JPH0333605A - Comparing/testing method of patterns and apparatus thereof - Google Patents

Abstract

PURPOSE:To enable tests corresponding to a variety of kinds of items to be tested by setting the optimum threshold value for binarization for every kind of the items to be tested. CONSTITUTION:A threshold value calculating means 2a of this apparatus calculates the optimum threshold value for every kind of testing items. For example, if the testing item is of a kind not yet tested, a threshold value calculating command 10 is sent to the means 2a, where the threshold value is calculated. The calculated value is input to a threshold value memory means 26. On the other hand, if the testing item is a kind already tested, the already- obtained threshold value is input to the threshold value memory means 26. Thereafter, a binarizing means 25 binarizes a differential image of the detected images of the patterns of the testing item to test the item. Thus, it becomes possible to change the testing conditions in compliance with the complexity of the patterns of items of each kind, the finish or result of the items, the contrast of the detected images of the patterns of the items of each kind and the like, thereby enabling tests corresponding to many kinds of items.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention detects pattern defects by using a binarized image of a difference image between detected images of patterns that should be the same in an inspection object such as an LSI wafer. The present invention relates to a pattern comparison inspection method and apparatus for detecting 8- and the like.

[Conventional technology]

A conventional pattern comparison inspection method and apparatus for inspecting defects on LSI wafers and the like by comparing grayscale images of actual patterns has been constructed as shown in FIG.

FIG. 19 shows the constituent countries of an example of a conventional pattern comparison inspection method and apparatus. In FIG. 19, 3 is a stage, 4 is an imaging means, 6 is an image storage means, 7 is a difference image detection means, 8 is a binarization means, 9 is a threshold value storage means, 14 is a detected image signal, and 15 is a stored image signal, 16 is the difference image signal, 1
7 is a threshold value, and 18 is a binary image.

With this configuration, an image is detected by the imaging means 4 without moving the stage 3 on which the LSI wafer, etc. to be inspected is mounted, and this image is stored in the image storage means 6 for the comparison unit, thereby detecting a pattern. Image signal 14 and stored image signal 1
Get 5. Next, the difference image detection means 7 calculates a difference image signal 16 between these two image signals 14 and 15, and the binarization means 8
By binarizing this difference image signal 16 using the threshold value 17 stored in the threshold value storage means 9, a binary image signal 18 is obtained.
get. This binary image signal 18 is used to inspect patterns of LSI wafers and the like for defects.

FIG. 20 is an explanatory diagram showing an example of the image detected in FIG. 19. In FIG. 20, the same reference numerals indicate corresponding parts throughout the drawings, and 91 and 92 indicate defects. For example, a detected image 14 as shown in FIG.
Assume that a, a stored image 15a, and a difference image 16a are obtained, and that there is a defect 91 in this detected image 14a. Therefore, if we take the signal waveform 90a of the A-A' section that includes the defect 91 in the difference image 16a, the difference in shading of the signal waveform 90a becomes large in the defect 91 section, and here it is binarized using an appropriate threshold value Vl. Then, only the defect 91 can be detected, and such a binary image 18a can be obtained. The value v1 during this time is called a defect determination threshold. Further, it is assumed that a detected image 14b, a stored image 15b, and a difference image 16b as shown in FIG. 20 are obtained for a certain product Y, and in this case as well, there is a defect 92 in the detected image 14b. Therefore, defect 9 in the difference image 16b
If you take the signal waveform 90b of the B-B' part including 2, the defect 92
There is a large difference in shading in the signal waveform 90b between part 0 and other parts, and if the binarization is performed using the threshold V1 for type X, not only the defect 92 but also the normal part of the pattern will be detected. Erroneous detection occurs in which a defect is mistakenly detected.

For this reason, for example, it is necessary to binarize with a high threshold value V2,
In this way, only the defect 92 can be detected as in the binary image 18b obtained by binarizing using the new threshold value V2. However, in the case of the difference image 16a for product X, the threshold value V
If the image is binarized at 2, the defect 91 cannot be detected and will be missed. Therefore, it was necessary to set the defect determination threshold V to an optimal threshold for each type of product to be inspected. A related method for determining this type of binarization threshold includes, for example, Japanese Unexamined Patent Publication No. 62-282387.

[Problem to be solved by the invention]

The above-mentioned conventional technology does not take into consideration the method of setting the optimal threshold corresponding to other product types in the comparative inspection using the binarized image of the difference image of the detected image of the pattern that should be the same in the inspection object. Inspectors used to set thresholds for each product through trial and error11-, but the thresholds set through trial and error were not always optimal, making it impossible to conduct highly reliable inspections. There was a problem with the setup, which required a huge amount of time.

SUMMARY OF THE INVENTION An object of the present invention is to provide a pattern comparison inspection method and apparatus that can perform inspections compatible with various types of inspection objects.

Another object of the present invention is to provide a pattern comparison inspection method and apparatus that can perform accurate inspections corresponding to a wide variety of inspection objects.

Still another object of the present invention is to provide a pattern comparison inspection method and apparatus that can perform accurate inspection at high speed for a wide variety of inspection objects.

[Means to solve the problem]

In order to achieve the above object, the pattern comparison inspection method and apparatus according to the present invention are provided with a means for calculating an optimal threshold value for each product type to be inspected so as to correspond to a wide variety of products, and a threshold value for a product type that has not yet been inspected. After calculating the threshold value for each of the 12 items that have already been inspected, the difference image between the detected images of the pattern to be inspected is binarized and inspected. It is.

In addition, in order to achieve the above-mentioned other objects, the pattern comparison inspection method and apparatus according to the present invention provide a pattern comparison inspection method and apparatus that can measure the area, etc. of a difference image of a pattern detection image or its binarized image so as to perform accurate inspection compatible with a wide variety of products. A means is provided to calculate a threshold value that does not erroneously detect a normal pattern or overlook a pattern defect from the feature amount, and the inspection is performed by using the threshold value to binarize the difference image of the detected image of the pattern to be inspected. This is how it was done.

In addition, in order to achieve the above-mentioned and other objects, the pattern comparison inspection method and apparatus according to the present invention provide a plurality of threshold values and a difference image of a pattern obtained by the threshold values so as to perform accurate inspection corresponding to a wide variety of products at high speed. A means is provided to update the binarization threshold based on the relationship between the feature amount of the binarized image, the false detection rate of normal patterns, and the overlooking rate of pattern defects, and the threshold value that satisfies the update termination condition 3 is set. Inspection is performed by converting the data into binarized data.

[For production]

The above-mentioned pattern comparison inspection method and apparatus calculate and detect the optimal threshold value for binarizing the difference image of the pattern detection images every time the product to be inspected changes, thereby detecting the complexity of the pattern of each product, the degree of completion, etc. In addition, since the inspection conditions can be changed according to the contrast of the detected image of the pattern of each product type, it becomes possible to perform inspections compatible with a wide variety of products.

In addition, the pattern comparison inspection method and apparatus described above use information on the results of comparison inspection using a binarized image of a difference image of a detected pattern image, and detects characteristics such as the area of a pattern difference image or its binarized image. By binarizing and inspecting the difference image using an optimal threshold that does not mistakenly detect normal patterns from the amount or overlook pattern defects, it is possible to mistakenly detect normal patterns in the detected areas. 14- Since there are no defects included in the pattern or defects in the pattern overlooked, accurate inspection becomes possible.

In addition, the pattern comparison inspection method and apparatus described above detect pattern difference images based on the relationship between a plurality of threshold values, the feature values of the binary image observed using the threshold values, the false detection rate of normal patterns, and the missed rate of pattern defects. By converging while updating the threshold values, the range of optimal thresholds can be quickly narrowed down and optimized, compared to methods that simply update the thresholds in ascending order of values or in order of increasing values, or methods that update thresholds through trial and error. Since the threshold value can be determined quickly, the overall work time can be shortened and high-speed inspection can be performed.

〔Example〕

Embodiments of the present invention will be described below with reference to FIGS. 1 to 17.

FIG. 1 is a block diagram showing a first embodiment of a pattern comparison inspection method and apparatus according to the present invention. In Figure 1,
1a is an inspection device main body, and 2a is a threshold value calculation means. 3
It is possible to scan the LSI wafer to be inspected by moving the stage 15-. 4 is an imaging means that detects an image of the LSI wafer in synchronization with the movement of the stage 3;

This image is called a detected image 14. Reference numeral 5 denotes a control unit that controls the movement of the stage 3 and the imaging means 4 using a stage movement command 12, an imaging command 13, and the like. Reference numeral 6 denotes an image storage means which stores the detected images 14 in units of comparison, thereby making it possible to compare adjacent identical patterns. This stored image will be referred to as a stored image 15. 7 is a difference image detection means that detects a difference image 16 between the detected image 14 and the stored image 15. 8 is a binarization means, 9 is a threshold value storage means, and the binarization means 8 binarizes the difference image 16 using the threshold value 17 stored in the threshold value storage means 9, and outputs a binary image 18 as an inspection result. . By switching the switch 20, the detected image 14, the difference image 16, and the binary image 18 can be selected and the image 21 can be sent to the threshold calculation means 2a. In the threshold value calculation means 2a, the threshold value calculation command lO
The operation is started by sending an image capture command 11 to the inspection apparatus main body 1a, obtaining an image 21, calculating a threshold value using the image 2116-, and setting a calculated threshold value 19 in the threshold value storage means 9.

FIG. 2 is a flowchart of the pattern comparison inspection method shown in FIG. The operation of each part in FIG. 1 will be explained with reference to the flowchart in FIG. 2. First, the inspector checks whether the type of LSI wafer to be inspected is a type that has been inspected before (step 100), and if it is a type that has been inspected before, the optimum threshold value has already been calculated. This threshold 19
is input from the threshold calculation means 2a to the threshold storage means 9 (step 101), and an actual test is performed (step 105).
. Further, in the case of a product type that has not been inspected before, the inspector sends a threshold value calculation command 10 to the threshold value calculation means 2a to start threshold value calculation (step 102). These judgments (
Step 100) may not be performed by the inspector, but may be performed automatically by registering the product name and threshold value in the storage means. Next, the threshold calculation means 2a sends an image capture command 11 to the control section 5, and the control section 5 sends a stage movement command 12 to the stage 3, and at the same time sends an imaging command 13 to the imaging means 4. By capturing the images together, the image 21 is input to the threshold calculation means 2a (step 103). Next, the threshold value calculation means 2a calculates the threshold value using the image 21, and the image 21 can be the detection image 14, the difference image 16, the binary image 18, etc., and these can be switched with the changeover switch 20. It is. This calculated threshold value 19 is calculated by the threshold value storage means 9
(step 105).

After setting this threshold value, the actual inspection is performed in the following steps. First, the stage 3 is moved by the stage movement command 12 from the control section 5, and at the same time, the detected image 14 from the imaging means 4 is inputted by the imaging command 13, and this detected image 14 is sent to the difference image detection means 7, and the image storage means 6 Remember it. The stored image 15 is then delayed by a certain period of time.
By taking out the difference image 1, the difference image detection means 7 can obtain a difference image 16 between the detected image 14 in which adjacent patterns that should be originally detected and the stored image 15 are obtained.
6 is binarized by the binarization means 8 using the threshold value 17 stored in the threshold value storage means 9, thereby making it possible to obtain a binary image 18 in which only the defective portion of the pattern is detected (Step 8-Tub 105). . In this way, it is possible to perform inspections that correspond to a wide variety of inspection targets.

FIG. 3 is a first configuration side view of the threshold value calculation means 2a of FIG. 1. The threshold value calculation means 2a of the second construction drawing will be explained in detail with reference to FIG. In this first configuration example, the threshold value is calculated by inputting the detected image 14 as the image 21 from the inspection apparatus main body 1a. In FIG. 3, 22 is a selector which outputs the detected image 14 to one of images 30a and 30b in response to a select signal 35, and 23a and 23b are image storage means which respectively output images in which adjacent patterns that should be originally the same are detected. Remember. 24 is a difference image detection means that detects two images 31a.
, 31b, and 25 is a binarization means which stores the difference image 32 in a threshold value storage means 26 and outputs a binary image 33 which has been binarized using a threshold value 38. A labeling means 27 labels high-level pixels in the binary image 33 for each connected component (4-connected or 8-connected) and outputs a labeled image 34. 28 is a threshold value calculation control unit which starts and controls threshold value calculation according to the threshold value calculation command 10;
Reference numeral 29 is a threshold value updating means that calculates a threshold value 37 based on the maximum area 40.

19-39 is a maximum area calculation means which calculates the area by scanning each labeled pixel from the labeling image 34, and then outputs the maximum area 40.

Next, the operation of each part of the threshold value calculation means 2a will be explained. First, the threshold value calculation control section 28 receives the threshold value calculation command 10, and then sends the image capture command 11 to the inspection apparatus main body section 1, and inputs the detected image 14 to the selector 22.

At this time, by keeping the select signal 35 on the a side, the image 30a is stored in the image storage means 23a. Then, by sending the image capture command 11 again and moving the stage 3 in the inspection device main body 1, the previously stored image 30
The detected image 14 in which an adjacent pattern which should originally be the same as a is inputted to the selector 22, and at this time, by setting the select signal to the b side, the image 30b is stored in the image storage means 23b in the storage means 30b. This completes the image input (step 103 in FIG. 2). Next, an image 31a in which patterns that should be originally the same are detected by the difference image detection means 24.

A difference image 32 of 31b is detected. This difference image 32 is converted to a threshold value 38 stored in the threshold value storage means 26 by the binarization means 25.
0- to obtain a binary image 33, which is labeled by a labeling means 27 to obtain a labeled image 34, and further by counting the number of pixels with each label by a maximum area calculation means 39. Output area 40.

This maximum area 40 will be referred to as the mismatch area S here. The mismatch area S is one of the feature amounts of the difference image 16 (binary image 1g). Therefore, by changing the threshold value V, the mismatch area S
The result obtained is as shown in Figure 4.

FIG. 4 is an explanatory diagram showing the relationship between the threshold value V38 in FIG. 3 and the mismatch area S40. In Fig. 4, the unvalue area S(
Pixels) are in a monotonically decreasing relationship with respect to the threshold value ■ (gradation) in that as the threshold value ■ increases, the mismatch area S decreases or remains unchanged. Here, let the minimum detection area Sd be the minimum detection area for pattern defect determination, and find the minimum threshold value vc such that the mismatch area S is less than this minimum detection area 84. is the minimum threshold value that will not cause false detection in a normal area when there is no pattern defect in the image plate used for calculating this threshold value. Such a minimum threshold value V. The update threshold 37 is updated by the threshold update means 29 until the update threshold 37 is found, and the maximum area calculation means 39 repeatedly calculates the discrepancy area S40, thereby determining the minimum threshold that does not cause false detection in the normal part of the pattern.
. can be calculated, allowing accurate pattern inspection.

This minimum threshold Vc is for an image of a part of the inspection target, but by taking the maximum value among the calculated thresholds Vc obtained by calculating from several representative parts of the inspection target, It is possible to obtain a minimum threshold value vo that does not cause erroneous detection of normal parts even if a wide area of the area is inspected. For this purpose, it is sufficient to confirm in advance that there are no defects in all the locations used for threshold value calculation by visual inspection or the like.

Such a minimum threshold■. FIG. 5 shows the configuration of the threshold value updating means 29 for calculating .

FIG. 5 is a side view of the configuration of the threshold value updating means 29 in FIG. 3.

In FIG. 5, numeral 41 is a calculation unit that calculates an updated threshold value 37 using a formula that will be explained later. 42 is a threshold storage means;
Reference numeral 43 denotes a mismatch area storage means that stores all the threshold values 37 for calculating the mismatch area so far and the mismatch area 40 at that time. Reference numeral 44 denotes a termination determination unit which reads out all sets of these threshold values 45 and non-matching areas 46 with respect to these, determines whether or not termination conditions described later are satisfied, and ends if the termination conditions are satisfied. A signal 36 and a calculated threshold value 19 are output.

FIG. 6 shows the calculation threshold V of FIG. 3 (FIG. 5). 19 is a flowchart of the calculation method of No. 19. The procedure for calculating the calculation threshold c19 of the threshold updating means 29 in FIG. 3 (FIG. 5) will be explained with reference to FIG. 6. The method for calculating the mismatch area S40 has already been explained above, but this time, the threshold updating means 29 first calculates the initial value V1 of the threshold V38 (step 200), and then calculates the mismatch area Sn (step 201).
After determining whether the termination condition is satisfied (step 20
2) Calculate the threshold value Vn with n = n+1 (step 203), and sequentially change this threshold value V38 in various ways to calculate the discrepancy area S4 for each threshold value Vll V2..., Vn
The operation of calculating each mismatch area SI+S2+...tsn of 0 (step 201) is repeated until the end condition is satisfied (step 202). The termination condition at 23 (step 202) is as follows.

V, -VI11=1 and s, < s d and SWl
≧Sd where Q, m=1 to n, and the minimum detection area Sd.

Threshold V when this termination condition is met. =VA, the calculation threshold value V019 is obtained, and the threshold calculation is completed. At this point, the threshold value updating means 29 sends the end signal 36 to the threshold calculation control unit 28, and the calculation threshold value V is determined. 19 is sent to the inspection device main body section 1.

FIG. 7 is an explanatory diagram of the secant method and the combination of the secant method and the scissors method as calculation methods of the calculating section 41 of FIG. The calculation procedure of the calculation method of the calculation unit 41 of the threshold value updating means 29 shown in FIG. 5, which is a combination of the secant method and the scissors method, will be explained with reference to FIG. First, in the secant method, the initial value Vll V2 of the threshold ■ (gradation) is the difference image 16
The threshold value vn is calculated or inputted from a value such as an integral multiple of the standard deviation of , and from the threshold value V3, the threshold value vn is updated by the following formula.

24- This formula sets n=3, and as can be seen by looking at the point Pl*P2+P3 in Figure 7, the threshold is V
3 corresponds to determining the point S3 of the mismatch area S3. At this time, depending on the object to be inspected, it is also possible to take the logarithm of the mismatch area S as follows to speed up the convergence. Next, in addition to this process, by combining the scissors method, for example, the point ``n-1+P n-2'' in Figure 7 is
When a r 5n-z<S a, the following equation is used in which the subscript n-2 of equation (1) is changed to n-3.

This is the process of the scissors method, This process ? 5- Always use the optimal threshold ■. can be converged to.

FIG. 8 is an explanatory diagram of the scissors method of calculation performed by the arithmetic unit 41 of FIG. Arithmetic unit 4 of threshold value updating means 29 in FIG.
The calculation procedure of the scissors method of calculation method 1 will be explained with reference to Figure 8. We have already explained the calculation formula (3) of the scissor method in combination with the secant method, but the difference between the scissor method and the secant method is that the initial value of the threshold V is V I + V 2
By setting Vl sufficiently small and V2 sufficiently large, the optimum threshold V is set between the initial value V I r V 2 from the beginning, where S, ≧ Sd, S2 < Sd of the mismatch area S. The point is that calculation is performed using (3) under the condition that .

FIG. 9 is an explanatory diagram of the dichotomy of the calculation method of the arithmetic unit 41 of FIG. The calculation procedure of the bisection method used by the calculating section 41 of the threshold updating means 29 shown in FIG. 5 will be explained with reference to FIG. 9. In this dichotomy, by giving a sufficiently small Vl and a sufficiently large V2 as the initial value VIIV2 of the threshold V, as in the previous scissors method, the threshold V3 becomes the initial value V I + V, with S+>Sd and S2<Sd. Take the midpoint of 2 - + and Vs
= (VllV2)/2, 26 (vn-2+vn-1)/2, and S n < S
d and S n-2>S n-3 or Sn≧sd and S n-2<S n-3, then point P. -2 and P. −
The operation of exchanging 3 is repeated sequentially. This operation ensures that the optimal threshold value V is reached. converges to.

In the calculation method of the arithmetic unit 41 shown in FIGS. 7 to 9 above,
The secant method in Figure 7 has an optimal threshold ■. It has the advantage that it moves quickly near , but it has the disadvantage that it no longer converges when the slope approaches the horizontal due to the relationship between the threshold value V and the mismatch area S. The scissors method shown in Fig. 8 always has an optimal threshold value V. It has the advantage that it converges to , but it has the disadvantage that it converges slowly because it is necessary to have a large interval between the initial values V, , V2. Therefore, as explained in FIG. 7, by using a method that combines the second method and the scissors method, the optimum threshold value V is first determined by the second method. After moving quickly to the vicinity, the sandwiching method can be applied in a sufficiently limited range where the optimal threshold value Vc exists, so the threshold value can be changed more quickly than when changing the threshold value vn monotonically in descending order or in descending order. V. can be calculated.

FIG. 10 shows a second configuration side entrance of the threshold value calculation means 2a of FIG. 1. In this second configuration example, unlike the configuration example shown in FIG. 3, the threshold value 19 is calculated by inputting the difference image 16 of images in which patterns that should originally be the same in the inspection target are detected from the inspection device main body 1a. Therefore, a difference image storage means 60 is provided, but the selector 22 and image storage means 23a and 23b shown in FIG.
Since the difference image detecting means 24 and the difference image detecting means 24 are not required, there is an advantage that the structure of the threshold value calculating means 2a is simplified compared to the structure example shown in FIG.

The procedure for calculating this threshold value 19 is the same as that shown in FIG.

FIG. 11 shows a third component side entrance of the threshold value calculation means 2a of FIG. 1. In this third configuration example, unlike the configuration example in FIG. However, in this configuration, it is necessary to input the binarized image 18 every time the threshold value 19 is changed. In this configuration example, the selector 22 and image storage means 23a shown in FIG. 3 are used.

=28 23b, the difference image detection means 24, the binarization means 25, and the threshold value storage means 26 are not required, so the advantage is that the configuration of the threshold value calculation means 2a is simpler than the configuration example shown in FIG. be. The procedure for calculating this threshold value 19 is the same as that shown in FIG.

The twelfth is a configuration diagram showing a second embodiment of the pattern comparison inspection method and apparatus according to the present invention. In FIG. 12, 1b is the main body of the inspection device, and 2b is threshold calculation means.

61 is a coordinate calculation means that calculates the position based on the binary image 18 and the movement amount 64 of the stage 3! 7i! Calculate 62.

This configuration is almost the same as that in FIG. 1, but the difference is that the inspection device main body 1b has a coordinate calculation means 61 and a movement amount 64 of the stage 3, and the changeover switch 20 in FIG. 1 is not present. The internal configuration of the threshold value calculation means 2b is different, and the inspection command 63 is provided in place of the image capture command 1 shown in FIG.

FIG. 13 is a first flowchart of the pattern comparison inspection method of FIG. 12. The operation of each part in FIG. 12 will be explained using flowchart 1- in FIG. 13. First, the inspector checks whether the product to be inspected has ever been inspected (step 100), and if it has been inspected, the optimum threshold value has already been calculated, so this threshold value 19 is stored in the threshold storage means 9. Step 10 (2) is entered, and actual inspection is performed (Step 105). Furthermore, in the case of a product type that has not been inspected before, the inspector sends a threshold value calculation command 10 to the threshold value calculation means 2b to start threshold value calculation (step 102). These determinations (step 100) may not be performed by the inspector, but may be performed automatically by registering the product name and optimal threshold value in the storage means. The procedure up to this point is the same as the flowchart in FIG. Next, threshold value calculation means 2a
Then, an inspection command 63 is sent to the control section 5 to perform a trial inspection. In this trial inspection, an appropriate threshold value 17 is either set in the threshold storage means 9 or calculated. This trial inspection performs the same operation as the actual inspection (step 105), which will be explained below. First, the stage 3 is moved by the stage movement command 12 from the control section 5, and at the same time, the detected image 14 is inputted from the imaging means 4 by the imaging command 13, and this detected image 14 is sent to the difference image detection means 7 and stored in the image storage means 6. Then set it to 0. Then, after a certain period of time delay, memory image 1
5, the difference image detecting means 7 can obtain a difference image 16 between the stored image 15 and the detected image 14 in which adjacent patterns of the inspection object that should be the same are detected. This difference image 16 is binarized by means 8 and threshold storage means 9.
By performing binarization using a threshold value 19 stored in , a binary image 18 in which pattern defect candidates are detected is obtained. The coordinate calculation means 61 calculates this binary image 18 and the movement amount 64 of the stage 3.
The coordinates 62 of the pattern defect candidate are calculated and output based on , and the trial inspection ends here (step 106). Next, a visual check is performed, but in this visual check, the stage movement command 12 from the control unit 5 is determined based on the coordinates 62 of the defect candidate.
The stage 3 and the imaging means 4 are operated according to the imaging command 13, and an image of the defect candidate is captured, and the inspector observes this on a monitor or the like to determine whether it is a true defect or a false detection. Then, the erroneously detected coordinates 62 are sent to the threshold calculation means 2b. This visual confirmation may be performed by directly observing with a microscope or the like instead of using an image (step 107). Next, the threshold value calculation means 2b calculates the threshold value based on this false detection rate 68 standard 62, and the calculated threshold value 19 is set in the threshold value storage means 9.Step 10
8) Then perform the actual inspection (step 105)
. In this way, it is possible to perform inspections that correspond to a wide variety of products to be inspected.

FIG. 14 is an explanatory diagram showing the relationship between the threshold value V l 9 in FIG. 13 and the false detection rate f. In FIG. 14, the false detection rate f of pattern defect candidates is defined as the number of false detections per chip to be inspected, and this false detection rate f (pieces/chip)
When examining the relationship between V and the threshold value V (gradation), it becomes a monotonically decreasing relationship as shown in the figure. Here, the minimum threshold ■ that makes the false detection rate f smaller than a certain appropriate setting value fd. is determined as the optimal threshold value.

This minimum threshold value V. Threshold calculation means 2b for calculating
The configuration is shown in FIG.

FIG. 15 is a side view of the configuration of the threshold value calculation means 2b of FIG. 12. In FIG. 15, 28 is a threshold value calculation control section, 29 is a threshold value updating means, and 65 is an erroneously detected coordinate storage means that stores the erroneously detected coordinates 62 sent from the coordinate calculation means 61. 6
Reference numeral 6 denotes an erroneous detection rate calculation means which calculates an erroneous detection rate 68 from the erroneous detection coordinates 62 to 3 or the erroneous detection coordinates 67.

69 is a selector which detects incorrectly detected coordinates 6 by a select signal 70.
2 to the erroneous detection coordinate storage means 65 or the erroneous detection rate calculation means 6
Send to 6.

Next, the operation of calculating the threshold value shown in FIG. 15 will be explained. First, after completing the visual confirmation (step 107 in FIG. 13), the erroneously detected coordinates 62 are input from the inspection apparatus main body 1. Selector 6
At step 9, the erroneously detected coordinates 62 are sent to the erroneously detected coordinates storage means 65 by the select signal 70 and are stored therein. Next, the false detection rate calculation means 66 counts the false detection coordinates 62 and calculates the false detection rate f68. Next, the threshold value updating means 29 updates the threshold value 19 based on the false detection rate 68, and updates the threshold value 19 based on the false detection rate 68.
Set in the threshold storage means 9 of b. Here, the threshold calculation control section 28 sends an inspection command 63 to the inspection apparatus main body section 1b to inspect the same location as the trial inspection (step 106 in FIG. 13). Next, the coordinates 62 of the defect candidate as a result of the inspection are received, and the selector 62 sends the coordinates 62 by a select signal 70 to the false detection 1$ calculation means 66, where it is stored in the false detection coordinate storage means 65. A false detection rate f68 is calculated by comparing with a certain false detection coordinate 67. Next, the threshold value updating means 29 calculates a third value based on this false detection rate f68.
The same termination determination (step 202 in FIG. 6) as in the case of the mismatch area S40 in the figure (FIG. 5) is performed. If the termination condition is not satisfied here, the threshold value 19 is further updated and the same process is repeated. When the termination condition is satisfied in this way, the calculation threshold value V is such that the false detection rate f (pieces/chip) becomes equal to or less than the set value fd. 19 in the inspection device main body 1b to complete the threshold calculation process (step 108 in FIG. 13).
.

As described above, according to the embodiment shown in FIG. 12 (FIG. 15), the threshold value ■ does not cause false detection. Accurate inspection can be carried out with
You can check the cause of the failure and take effective countermeasures. Further, the configuration of the threshold value updating means 29 in FIG. 15 may be changed to the configuration shown in FIG. 5 by replacing the mismatch area storage means 43 with a false detection rate storage means and inputting the false detection rate 68. The operation of the threshold value updating means 29 in this case is also the same as above, and the calculation method of the calculation unit 41 of the threshold value updating means 29 is the method using the secant method and the scissor method in combination, the method using the scissors method alone, and the two methods.
Threshold value V in division method etc. can be calculated. Here, the threshold value V is calculated using the false detection rate f68 by the threshold value calculation means 2b. However, the threshold value ■o can be obtained by similar processing using other evaluation quantities.

FIG. 16 is an explanatory diagram showing the relationship between the threshold value VI9 in FIG. 13 and the defect overlooking rate M. In FIG. 16, the missed pattern defect rate M is defined as, for example, the number of missed defects per chip to be inspected, and the relationship between this missed rate M (pieces/chip) and the threshold value V (gradation) is examined as shown in the figure. is a monotonically increasing relationship. Here, an appropriate setting value M with a missed rate M,
The maximum threshold value V that is less than. is determined as the optimal threshold value. This allows almost all defects to be found, allowing all causes of defects to be checked and countermeasures to be taken.

However, in order to calculate this threshold value, it is necessary to know the coordinates of the defect in the pattern.
Trial inspection (step 1) as shown in flowchart 1~ in Figure 3.
06) and visual confirmation (step 107), or a test pattern in which defects are embedded may be prepared.

5. FIG. 17 is a second flowchart of the pattern comparison inspection method of FIG. 12. In FIG. 17, first check whether the variety has been tested before (step 100);
If the product has been previously inspected, the threshold value is input (step 101), and the actual inspection is performed (step 105). In addition, in the case of a product that has not been inspected before, threshold value calculation is started (step 102), an image is inputted (step 103), a threshold value is calculated from that image (step 104), and a trial inspection is performed using this threshold value (step 104). 106), then perform a visual check (step 107).
), calculate the threshold value from the false detection rate f (step 1.0
8) Perform actual inspection (step 105). In this case, both of the aforementioned threshold value calculation means 2a and 2b are required. According to this inspection method, the accuracy of the threshold value can be improved.

FIG. 18 is the third flowchart 1 of the pattern comparison inspection method of FIG. 12. In Figure 18, first check whether the variety has been inspected before (step 100).
If the product has been previously inspected, enter the threshold value (step 1.01) and perform the actual inspection (step 105). In addition, in the case of a product that has not been inspected before, start threshold calculation (step 102), input an image (step 103), calculate the threshold using this image (step 104), and perform a trial inspection using this threshold (step 1 ( 16) Next, perform a visual check (step 1.07), input the image of the location where the false detection found here has occurred (step 109), calculate the threshold value from this image (step 108), and perform the actual inspection. conduct(
Step 105). According to this inspection method, the accuracy of the threshold value can be similarly improved.

In the above embodiment, an example was explained in which the threshold value V is calculated using the mismatch area S as the feature amount of the difference image 16, but the invention is not limited to this, and the threshold value calculation may be performed using other feature amounts. For example, the difference image 1 may be used as other feature quantities.
6 (binary image 18), projection length, volume (sum of brightness of pixels whose brightness exceeds a certain value), etc.

〔Effect of the invention〕

According to the present invention, it is possible to carry out inspection by setting the optimum threshold value for binarizing the difference image of patterns that should be the same depending on the product to be inspected, so it is possible to handle a wide variety of products.

Furthermore, since inspection can be performed using an optimal threshold value that does not erroneously detect normal patterns, it is possible to improve the accuracy of inspection.

Further, since the threshold value can be updated and converged based on the relationship between a plurality of threshold values and the amount of false detections observed by the threshold value, there is an effect that the optimal threshold value can be calculated at high speed.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing a first embodiment of the pattern comparison inspection method and apparatus according to the present invention, FIG. 2 is a flowchart of the inspection method of FIG. 1, and FIG. 4 is an explanatory diagram of the relationship between the threshold in FIG. 3 and the mismatch area, and FIG. 5 is the configuration side of the threshold updating means in FIG. 3.
Figure 6 is a flowchart 1 of the threshold calculation method in Figure 3 (Figure 5), Figure 7 is an explanatory diagram of the secant method in the calculation unit in Figure 5, and Figure 8 is an illustration of the secant method in the calculation unit in Figure 5. Fig. 9 is an explanatory diagram of the dichotomy of the calculating section in Fig. 5, Fig. 108 is the second component survey of the threshold calculation means in the engineering drawing, Fig. 11 is the illustration of Fig. 1. FIG. 12 is a block diagram showing a second embodiment of the pattern comparison inspection method and apparatus according to the present invention, and FIG. Flowchart, Figure 14 is an explanatory diagram of the relationship between the threshold value in Figure 13 and the false detection rate, Figure 15 is a configuration survey of the threshold calculation means in Figure 12, and Figure 16 is the relationship between the threshold value in Figure 13 and the missed rate. Fig. 17 is an explanatory diagram of the second inspection method of Fig. 12.
18 is a third flowchart of the inspection method of FIG. 12, FIG. 19 is a block diagram showing an example of a conventional pattern comparison inspection method and apparatus, and FIG.
FIG. 9 is an explanatory diagram of the detected image example of FIG. 9;

Claims (1)

[Scope of Claims] 1. In a pattern comparison inspection method using a binarized image of a difference image between images in which patterns that should originally be the same are detected, optimization of binarization for each type of inspection subject is provided. A pattern comparison inspection method characterized by calculating and setting a threshold value. 2. Consisting of a means for detecting a pattern of the inspection object that should be originally the same, a means for storing the detected image, a means for detecting a difference image of the detected image, and a means for binarizing the difference image. A pattern comparison and inspection apparatus, characterized in that the pattern comparison and inspection apparatus is provided with means for calculating an optimum threshold value for binarization for each product type to be inspected, and means for setting the threshold value. 3. In a pattern comparison inspection method using a binarized image of difference images of detected images that should originally be the same pattern of the inspection target, from the feature amount of the difference image or binarized image obtained from multiple detected images. A pattern comparison inspection method characterized by calculating and setting an optimal threshold for binarization. 4. Consisting of means for detecting patterns that should be originally the same in the inspection object, means for storing the detected image, means for detecting a difference image between the detected images, and means for binarizing the difference image. In a pattern comparison inspection device, a means for calculating a feature amount of a difference image or a binarized image obtained from a plurality of detected images, a means for calculating an optimal threshold value for binarization from the feature amount, and a means for setting the threshold value. 1. A pattern comparison inspection device comprising: means. 5. In a pattern comparison inspection method using a binarized image of a difference image of images in which a pattern of the inspection target that should be originally the same is detected, a plurality of binarization thresholds and a binarized image obtained by the thresholds are used. A pattern comparison inspection method characterized in that an optimal threshold is set by updating a threshold for binarization based on a relationship with a feature amount. 6. Consisting of means for detecting patterns that should be originally the same in the inspection object, means for storing the detected image, means for detecting a difference image between the detected images, and means for binarizing the difference image. In a pattern comparison inspection device, a means for updating a binarization threshold based on a relationship between a plurality of binarization thresholds and feature amounts of a binarized image obtained by the threshold, and a means for determining completion of the update. and means for setting the updated threshold value. 7. In a pattern comparison inspection method using a binarized image of a difference image of images in which a pattern of an inspection target that should be originally the same is detected, a plurality of binarization thresholds and a binarized image obtained by the thresholds are used. A pattern comparison inspection method characterized in that an optimal threshold is set by updating a binarization threshold in relation to the area of the area. 8. In the pattern comparison inspection method according to claim 7, as a method of updating the threshold value, the area of the binarized image is in a monotonically decreasing relationship with respect to the threshold value, but when the threshold value becomes a certain set value. In the method of calculating the area value, calculate the area value for two threshold values, update the threshold value from the intersection of the straight line connecting them and the area setting value, and when the area exceeds the setting value, A pattern comparison inspection method characterized by using a sequential calculation method in which updates are made by connecting two points so that they always sandwich a set value. 9. The pattern comparison inspection method according to claim 7, wherein one of a secant method, a scissors method, and a bisection method is used as a method for updating the threshold value. 10. In the pattern comparison inspection method according to claim 7,
A pattern comparison inspection method characterized by using a combination of at least two of the secant method, the scissors method, and the bisection method as a method for updating a threshold. 11. Consisting of means for detecting patterns that should be originally the same in the inspection object, means for storing the detected image, means for detecting a difference image between the detected images, and means for binarizing the difference image. In a pattern comparison inspection device, a means for labeling a binarized image, a means for determining a maximum area of connected components in the label, a means for storing the maximum area, a plurality of thresholds for binarization, and the threshold value. means for updating the binarization threshold based on the relationship with the maximum area of the binarized image obtained by the method, means for determining the completion of the update, and means for setting the threshold at which the update has been completed. A pattern comparison inspection device characterized by: 12. In a pattern comparison inspection method using a binarized image of a difference image of images in which patterns that should originally be the same are detected, the false detection rate of normal patterns is calculated using information on the results of comparison inspection. A pattern comparison inspection method characterized in that a binarization threshold is set so that the false detection rate is less than a set value. 13. In a pattern comparison inspection method using a binarized image of a difference image of images in which a pattern of an inspection target that should be originally the same is detected, a plurality of binarization thresholds and a binarized image obtained by the thresholds are used. A pattern comparison inspection method characterized in that an optimal threshold is set by updating a binarization threshold in relation to a false detection rate of a normal pattern. 14. Consisting of a means for detecting a pattern that should be originally the same in the inspection object, a means for storing the detected image, a means for detecting a difference image of the detected image, and a means for binarizing the difference image. In a pattern comparison inspection device, a means for updating a binarization threshold based on a relationship between a plurality of binarization thresholds and a false detection rate of a normal pattern of a binarized image obtained by the threshold; A pattern comparison inspection device comprising: means for determining completion; and means for setting a threshold value at which the update has been completed. 15. In a pattern comparison inspection method using a binarized image of a difference image of images that detect patterns that should originally be the same on the inspection target, the pattern defect oversight rate is calculated based on the information of the comparative inspection results. , a pattern comparison inspection method characterized in that a binarization threshold is set so that the missed rate is less than a set value. 16. In a pattern comparison inspection method using a binarized image of a difference image of images in which a pattern of the inspection target that should be originally the same is detected, a plurality of binarization thresholds and a binarized image obtained by the thresholds are used. A pattern comparison inspection method characterized in that an optimal threshold is set by updating a binarization threshold in relation to a missed rate of pattern defects. 17. Consisting of a means for detecting a pattern that should be the same in the inspection target, a means for storing the detected image, a means for detecting a difference image of the detected image, and a means for binarizing the difference image. In a pattern comparison inspection device, means for updating a binarization threshold based on the relationship between a plurality of binarization thresholds and a missed rate of pattern defects in a binarized image obtained by the threshold, and a means for updating the binarization threshold; A pattern comparison inspection device characterized by comprising: means for determining, and means for setting the updated threshold value. 18. Consisting of a means for detecting a pattern that should be originally the same in the inspection object, a means for storing the detected image, a means for detecting a difference image of the detected image, and a means for binarizing the difference image. The pattern comparison inspection device is characterized by being provided with a means for calculating an optimum threshold value for binarization for each product type to be inspected, a means for storing an optimum threshold value corresponding to the product type, and a means for setting the threshold value. Pattern comparison inspection device. 19. means for detecting patterns that should be originally the same in the inspection object; means for storing the detected image; means for detecting a difference image between the detected images; means for binarizing the difference image; A pattern comparison inspection device comprising means for comparing the coordinates of a binarized image, characterized in that it is provided with means for calculating an optimal threshold value for binarization for each product type to be inspected, and means for setting the threshold value. Pattern comparison inspection device.