones and to significantly reduce the number of required equipment such as the number of amplifiers.

[Brief explanation of drawings]

Figures 1 to 4 are front views showing the company's conventional example, Figures 5 and 6.
Figure 5 is a plan view and Figure 6 is a front view of an example in which the apparatus of the present invention is used to inspect the body of a glass bottle. FIG. 7 is a front view showing an example in which the device of the present invention is used to inspect the bottom of a glass bottle, FIGS. t1g8 to 14 are front views showing examples of the diode array according to the present invention, and FIG. This is a circuit diagram showing the mechanism. l... Glass bottle, 2... Diffusion plate, 3... Lens system, 4... Image forming surface, 5... Real image, 6... Tiode array, 7... Dead zone, M8, A ,...B,”IB,
... Unit element, H9 engineering, J, Ni, II, M... Defect image, Applicant Yama'mura Glass Co., Ltd. Figure 1 Figure 2 Figure 3 Figure 4 Mouth Mouth Roro Roro Figure 5 Figure 7 Figure 8 Figure 9 Figure 13 Figure 2

Claims (1)

[Claims] (1) The object to be inspected is irradiated with diffused light, the transmitted light is imaged using a lens system, and the optical image is converted into an electrical signal by a photoelectric converter placed on the imaging plane. A device for determining the quality of an object to be inspected, which uses a diode array in which vI number of unit elements with a narrow dead band width are arranged as the photoelectric converter, and is provided with an analog arithmetic processing circuit between adjacent unit element outputs. A defect detection device for a transparent body (2) The diode array uses parallelogram unit elements, and the corresponding sides of each unit element are aligned on the same straight line and arranged in a row. A defect detection device for a transparent body according to claim 1 (3) A defect detection device for a transparent body according to claim 1 in which the diode array has unit elements arranged in two rows (4) The diode array has a parallelogram shape. A defect detection device for a transparent body according to claim 3, in which unit elements are used, and corresponding sides of each unit element are aligned on the same straight line. (5) A patent in which the positions of the unit elements are shifted between two rows. Transparent body defect detection device (6) according to claim 3 or 4, wherein analog calculation processing is performed for each output of two upper and lower adjacent unit elements. A defect detection device for a transparent body according to any one of claims 3 to 5, which performs analog calculation processing between outputs of near 1liI2 units on the left and right. Detection of defects in a transparent body according to any one of claims 3 to 5, which is performed for each element without duplication Claim 8 characterized by
Transparent object defect detection device described in section