JPS6048578A - Pattern detector - Google Patents

Abstract

PURPOSE:To prevent the noise produced by the regular reflection of a fine grain by illuminating a object right from above, giving photoelectric transducing and amplification to the same area to detect reflected light at two areas of an upper oblique side and synthesizing two signals after shading correction and a binary coding. CONSTITUTION:A sheet containing a pattern 3 is put on a table 14 and shifted horizontally by a motor 13 and via a ball screw 28. Then the sheet is irradiated right from above by an optical system consisting of a light source 15, a condenser lens, a half mirror 17 and a filter 18. The point of irradiation forms images to two sensors 20a and 20b through image forming lenses 19a and 19b having the same magnification ratio and set at the upper oblique side. Both sensors 20a and 20b and a motor driving circuit 27 are controlled synchronously with each other by a sensor driving circuit 21. The outputs of sensors 20a and 20b are amplified 22a and 22b, and the shading is corrected 23a and 23b for the correction of variance of both illumination and sensitivity and for the binary coding 24a and 24b. The outputs of both binary coding circuits are fed to an AND circuit 25 to produce an output signal. Thus it is possible to erase the regular reflection of metallic fine grain by the difference of timing. In such a way, the noise level can be lowered.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a method for detecting or inspecting a pattern formed on a plane, and in particular for metal particles having a surface that specularly reflects light. The present invention relates to a pattern detection device suitable for faithfully detecting a pattern shape formed of fine particles.

[Background of the invention]

Conventionally, methods for detecting patterns include 1) bright field (
There were 2) methods using dark-field (oblique) illumination, and 3) methods using transmitted illumination.

In the method using bright field illumination, as shown in FIG. The illuminated pattern 3 is then detected almost vertically upward. According to such a configuration, since the illumination direction and the detection direction almost match, a bright optical image can be obtained.

Generally, in photoelectric conversion elements such as two-dimensional image sensors and linear sensors, the signal-to-noise ratio (S ratio) is proportional to the product of the amount of incident light and the time required to detect one screen (12 inches for linear sensors). Therefore, according to a method using bright field illumination that can obtain a bright optical image, it becomes possible to detect a pattern image signal with a high S/N ratio at high speed.

However, when this method is used to detect a pattern formed by fine metal particles with specular reflection surfaces, as shown in Figure 2, the background becomes a pattern! If the pattern is made of a material that is detected brightly, the pattern may partially shine, as shown as bright spot noise 5, or conversely, the background may be detected darker than the pattern, as shown in Figure 3. If made of material, as shown as scotoma noise 6,
The pattern is detected partially dark, making it difficult to distinguish it from a pattern with fine holes. therefore,
When this method is used to inspect the shape of a pattern or the holes inside the pattern, there is a drawback that the reliability of the inspection is significantly reduced.

Further, in the method using dark field illumination, as shown in FIG. 4, a parabolic concave mirror 7 or the like is used to illuminate the pattern in an oblique direction, and the pattern is detected almost vertically upward. With this method, when there is a step 8 between the pattern and the background as shown in FIG. 5, the step part can be detected particularly brightly, so it is useful when detecting or inspecting the outline of the pattern or the unevenness of the surface. is effective, but as in the case of bright field illumination, it is not suitable for detecting patterns formed of materials with specular reflection surfaces, such as fine metal particles.

In the method using transmitted illumination, as shown in FIG. 6, the object on which the pattern 3 is formed is illuminated from the back side, and a silhouette image of the pattern is detected. This method has the ability to correctly detect the shape of a pattern even if it is formed of a material with a specular reflection surface such as fine metal particles, but if the butter 7 is formed on both the front and back sides, it may be difficult to identify it. Furthermore, if the light transmittance of the material forming the background of the pattern is low, only a dark silhouette image can be detected, and a high-speed, high-S detection image cannot be obtained.

[Purpose of the invention]

An object of the present invention is to provide a device that can faithfully and quickly detect a pattern shape formed by fine particles such as metal particles having a surface that specularly reflects light without being affected by specularly reflected light from the pattern surface. There is a particular thing.

[Summary of the invention]

In order to achieve the above object, a pattern detection device according to the present invention includes a means for illuminating an object, a means for converting an optical image of the illuminated object into an electrical signal, and a pattern detection device formed on a surface of a planar object. In the device for detecting a pattern, the illumination means is arranged and configured to illuminate the object from the side where the pattern to be detected is formed, and the pattern to be detected of the object is similarly formed at the same location of the illuminated pattern. two sets of imaging means that detect the pattern from two different directions and convert the brightness of the optical image of the pattern into electrical signals, and synthesize the corresponding parts of the two electrical signal patterns obtained from the imaging means. The gist is to have a means for synthesizing.

In an advantageous embodiment of the invention, the imaging means is a linear sensor, and the synthesizing means includes means for binarizing each of the two electrical signals, and forming a logical product or a logical sum of the obtained binary signals. or means for outputting a value corresponding to the larger or smaller of the two electrical signals.

The present invention has been made based on the following findings of the present inventors.

The following explanation will be limited to patterns formed of fine metal particles. Of course, the same applies to fine particles other than gold nine.

Here, we will reconsider the detection method using bright field illumination.

FIG. 7(&) shows an example of a cross section of a pattern formed of fine metal particles, and FIG. 7(b) is an enlarged view of a portion surrounded by a circle. As shown in the figure, the shape of the metal fine particles 9 generally consists of several planes (open planes 59) 10. Therefore, the direction of specular reflection of the illumination light on the surface of each fine particle follows the law of reflection: angle of incidence - angle of reflection. FIG. 8 shows this state, where 11 represents illumination light and 12 represents reflected light. Since the normal direction of the arm opening plane of the fine particles exposed on the pattern surface is completely random, the illumination light irradiated from any direction on the pattern will be specularly reflected in a completely random direction at each point on the pattern. do. Therefore, if the background is made of a material that is detected brighter than the pattern, the specularly reflected light 75: incident on the entrance pupil of the imaging lens of the detection optical system will be detected as shining on the pattern, resulting in This becomes a type of noise (hereinafter referred to as bright spot noise) on the detected image (FIG. 2). In addition, if the background is made of a material that is detected darker than the pattern, conversely, points on the pattern where reflected light does not enter the entrance pupil of the imaging lens of the detection optical system will be detected darkly, and as a result, this will also be the case. A type of noise on the detected image (hereinafter referred to as scotoma noise).

) (Figure 3).

Considering that the specularly reflected light of each point can only be detected from a certain direction, bright spot noise on the pattern will not appear at the same position in detection images from two different directions. . If the numerical aperture of the detection optical system is increased, in most cases dark spot noise will not appear at the same position on detection images from two different directions. therefore,
By combining detection images from two directions in a manner that cancels bright spot noise or dark noise, it becomes possible to detect a pattern without bright spot noise or dark spot noise. FIG. 9 schematically represents this, in which (a) is the detected pattern and (b) is the detected pattern.
and (C) are detected images from two different directions, (d
) is a composite image.

[Embodiment of the Invention] FIG. 10 shows an embodiment of the present invention. This embodiment is a pattern detection device particularly suitable for inspecting patterns of printed circuit boards such as green sheet patterns. In this example,
Pattern 3 is a white sheet (
For example, the butter 7
The purpose is to output the shape as a binary signal without being affected by bright spot noise.

The sheet is placed on a table 1 that can be horizontally fed by a motor 13.
4, and is illuminated from almost vertically above by an illumination device consisting of a mercury lamp 15, a mercury lamp lamp 29, a capacitor 16, a mercury lamp 2-17, and a filter 18. Of course, the light source is not limited to the mercury lamp 15, but other materials such as a halogen lamp may also be used. Further, in this case, it is not necessarily necessary to bend the nine-way path by the mirror 17. Filter 1
8 is used for the purpose of cutting wavelength components that have a significant effect on the material of the sheet or pattern or the detection optical system. Although it is not necessary to install an infrared absorption filter, good results are often obtained by using an infrared absorption filter to avoid the harmful effects of infrared rays. In this case, it is more effective to use a cold mirror (infrared transmitting mirror) as the mirror 17. The pattern thus illuminated is imaged onto two linear φ sensors 20a, 20b by imaging lenses 19a, 19b having the same magnification. The magnification is determined by the ratio between the required detection resolution and the image pitch of the linear sensor. The arrangement of the linear sensors 20a and 20b is such that optical images of the same magnification can be detected as described above, and the same location on the sheet is located for both sensors 2.
The images are formed on 0a and 20b. in this case,
The reading directions (scanning directions) of the signals of the linear sensors 20a and 20b are the same as shown by arrows J and a2 in FIG.

These two linear sensors 20a, 20b tt
They are operated synchronously by one sensor drive circuit 21.

Now, the analog image signals obtained in this way are input to amplifiers 221L and 22b, respectively, and the DC offset and amplitude are corrected. Next, the shading correction circuits 23a and 23b correct the p1 illumination unevenness and the sensitivity unevenness of each pixel of the sensor. The shading correction circuits 23a and 23b may be omitted if the effects of the two types of unevenness mentioned above can be ignored.

Various systems have been devised for the shading correction circuits 23a and 23b K, for example, the method disclosed in Japanese Patent Laid-Open No. 57-35
This is the circuit shown in No. 721. Next, binarization (b)
The signals are binarized using paths 24a and 24b. There are two methods for binarization: a method using a fixed threshold value and a method using a floating threshold value, and either method may be used. Next, the two binary signals generated in this way are synthesized using the AND circuit 5. Note that this means that the polarity of the circuit from the sensor to the binary signal output is positive. , assuming that the bright part is logic I (and if the polarity of the circuit is negative, then N
Uses an OR circuit. In the binary pattern signal obtained in this case, the pattern portion becomes logical. Therefore NOT
A circuit may be used to invert the signal. The motor 13 that horizontally moves the table is driven by a motor drive circuit 27, and the drive speed is determined by the scanning speed of the linear sensor, the reduction ratio of the ball screw path, and the detection resolution in the feed direction. The operation of the present embodiment described above will be explained with reference to FIG. 11 for the butter shown in FIG. 9. 1st
In Figure 1, (al) (A2), (bl) (b
2) + (cl) (c2) are linear sensors 20a, respectively.

20b1 amplifier "2a r 2zb, shape ink correction circuits 23a, 23b, binarization circuit 24a,
At the output of 24b, (al) (bl) (cl) (
dl) is the signal along Al-A2 in Fig. 9, (A2)
(b2) (c2) (d2) are signals along 81-82. First, as shown in Figure 11 (al) (A2), the DC offset and amplitude of the sensor signal are generally different for two different sensors, so these are corrected using an amplifier, as shown in (bl) and (b2). signal.

Furthermore, as shown in (cD (c2)), shading correction circuits 23a and 23b are used to remove signal undulations, and threshold values THI and TH2 are used to binarize Q(
cJXl! 2) So? ``!,!, Although expressed as a signal log, depending on the method of the shape ink correction (b) paths 23a and 23b, it may be a digital signal.
In the above, a method using a fixed threshold value was exemplified, but as described above, a method using a floating threshold value may also be used. (dl) (d2) are the binary signals obtained in this way. As mentioned above, the two sensors are detecting patterns at exactly the same location, so by taking the AND of the signals (dl,) and (d2), the bright spot noise part can be removed as shown in (ri). As described above, according to this embodiment, it is possible to obtain a binary image signal with a pattern free of bright spot noise with a relatively simple configuration. Therefore, the dimensional accuracy between the sensors in the pixel direction is extremely high, and as a result, alignment of the sensors becomes easy.

2. This example can be applied to a black pattern on a white sheet, but conversely, if a light-colored fine metal particle pattern is formed on a dark-colored sheet, this example can be modified, By replacing the AND circuit with an OR circuit, it is possible to construct a device that can eliminate the dark dots.

Next, a second embodiment will be described using FIG.

The table system, illumination system, and detection system are exactly the same as in the previous example. In the process of signal processing and synthesis, amplification and shape ink correction are performed in exactly the same manner as in Example 1, and then the minimum value circuit 30 outputs the smaller value of the two signals. The output becomes a bright and dark image signal 3J of a pattern from which bright point noise has been removed. Furthermore, by binarizing this signal using a binarization circuit array, a pattern binary image signal 26 can also be generated at the same time. As mentioned in the previous example, the binarization (b) path 24 can be used with either a fixed threshold method or a floating threshold method.

The operation of this embodiment will be explained in detail with reference to FIG. Note that this figure shows the pattern in Figure 9 as an example.
al) (A2) f'l: Seph f 20a,
20b, (bl) (b2) increased mf522g, 22
b, (cl) (02) is the shading correction circuit 2
3g, 231), (d) is the minimum value circuit 3o, (e)
is the output of the binarization circuit Sae. Also, (al) (bl
) (cl) is the signal along Al-A2 in Fig. 9, (A
2) (b2) (c2) is the signal along Bl -82. (al) (bl) (cl) (LL2) (b
2) (c2) is exactly the same as the operation of the previous example explained using FIG. (d) is the result of outputting the smaller value of the signals (cl) and (c2) by the minimum value circuit 3o, and a bright/dark signal with a pattern from which bright point noise has been removed is output. (d) The appearance of the blemishes is displayed as if it were an analog signal, but ((II,) (c2) is a digital signal using the method of shape ink correction (b) as described in the previous example. Therefore, (d) may be a digital signal.Also, by binarizing this signal using a fixed threshold TH, the pl2 transmission signal (e)
can be taken out. Although τ is not shown in the figure, as described above, binarization using a floating threshold may be used.

In this embodiment, as in the previous example, a diamond shape is considered for the case where a light-colored metal fine particle pattern is formed on a dark-colored sheet. By changing the fD minimum value (b) circuit to a maximum value circuit, it is possible to construct a device that can remove dark spot noise.

Next, a third embodiment will be described using FIG. 14. This is a modification of the first embodiment in terms of the sensor and the opening system, but the other configurations and operations are completely the same. The sensor has photomultiplier 'W 32a, 32b
Use. Since the photomultipliers tH32a, 32b are point sensors, the galvanometer mirrors 33n, 3
Set meal in a linear manner using 3b. Galvano mirror 33a
, 33b are synchronously swept in the directions bl and b in FIG. 14 using the synchronous control circuit 35. 36a, 36b indicate power supplies for the photomultiplier tubes 3za, 321+. The detection point on the pattern is set to be the same point for the two sensors by adjusting the position of the detection system, which is exactly the same as in the previous two examples. As in the previous two examples, a mercury lamp or the like can be used as the light source, but since the sensitivity of the photomultiplier tube is very high, a light source with low brightness such as the light emitting diode 31 can be used. Therefore, in this embodiment, in addition to the first effect of the first embodiment, it is possible to detect a pattern with high sensitivity even when the pattern is weak against strong light such as heat or ultraviolet rays, and it is also possible to detect a pattern with high sensitivity. This has the effect of reducing power consumption and heat generation.

Next, a fourth embodiment will be described using FIG. 15. This example is an example modified in the same way as the second example was modified from the first example to the third example,
The configuration and operation of the sensor and illumination system are exactly the same as in the third embodiment, and the operation and configuration of other parts are exactly the same as in the second embodiment. This embodiment has the same effect as the third embodiment in addition to the effect of the second embodiment in that bright and dark image signals and two-channel signals with a pattern from which bright spot noise has been removed can be obtained.

Also in the third and fourth embodiments, as in the first and second embodiments, dark spot noise occurs when a light-colored metal particle pattern is formed on a dark-colored sheet. Of course, it is possible to consider a modification to the detection device that eliminates this.

〔Effect of the invention〕

As explained above, according to the present invention, the shape of a pattern formed of fine particles such as metal particles having a surface that specularly reflects light can be detected with low noise. This has the effect of making inspection possible.

Furthermore, since it is basically a modification of the bright field illumination method, a bright optical image can be detected and high-speed pattern detection is possible.

Fig. 3 is a diagram showing noise in a detected image of a metal particle pattern, Fig. 4 is an explanatory diagram of the dark field illumination method, Fig. 5 is a perspective view of a pattern with steps, and Fig. 6 is an explanatory diagram of the transmitted illumination method. , FIG. 7 is a cross-sectional view of butter made of fine metal particles and an enlarged view of a part thereof, FIG. 8 is a diagram showing the direction of light reflection on the pattern, and FIG. 9 is a detailed explanation of the present invention. FIGS. 10, 12, 14, and 15 are block diagrams showing configurations of pattern detection devices according to four different embodiments of the present invention, and FIGS. 11 and 13. 6 are pulse waveform diagrams for explaining the operation of the apparatus shown in FIGS. 10 and 12, respectively.

1... Half mirror 12... Illumination system, 3... Pattern, 4... Detection system, 5... Bright spot noise, 6...
Dark point noise, 7... Parabolic concave mirror, 8... Step, 9.
...Metal fine particles, 10...Labor surface, 11...Illumination light, 12...Specular reflection light, 13...Motor, 14...
Table, 15... Mercury lamp, 16... Container 9
1/lens, 17...mirror, ]-8-...filter,
19a, 19b...imaging lens, 20a, 2
0b...Linear sensor, 21...Sensor drive circuit, 22a, 22b...Amplifier, 23g.

23b...Shape ink correction circuit, 24, 24a
, 24b...binarization circuit, 5...AND circuit,
G... 2 transmission signal output, γ... Motor drive circuit, A...
...Ball screw, 29...Mercury lamp power supply, concave...Minimum value circuit, 31...-Brightness image signal, 32IL, 32
b...Photomultiplier tube, 33a, 33b...Calvano mirror, ah...Light emitting diode, 35...Synchronization control circuit, 35a, 36b...Photomultiplier tube power supply agent Patent attorney Masami Akimoto Fig. 1 Fig. 2 Shoulder: 3 Fig. 4'';::, 5 Fig. F1' ↓ Fig. 6 Rainbow 8 Fig. 2.79 (a) 5 (C) Fig. 10 6 Fig. 11 Fig. 12 661 Figure 13; Figure 14

Claims (1)

[Claims] 1. An apparatus for detecting a pattern formed on a surface of a planar object, comprising means for illuminating an object and means for converting an optical image of the illuminated object into an electrical signal, The illumination means is arranged and configured to illuminate the object from the side on which the pattern to be detected is formed, and illuminates the same part of the illuminated pattern from two different sides of the object on which the pattern to be detected is formed. It has two sets of imaging means for detecting from two directions and converting the brightness of the optical image of the pattern into an electrical signal, and a synthesizing means for synthesizing corresponding parts of the pattern of two electrical signals obtained from the imaging means. Characteristic pattern detection device. 2. The pattern detection device according to claim 1, wherein the imaging means is a linear sensor. 3. Claim 1, wherein the synthesizing means comprises means for binarizing each of the two electric signals, and means for forming an AND or OR of the obtained binary signals. The pattern detection device described. 4. The pattern detection device according to claim 1, wherein the combining means is means for outputting a value corresponding to a larger value, a smaller value, or a smaller value of two electric signals.