JPH01137376A - Pattern check system - Google Patents

Abstract

PURPOSE:To realize the alignment with high accuracy of patterns by obtaining the position where the discordant count value is minimized in both horizontal and vertical directions and obtaining the position shifting quantity among plural directions based on the previous shifting quantities. CONSTITUTION:Two patterns are temporarily stored in buffer memories 4 and 7 respectively and at the same time a position error detecting circuit 8 counts the discordant value to obtain the position where the count value is minimized in both horizontal and vertical directions respectively. In case plural such positions are obtained, the previous alignment result is applied in the due direction for acquisition of the position error. Based on this position error, a position error correcting circuit 9 obtains the reading position of the memory 7 and correct the position error. Thus, it is possible to secure the position matching between two patterns even though these patterns are set in just a single direction. In such a way, the patterns can be checked with high efficiency.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to pattern inspection of printed boards, etc.
This invention relates to a method for determining the quality of a pattern by comparing two patterns.

[Conventional technology]

Conventionally, there is a method of inspecting patterns by comparing two patterns, but since there is an error in the alignment of the two patterns, Kagotsu Idojikai Gobunshi Vol. J69-
C, A 4 (April 1986) pp, 585-3
Methods for detecting alignment g4 differences are known, such as those discussed in US Pat. In other words, we sequentially cut out local areas within the range where positional deviation occurs from one pattern, count the discrepancies between each pixel in the local area and the other 10,000 patterns over a certain interval, and find the position on the local area where the counted value is the minimum. We are looking for positional deviation.

[Problem that the invention seeks to solve]

The conventional technique described above does not take into account the number of positions where the count value is the minimum, and there is a problem in that a plurality of positions are determined in a portion formed only by vertical and horizontal patterns.

The purpose of the present invention is to solve the above problems. % accurate positioning means is provided.

[Failure to solve the problem]

The above purpose is to temporarily store each of the two patterns in memory, count the discrepancies in the same way as in the conventional technology, and independently look at the position where the counted value is the minimum in the horizontal direction and in the vertical direction. This is achieved by adopting the previous alignment result in that direction, reading it from the memory, and comparing and determining the two patterns.

[Effect]

After the two patterns are temporarily stored in a memory and the amount of positional deviation is determined, the two patterns are aligned from the memory and read out for comparison and determination. By doing this, there is a pattern in the horizontal or vertical direction, rather than a situation where only horizontal or vertical patterns exist, and therefore no misalignment in the horizontal or vertical direction can be created. Even if the positional shift changes to a state where it can be detected, the corresponding pattern exists in the memory, so it can be followed. Also,
The position where the count value is minimum is viewed independently in the horizontal and vertical directions. For example, when horizontal distortion occurs in the pattern shown in Figure 2, the range for counting mismatches is continuous for block L (no alignment, the unit of judgment is called a block) and block t+1, but the judgment ranges overlap. However, in block i, there is only one position where the minimum count value is shown, but in block a
In test 1-)-1, only vertical patterns exist in the count range, so there are multiple positions in the vertical direction where the count value is minimum. In such a case, if there is only a vertical pattern in the judgment range, there is no problem with vertical alignment. Therefore, if a pattern other than the vertical pattern exists in the overlapped part with block i as shown in Figure 2, the vertical Directional alignment becomes an asset. In the present invention, when there are multiple positions where the count value is minimum in the &C vertical direction, the previous alignment result, that is, block i, is adopted in the vertical direction, and the horizontal alignment is Tsuku i -1
-j, making it possible to align both horizontally and vertically.

〔Example〕

An embodiment of the present invention will be described below using Figure @1, taking as an example a comparative calibration with a standard pattern without defects created from design data or the like. In FIG. 1, the video signal of the imaging device [1] is converted into a 21-digital signal by a binarization converter 2 to become a detection f pattern signal 3, which is input to a buffer memory 4. On the other hand, the Irs quasi-pattern stored in the standard pattern memory 5 is synchronized with 1, read out as the standard pattern signal 6, and inputted to the backup memory 7. In addition, with these processing and distortion,
6 and 6 are input to the positional deviation detection circuit 8 to determine the positional deviation, and based on the result, the 1-positional deviation correction circuit 9 lengthens the read position of the buffer memory 7 and corrects the positional deviation. The inspection pattern stored in the buffer memory 4 is input to the defect determination circuit 10 together with the standard pattern whose positional deviation has been corrected, and a defect determination is performed.

FIG. 3 shows the specific formation of positional deviation and horizontal output 1188, assuming that a maximum positional deviation of ±2 strokes occurs between the inspection pattern and the standard pattern. In FIG. 3, the standard pattern signal 6 passes through a shift register group 11 corresponding to the scan line, and a local area 12 of 5×5 pixels consisting of serial-in-parallel-out shift registers is successively cut out. On the other hand, the test pattern 3 is also inputted to the 3-bit shift register 14 after passing through the shift register group 13 corresponding to the length of the run f break, and the output of 14 and the output of 12
output 3.3 is the corresponding position of both patterns when there is no positional shift. 14 outputs and 12 5x5;2
For each of the five outputs, 25 EXOR[gl circuits 5 are used to find a mismatch, and the number of mismatches is counted using a 25 counter 161C. A positional deviation calculating circuit 17 inputs the counter 16 of 25 and calculates the positional deviation.

It consists of a processing device such as a microphone Q7'Q processor.

The processing flow is shown in FIG. In Figure 7 @ 4, the minimum count value of the counter to be found is MIN, the number of counters that indicate @small count value is CRT, and the distribution IJ that stores the address of the counter that indicates the minimum count of 1 is I, J. do. In this way, according to 70- shown in FIG.
When all five counters are processed, the misalignment trays ΔX and ΔY are filled with water.

The reason why ΔX and ΔY are each subtracted by 3 from the counter address stored in 1.1 is because there is no position shift when the minimum count value is superior at the counter address (3, 5). This is because it belongs to the species. Also, the fourth
In the figure, when there are multiple counters that show the minimum count value (CNT = 1), only 10,000 is set for ΔX and ΔY, but for those that are not set, the previous result is used. It is for use.

Next, what about the configuration of buffer memories 4 and 7? This will be explained using Figure 5. In FIG. 2, the rows 5 are aligned for each fixed section as explained by CC. Since each section has an overlapping part, each of the backup memories 4 and 7 is composed of two sides. That is, the 5th As shown in the figure, the test pattern signal 3 is input to the memories 20 and 21 via the gate circuits w618 and 19, and then to the memory 2o.

21 is read out via the selection circuit 22 and a defect determination is made. The gate circuits 18 and 19 are
In -1-1, 3 is input to the memories 20 and 21, and the selection circuit 22 selects the memory to be determined to be defective. Although FIG. 5 shows the buffer memory 4, the buffer memory 7 also has a similar configuration, except that the answer is different. FIG. 6 shows the relationship between the test pattern and the standard pattern. δZm'yk is the maximum displacement child. In addition, alignment is performed using the buffer memories 4 and 7, and defects are determined.
The results need to be output before defect determination for the next block begins, and the welding speeds from 4 and 7 need to be faster than the input speeds.

〔Effect of the invention〕

As explained above, according to the present invention, even if there are patterns in only one direction, it is possible to align two patterns, allowing for accurate pattern calibration. Also, the second
It is also possible to match the non-counting range shown in the figure with the judgment range, but in this case, another set of positional deviation detection circuits shown in Fig. There is.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the overall configuration of the present invention, FIG. 2 is a diagram explaining the problems of the prior art, FIG. 5 is a diagram showing a specific problem of the misalignment amount detection circuit, and FIG. 4a is a diagram showing the misalignment detection circuit. FIG. 5 is a diagram showing the buffer memory, and FIG. 6 is a diagram showing the relationship between the calibration pattern and the standard pattern. 1... Image fold fit 2... Binarization circuit 4.7... Buffer memory 8... Positional deviation amount detection circuit 9... Positional deviation correction circuit 10... Defect determination section 1 Diagram 2■ t0￥J4 diagram

Claims (2)

[Claims] 1. In a pattern comparison inspection method that compares and inspects two binary patterns, local areas in the range of positional deviation that may occur between the two patterns are sequentially cut out from the reference binary pattern, and each pixel in the local area and the test binary pattern are Find the discrepancy for each pixel, count it over a certain interval, and if there is one position where the count value is the minimum, calculate the horizontal and vertical position shift from that position, and if there are multiple positions, calculate the count value. Find out whether the minimum exists in the horizontal direction or in the vertical direction, use the previous position shift amount in multiple directions to find the position shift amount, and then calculate the position shift amount between the two based on this position shift amount. A pattern inspection method characterized by aligning patterns and determining defects. 2. In a pattern comparison inspection device that compares and inspects two binary patterns, a cutting means for sequentially cutting out local air within a range of positional deviation that may occur between the two patterns from a reference binary pattern; The mismatch between each pixel in the local area and the test binary pattern is determined for each pixel, counted over a certain interval, and the position where the counted value is the minimum is determined as 1.
If there are two or more positions, calculate the amount of position shift in the horizontal and vertical directions, and if there are more than one, find out whether the minimum count exists in the horizontal or vertical direction, and calculate the displacement in the direction where there are multiple positions. a positional deviation calculation means that adopts the previous positional deviation amount; a positional deviation correction means that aligns two patterns based on the positional deviation amount calculated by the positional deviation calculation means; 1. A pattern inspection method comprising: a defect determining means for comparing two patterns aligned by a correcting means to determine a defect.