JPS61256237A - Defect inspection for cyclic pattern - Google Patents

Abstract

PURPOSE:To enable discrimination of a defect in monochromatic manner, by performing a detection processing based on data as obtained by subtracting image data after the displacement of a cyclic pattern from that before the displacement thereof to inspect the cyclic pattern efficiently at a high accuracy. CONSTITUTION:A pattern 6 to be inspected is lighted by a transmission lighting section made up of a DC power source 9, an incandescent lamp 8 and a diffusion plate 7 and the inspection area is photographed with a TV camera 1. An image processor 2 converts the output signal of the TV camera 1 into digital from analog to be a digital image data and various processings including addition and subtraction on the screen are performed with a frame memory and an arithmetic unit. A control section 3 controls the image processor 2 and a pattern shifting mechanism composed of an X-Y stage 10 and a driving section 5. Then, when subtraction is done between image data before and after the shifting of the pattern, changes in the video signal due to a unit pattern with no defect are cancelled by changes in the video signal due to shading or the like in a photography system to reduce to virtually zero so that the value varies locally at only the defective point. Thus, the defect can be detected based on this fact.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to shadow masks used in cathode ray tubes for color televisions, color separation filters for color imaging devices, color filters for liquid crystal display panels, mesh-like electrodes used in electron tubes, For VDT filters, mesh filters, rotary encoders, linear encoders, 7-otomasks for IC, Fresnel lenses, lenticular lenses Units with certain optical properties and shapes (hereinafter referred to as unit patterns) are arranged in the -dimensional direction, or Industrial products that are regularly and repeatedly arranged in two dimensions, or industries in which unit patterns are repeatedly arranged while their optical properties, shapes, and arrangement pitches in one and two dimensions gradually change. It relates to a method for automatically inspecting products for defects such as scratches, pinholes, black spots, and dust.

[Background of the invention]

Conventionally, defect inspections of industrial products such as those mentioned above have been carried out visually using the naked eye or a microscope.
Inspecting a large number of products requires a large amount of manpower,
Furthermore, since it is a sensory test, there is a problem in that it lacks test accuracy and reliability.

In order to solve such problems, for industrial products with periodic patterns of equal pitch arrays, we investigated video signals obtained by microscopic imaging methods that sufficiently reveal the array units and defect shapes. Perform pattern recognition using
Alternatively, methods of detecting and inspecting defects by comparing signals obtained by similarly photographing a defect-free pattern have been proposed, and some examples have been implemented.
This method requires mechanical precision according to the size of the defect to be detected, and high-precision equipment is required to detect minute defects, which makes the equipment expensive and requires microscopic precision. Due to the large size of the image, the size of the screen that can be processed at one time is small, and it takes a lot of time to inspect the entire pattern to be inspected.

For products with periodic apertures, such as mesh electrodes for electron tubes, an optical 7-Lier transform spatial filtering method has been proposed that utilizes the diffraction phenomenon of light due to periodic patterns when irradiated with coherent light. However, although this method has excellent inspection speed and detection sensitivity, it requires creating a spatial filter for each pattern to be inspected, and requires a precise optical system, making the equipment expensive. Although defects can be detected, there are many problems such as the inability to determine the size relationship of the defect aperture with respect to a reference value, and a new inspection method has been desired.

[Purpose of the invention]

SUMMARY OF THE INVENTION The present invention aims to solve the above-mentioned problems, and to provide an inspection method that can efficiently and accurately inspect periodic patterns, and can discriminate between black and white defects.

[Summary of the invention]

In order to achieve this objective, the present invention provides an imaging field of view of a large area in which the unit pattern of the periodic pattern is not sufficiently resolved or not resolved at all.

The resulting video signal includes defect information, and the random noise reduction effect of the video signal is achieved by screen addition processing, which is an image processing method, and when the pattern to be inspected is moved in a predetermined manner, the video of the defect-free portion is The signal changes little before and after the pattern movement, and when you subtract the image data before and after the movement.

Changes in video signal No. 1g due to the unit pattern in the defect-free area and changes in the video signal due to shinning in the imaging system and dust in the optical system are erased, and the image data in the defect-free area becomes almost O, indicating that there is no defect. This method is characterized by the fact that it has been found that the value changes locally only in the area, and defects can be detected based on this.

[Embodiments of the invention]

Hereinafter, the present invention will be described in detail based on Examples.

FIG. 1 shows an embodiment of the periodic pattern inspection method according to the present invention, in which a periodic opening unit pattern 1 as shown in FIG.
This shows the method applied to an example of inspecting abnormalities in the open area of a pattern with a DC power supply 9.
The pattern to be inspected 6 is illuminated by a transmissive illumination unit composed of an incandescent lamp 8 and a diffuser plate 7, and the inspection area is photographed by the left camera of the TV. The image processing device 2 A/D converts the TV left camera output signal into digital image data, and performs various image processing including addition and subtraction on the screen at high speed using a frame memory and an arithmetic unit.

Reference numeral 3 denotes a control section that controls the pattern moving mechanism composed of the image processing device 2, the XY stage 10, and the drive section 5. In FIG. 2, 12 and 13 represent unit patterns having defects.

Next, the operation of detecting defects in pattern 8 according to this embodiment will be explained. In order to simplify the explanation, here, we will use a shooting condition 1 in which changes due to unit apertures in the video signal from the TV left camera can be ignored. and the direction of displacement is T
A case will be described in which the displacement distance of the pattern is an integral multiple of the pixel pitch in the scanning line direction of the V camera 11.

First, Fig. 3 (α) is a diagram showing the light transmission weight distribution on a straight line passing through a defective part of the pattern, and 14 is a diagram whose opening area is larger than that of the normal pattern 11, as shown in pattern 13 of Fig. 2. 15 shows a change in light transmittance due to a defect (hereinafter referred to as a white dot), and 15 shows a change in light transmittance due to a defect (hereinafter referred to as a black dot) whose opening area is smaller than that of the normal pattern 11, as shown in pattern 12)C in Figure 2. shows.

Next, FIG. 3(b) is a diagram showing a video signal scanned on the same line as FIG. Random noise generated in signal processing circuits,
and a local change 16 in the signal due to dust or the like attached to the optical system.

In addition, FIG. 3(C) is a diagram showing the result of screen addition processing performed by the image processing device 2. /T
This shows that it has decreased to .

Furthermore, Fig. 3(d) shows the result of screen addition processing by displacing the pattern, and as the pattern moves, signals due to defects on the pattern also move, but shading in the imaging system and dust in the optical system also move. This shows that the position of signal 16 has not changed.

Figure 3 (-) is from Figure 3 (C) to Figure 3 (d).
) shows the result of subtracting the image data in Figure 3 (C1, (d)) and the shading included in the data in Figure 3 (C1, (d))
The components that do not change due to the movement of the pattern as shown in 16 are erased, and the signal due to the change in light transmission of the pattern and
Only the reduced random noise component remains, resulting in
Signals due to defects appear at positions separated by a number of pixels according to the amount of movement of the pattern, with almost the same difference in value from the average value in the vicinity, and with reversed signs, and the order in which the signals are reversed depends on the type of defect (white dots). .

It can be seen from the black dots that it is reversed.

The above describes an example of subtracting between two screen image data that have been subjected to screen addition processing before and after pattern displacement. Even if the number is subtracted, the result is exactly the same, and the processing can be performed using only one side of the frame memory.

Therefore, if you observe the image data processed above on the TV monitor 4, you can easily recognize the defect because the brightness changes locally in the defective area, and furthermore, the brightness of the defective area is reversed with respect to the surrounding area. Defect types (white dots, black dots) can be identified in this order.

In addition, from this one-shot data, subtraction of neighborhood average values or differential processing, especially when the pattern movement is parallel movement, image processing that calculates the difference between corresponding pixels before and after the movement of one point on the pattern. When performing this, as shown in FIG. 3 V), the gradual change component of the image data is removed,
Defects can be automatically detected by comparison with a predetermined threshold.

By the way, the above describes a method for erasing signals other than those caused by defects by addition, pattern movement, and subtraction of screen data, but depending on the shooting conditions, changes in the video signal due to unit patterns may not be ignored, and in this case As explained above, when the pixel pitch is moved by an integral multiple, the unit pattern signal remains in the image data after the subtraction, as shown in FIG. 4(b), and minute defects cannot be detected.

Note that FIG. 4 (α) is data before subtraction.

However, in such a case, for example, for a periodic pattern with an equal pitch, rotate in an integral multiple of the array pitch, for a striped pattern, rotate in the stripe direction, for a concentric pattern, rotate at its center, rotate the unit pattern, and rotate the unit pattern on the circumference. For arrayed patterns, if a rotation of an integral multiple of the array angle is set as a condition for each pattern movement with the center of the array as the rotation center, the image data of each defect-free pattern will not change before and after the movement. , Figure 5 (α).

(,6+, the image data after addition and subtraction is shown in Figure 3 (
Results similar to those shown in -) are obtained, making it possible to detect minute defects.

Next, the movement distance of the pattern will be explained. FIG. 6 is a diagram explaining the relationship between the optical images of the defect patterns 17 and 18, the pixels P, and the relationship between the image data D after addition and subtraction.
α) means that the amount of pattern movement is an integer multiple of the pixel pitch, Figure 6 (
b) shows an example when it is not an integral multiple.

In Fig. 6(a), the image data D of the defective part is vertically symmetrical with respect to the neighborhood average value, but if it is not an integral multiple, then as shown in Fig. 6(a), the image data D of the defective part is
Since the rate at which changes in are distributed to adjacent pixels differs before and after pattern movement, the vertical symmetry with respect to the neighborhood average value is lost, which causes errors in automatic detection processing.

Therefore, when moving the pattern, it is preferable that the positional relationship between the defective image and the pixel that receives it is the same before and after moving the pattern. For example, when moving in the pixel arrangement direction, You will get better results if you set it up.

Therefore, if the change in the video signal due to unit butter/ can be ignored, the moving distance should be an integral multiple of the pixel pitch, and if it cannot be ignored, then the moving distance should be a common multiple of the pattern arrangement pitch and the pixel pitch. It is possible to satisfy both the conditions of erasing image data and making the moving distance an integral multiple of the pixel pitch.

By the way, the displacement of the pattern required in the above inspection method of the present invention is based on the essential condition of erasing the information of the defect-free pattern from the image data after addition and subtraction, and under this condition, if possible, the defect image This means that a better result should be obtained if the positional relationship between the pixel and the pixel that receives it is the same before and after the movement. Therefore, as long as these conditions are satisfied, the direction of movement does not necessarily have to coincide with the pixel arrangement direction.

Next, as an embodiment of the present invention, a method for reducing fluctuations in defect signal level caused by the positional relationship between a defect image and a pixel will be described.

FIG. 7(α) shows a state in which the image of the defect pattern 17 is located at the center of one pixel P, and FIG.

In this way, even if the defect is the same, the signal change due to the defect is mostly concentrated in one pixel as shown in Fig. 7 (α), and in the case of 4 pixels as shown in Fig. 7 (b). There is a difference of almost four times in the defect signal level between the two cases, and as a result,
This causes a problem that the reproducibility of defect detection is low. However, most of the dispersion of such defect signals to surrounding pixels is within a 3 x 3 pixel area centered on the defect image, and the influence on the outside can be ignored, so image processing By performing spatial filtering processing, which is often used in Changes in level can be reduced.

Next, as yet another embodiment of the present invention, a method for automatically detecting defects and determining types will be described.

In Figure 8 (α), the direction in which the pattern is displaced is the same as the pixel arrangement direction, the movement distance is twice the pixel pitch, and after performing screen addition processing, the pattern is moved to the right side in the figure, and the addition This shows an example of the image data obtained by subtracting the same number of frames as , where 20 shows a white spot defect, 21 shows a local change in image data due to a black spot defect, and the other parts are the arrangement of unit patterns. It represents changes in image data (low frequency changes) in response to gradual changes in pitch, size, etc.

Here, if we look at the image data of the defective part in detail, pixel A (22, z2'), which corresponds to the defect during screen addition, and pixel B (23, 23'), which corresponds to the defect during subtraction, are pixels that have defect information. The displacement distance of the pattern, that is, the number of pixels (2 pixels) corresponding to 2 pixels, appears apart, and each pixel data IA p IB p The pixel C corresponding to the midpoint.
If the neighborhood average value of (24, 24') is IC, then ■^-
IC and IB-IC have opposite signs and almost the same absolute value, and the order of the signs corresponds to the type of defect (white dot, black dot), and the absolute value of I, -IB is proportional to the size of the defect. It turns out that it does. It can also be seen that the difference between the average value of M and IB and IC is a sufficiently small ratio to the value of M-IB. Therefore, image data processing that yields s IA-IB, for example, I(7'
) =ICj-1')=I(j+1>, the 8th
As shown in the figure (Al, ■(]゛-1), I()°+
1) are pixels (25, 25') that are respectively ■^pIB
), the data shows the type and size of the defect.
Furthermore, since the aforementioned low frequency changes are also reduced, it becomes possible to detect defects and determine the type by comparing them with a certain threshold value.

However, at this time, as shown in Fig. 8 (h) K,
On both sides of the pixel to be detected as a defect, there are values of I with opposite signs.
Since half of the data (26, 26') of A-IB is generated, false defects will also be detected for defects with a signal level that is more than twice the threshold value, making it difficult to detect and type defects. This causes inconvenience in the judgment.

Therefore, the above ■A t IB p IC! In (IA
+IB) / 2-Using the fact that IC is less than a certain ratio with respect to IA-IB, for example, IC)"(l(j-1)"IO'+x) for @8 diagram (α) ) /
2-It>) (where I(7') is the neighborhood average value of Iσ+) to obtain Figure 8 (C1), and compare it with a predetermined threshold to determine which of the defect detection pixels including the false defect. , Said (■
By selecting pixels whose Io is less than or equal to a predetermined value, the false defects can be excluded, and stable defect detection and type determination can be performed.

Next, an example of defect inspection of a periodic pattern according to the above embodiment is shown in which the diameter of the unit pattern is 100 μm.
A periodic pattern with an arrangement pitch of 300μ corners as shown in Fig. 2 was illuminated with transmitted light as shown in Fig. 1, and a TV right camera using a camera tube was used to capture the image at 300 x 220.
The area of tm was photographed, and within a total of approximately 5 seconds of photographing time and image processing equipment, defects in which the aperture diameter of the unit pattern differed by 5 μm were automatically detected, and it was possible to determine the size of the aperture relative to the surroundings. .

Furthermore, as the TV photographing device and illumination method used in the present invention, any device can be used as long as it can obtain a video signal including defect information. TV cameras using trench elements, imaging equipment using image dissectors and flying spot tubes, X@TV imaging equipment, electron microscope imaging equipment, etc. can be used.The illumination methods include transmitted light illumination, transmitted darkness, etc. Field illumination, reflective dark field illumination, specular reflection illumination, etc. are used, and the illumination light properties include coherence, spectral characteristics, and polarization, all of which are designed to increase the S/N of the signal due to the defect to be detected. Alternatively, it can be carried out by selecting K so that the signal level due to defects that do not need to be detected is low.

〔Effect of the invention〕

As explained above, according to the present invention, minute defects in various industrial products having periodic patterns are photographed with a wide field of view such that unit patterns constituting the periodic pattern are not resolved. By processing the image data,
It becomes possible to automatically detect and determine the type (white dots, black dots), resulting in improvements in inspection accuracy, reliability, and efficiency.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the periodic pattern inspection method according to the present invention, FIG. 82 is an explanatory diagram showing an example of the periodic pattern, and FIGS. 3(α) to 4(α) ,
(A), Figure 5 (αl, (bl), Figure 6 (αl
, (h), FIG. 7 (b), and FIGS. 8 (α) to (C) are explanatory diagrams each showing the operation of the present invention. 1...TV camera, 2... Image processing device, 3... Control unit, 4... TV monitor, 5... Drive mechanism, 6... Test subject body,
7...Diffusion plate, 8...Lamp, 9...
...DC power supply, 10... Stage, 11.
... Unit pattern, 12 , 13 ... Defect, 14 , 15 ... Transmittance change due to defect,
16...Signal change due to dust etc. in optical system, 17
...Defect image, 27...Orchid value, 28...
...Neighborhood average value. To 1'/''@1 Figure 2-〇 〇〇〇-11 Figure 3 Figure 4 Figure 5 (0) CQ) Figure 6 (0) (b) Figure 7 Figure 8 f Procedural amendment (voluntary) June 1985≠date

Claims (11)

[Claims] (1) A method for inspecting local defects in a periodic pattern consisting of a repeating arrangement of unit patterns, comprising: an imaging means for decomposing the periodic pattern into pixels to obtain image data in units of frames; means for displacing the periodic pattern by a predetermined distance in a predetermined direction, and defect inspection based on image data obtained by subtracting image data obtained by imaging after the periodic pattern is captured from image data obtained by imaging before the periodic pattern is displaced. 1. A periodic pattern defect inspection method, characterized in that the periodic pattern defect inspection method is configured to perform processing. (2) In claim 1, the periodic pattern is characterized in that the periodic pattern is arranged such that at least one of the shape, size, and arrangement pitch of the unit pattern changes at a predetermined rate. Defect inspection method for periodic patterns. (3) In claim 1 or 2, the image data obtained by imaging before the periodic pattern is displaced and the image data obtained by imaging after the displacement are both image data spanning multiple frames obtained by imaging multiple times. A periodic pattern inspection method characterized in that the image data is obtained by adding each pixel data. (4) In claim 1 or 2, the imaging conditions by the imaging means are such that the influence of changes in each unit pattern appearing in data for each pixel of the imaged image data is sufficiently small that it can be ignored. periodicity characterized by being configured such that the predetermined direction coincides with the arrangement direction of the pixels, and the predetermined distance is an integral multiple of the pixel pitch. Pattern defect inspection method. (5) In claim 1 or 2, the predetermined direction is configured such that the predetermined direction coincides with the arrangement direction of the unit patterns, and the predetermined distance is an integral multiple of the arrangement pitch of the unit patterns. A periodic pattern defect inspection method characterized by: (6) In claim 1 or 2, the predetermined direction coincides with the arrangement direction of the unit patterns, and the predetermined distance is a common multiple of the arrangement pitch of the unit patterns and the pixel pitch of the image data. A periodic pattern defect inspection method characterized by being configured so that: (7) In claim 1 or 2, the periodic pattern is a striped pattern, the predetermined direction is the arrangement direction of each stripe of the striped pattern, and the predetermined distance is A periodic pattern defect inspection method, characterized in that the periodic pattern is configured to have a pixel pitch that is an integral multiple of the pixel pitch of the image data. (8) In claim 1 or 2, the periodic pattern is a concentric pattern, and the predetermined direction is a rotational movement direction with the center of the concentric pattern as a rotation axis. A periodic pattern defect inspection method characterized by: (9) In claim 1 or 2, the periodic pattern is a pattern in which unit patterns are arranged on a circumference, and the predetermined direction coincides with the arrangement direction of the unit patterns. 1. A method for inspecting defects in periodic patterns, characterized in that the predetermined distance is an integral multiple of an array pitch on a circumference of the unit pattern. (10) The periodicity according to claim 1 or 2, wherein the defect detection process is configured to include an addition process of neighboring pixel data of up to 3 x 3 pixels. How to inspect patterns. (11) In claim 3, the defect detection process detects, with respect to the image data obtained by the subtraction, a pixel corresponding to one point on the periodic pattern during screen addition.
, when the corresponding pixel during screen subtraction is B, and the pixel corresponding to the midpoint of pixels A and B is C, the difference between the neighborhood average value of this pixel C and the average value of the data of pixels A and B is Pixel A
A point that is sufficiently small compared to the difference between the data of pixels A and B, and the difference between the data of pixels A and B is at least a predetermined level is determined to be a defect, and the sign and absolute value of the difference between the data of pixels A and B are determined. A method for inspecting a periodic pattern, characterized in that the method is configured to recognize the type and size of a defect.