JPH036673A - Method and device for pattern inspection - Google Patents

Abstract

PURPOSE:To inspect a pattern to be inspected with high accuracy and in real time by correcting the address of a reference pattern which is previously stored based on the position deviation data and in accordance with the pattern to be inspected and judging the noncoincidence between the picture data on the pattern to be inspected and the reference pattern. CONSTITUTION:A position mark 1b put in accordance with a pattern 1a to be inspected is detected, and various position deviation data are calculated for the position of the detected mark 1b against the position of a reference mark set previously. Then the address of the reference pattern stored previously in accordance with the pattern 1a is corrected based on those calculated position deviation data. The reference pattern data is read out of a reference pattern setting part 18 based on the corrected address. Then the noncoincidence is judged between the reference pattern data and the picture data on a wiring pattern. The quality of the pattern 1a is judged from the decided discordance state. Thus it is possible to judge the quality of the patterns 1a fetched via a video camera 2 in real time and with high accuracy.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a pattern inspection method and an apparatus therefor applied to visual inspection of wiring patterns on printed circuit boards.

[Prior Art] Conventionally, wiring patterns formed on circuit boards such as film boards and hard boards have been subject to problems such as disconnections, short circuits, and
Appearance inspections such as thickening and thinning are carried out, but most of these inspections involve pattern inspection processing, which captures the wiring pattern appearance as image data and compares it with a reference pattern to determine pass/fail. It has been adopted.

However, conventionally, as this type of inspection method, two TVs
Using the left camera, one TV left camera X1y reads the position mark of the inspected member on the θ table, moves the xSy and θ table according to this position data, and matches the position mark with the reference position of the pattern inspection processing means. Then, from this state, the image of the object to be inspected is read from the other TV left camera, or the position mark and image data of the object to be inspected are simultaneously read, and these data are stored in the memory and the position mark is read. There is also a method that corrects the positional deviation of image data.

[Problems to be Solved by the Invention] However, in the former method, in order to match the position mark of the inspected member with the reference position for image processing,
Since the s θ table is moved, it is possible to improve the precision of pattern inspection, but the work efficiency is poor (and the table drive mechanism is complicated, so there is a disadvantage in terms of price). In addition, since the latter corrects positional deviations for the memory contents that store position marks and image data, it is advantageous in terms of price since it does not require a table drive mechanism compared to the former. However, since the image data that has been read is stored in the memory and then corrected for the upper position, the processing speed is still not fast enough compared to real-time processing. When storing things in memory, it becomes complicated to correct misalignment of the memory contents.
A decrease in work efficiency due to the time required was unavoidable.

The present invention was made in view of the above-mentioned circumstances, and an object of the present invention is to provide a pattern inspection method and apparatus that can perform pattern inspection at extremely high speed while maintaining accuracy, and is also advantageous in terms of cost. do.

[Means for Solving the Problems] The present invention detects a position mark attached to a pattern to be inspected and calculates various positional deviation data for the detected mark position with respect to a preset reference mark position. Then, based on this positional deviation data, the address of the reference pattern stored in advance corresponding to the pattern to be inspected is corrected, and the mismatch between the reference pattern data read out by this correction address and the image data of the pattern to be inspected is corrected. I am trying to detect it. Moreover, the positional deviation data calculated with respect to the reference mark position set in advance for the mark position is the amount of positional deviation in the X%Y direction and the rotational direction with respect to the position coordinate data of the reference mark. Furthermore, the correction address is obtained by using affine transformation based on the correction amount parameters of the positional deviation amounts in the xSy direction and the rotational direction with respect to the position coordinate data of the reference mark at the mark position and the address corresponding to the image data of the pattern to be inspected. I try to ask for it.

On the other hand, the present invention provides a position mark detection means for detecting a position mark attached corresponding to a pattern to be inspected, and positional deviation data regarding the mark position detected by the position mark detection means with respect to a preset reference mark position. a positional deviation calculating means for calculating the positional deviation, and an address correcting means for correcting the address of the reference pattern stored in advance in accordance with the inspection pattern based on the positional deviation data calculated by the positional deviation calculating means. , an image data output means for outputting image data from the pattern to be inspected; and determining a mismatch between the image data output from the image data output means and the reference pattern data read out based on the address corrected by the address correction means. It consists of a means of determining inconsistency.

[Operation] As a result, the address of the reference pattern stored in advance corresponding to the pattern to be inspected is corrected based on the positional deviation data calculated with respect to the reference mark position set in advance for the mark position, and the address of the reference pattern stored in advance corresponding to the pattern to be inspected is corrected. Since the mismatch between the image data of the pattern and the reference pattern data read out based on the corrected address is determined, the inspection pattern inspection can be carried out with high accuracy and efficiency in real time.

[Example] An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 shows the circuit configuration of the same embodiment.

In the figure, reference numeral 1 denotes a circuit board which is a member to be inspected. This circuit board 1 is made of a strip-shaped flexible board as shown in FIG.
It also has a position mark 1b corresponding to each spacing line pattern 1a. Such a circuit board 1 is then transported in the horizontal direction shown in FIG. 2 while inspecting the wiring pattern 1a.

A video camera 2 such as a COD camera and a correction camera 3 are arranged opposite to such a circuit board 1. Here, the video camera 2 captures an image of the wiring pattern 1a of the circuit board 1, and outputs this image data in synchronization with the external synchronization signal φ5T. Further, the correction camera 3 captures an image of the position mark 1b.

In this case, as shown in FIG. 2, the video camera 2 and the correction camera 3 are moved in the direction of the arrows shown in the figure, respectively, to the supporting part 4.5 provided in a two-layer structure in a direction perpendicular to the moving direction of the circuit board 1. It is set freely.

The image data output from the video camera 2 is preprocessed by a preprocessing section 6 and converted into digital data by an A/D conversion section 7, and then provided to a time correction circuit 8. The time correction circuit 8 delays the output of the A/D converter 7 by the time necessary for address correction processing, which will be described later, and supplies the delayed output to the mismatching dot detection circuit 9.

On the other hand, the video output of the position mark 1b captured by the correction camera 3 is preprocessed by the preprocessing unit 10 and then processed by the A/D converter 11.
After being converted into digital data, the S/P converter 12
given to. The S/P converter 12 converts serial data provided from the A/D converter 7 into parallel data.

The parallel data from the S/P converter 12 is then given to the positional deviation calculator 13.

The output of the mark position setting section 14 is given to the positional deviation calculation section 13 . This mark position setting section 14 stores position coordinate data X of a reference mark set in advance for a mark position 1b attached to the wiring pattern 1a. , 'lo are stored, and the position coordinate data X is stored according to a command from the control unit CTR. ,y.

It is designed to output .

The positional deviation calculation unit 13 calculates positional coordinate data X. , yo, the position mark 1 captured by the correction camera 3
This is to calculate the positional deviation of b.

In this case, the positional deviation calculation unit 13
4. Coordinate data X of the reference position mark. , Yo, the positional deviation Mkll − of the position mark 1b in the X direction, the positional deviation amount g in the y direction, and the positional deviation amount θ in the rotational direction are calculated, respectively. The positional deviation ml in the X direction, the positional deviation EiN in the y direction, and the positional deviation amount θ in the rotational direction are expressed as position coordinate data x0, yo and the image position X. Determining from Yo − is already known and will be omitted here. After each positional deviation J4LD t , Ω, and θ are calculated, the positional deviation calculation unit 13 further performs calculation to obtain a correction amount.

Here, when each correction amount of the XSy address is found using affine transformation with reference to FIG. Displacement amount p,,y
Amount of positional deviation in direction! , then .

Then, by processing the terms in the matrix of the above equation, the corrected address is given by: As a result, the actual correction matrix becomes
osθx−sinθy+(1゜COSθ−0, sinθ
)...(3) Y-sinθX+CO5θ”/+(g, sinθ+l,
cos θ)...(4) is obtained. Among the items shown in equations (3) and (4), items other than x and y can be calculated for the current pattern to be inspected read by the correction camera 3. In other words, for the corrected address X, cos
θ, −5inθ and (J7.cosθ−g, sinθ
), and for the correction address Y, sin θ, cos
θ and (j7.sin θ+Nycos θ) can be calculated in advance.

Here, returning to FIG. 1, after the positional deviation calculation unit 13 calculates the positional deviation amounts px and pθ, the above-mentioned (
3) Each item corresponding to formula (4), CO8θ, -s in
θ, (ρ cosθ−Ω sinθ) sinθ, (
ΩxsinO+, Q, cos θ) are calculated, and the results of these calculations are written into the register 141.

The output of the register 141 is given to the X address correction section 16 together with the address data X of the scanning address generation section 15, and to the Y address correction section 17 together with the address data y.

Here, the scanning address generation unit 15 generates address data
sY, which here corresponds to the address of the image data of the wiring pattern 1a output from the video camera 2.

FIG. 4(a) shows a specific circuit configuration diagram of the X address correction section 16. In this case, the multiplier 161 multiplies c-osθ based on the address
Based on the address so of the scanning address generation unit 15 and the positional deviation amount θ in the rotational direction,
is multiplied by multiplier 162, and these multipliers 161 and 162
The adder 163 adds the outputs of , and the addition result and (,Q
mcos θ-g, sin θ) is supplied to the adder 164, and the corrected address X is obtained by adding it.

Further, FIG. 4(b) shows a specific circuit configuration diagram of the y address correction section 17. In this case, based on the address y of the scanning address generation unit 15 and the amount of positional deviation θ in the rotational direction written in the register 141, the multiplier 1
71, and based on the address y of the scanning address generation unit 15 and the positional deviation amount θ in the rotational direction,
θ is multiplied by multiplier 172, and these multipliers 171.17
The outputs of 2 are added by an adder 173, and the addition result and (Ω
xsinθ0g, cosθ) is supplied to the adder 174, and the corrected address Y is obtained by adding it. As is clear from FIGS. 4(a) and 4(b), the X address correction unit 16 and the y address correction unit 17 each have a multiplication-addition-addition calculation circuit, so that the The image data is delayed by this calculation processing time. A time correction circuit 8 disposed between the video camera 2 and the mismatched dot detection circuit 9 is required to match the delay and timing in each of the correction sections 16 and 17. For example, if the processing time of one arithmetic unit is shifted by 50 to 100 nanoseconds, the time correction circuit 8 delays the time by 150 to 300 nanoseconds.

Returning to FIG. 1, the corrected address XSY from the X address correction section 16 and the y address correction section 17 is given to the reference pattern setting section 18. Here, reference pattern setting section 1
A reference pattern 8 corresponding to the wiring pattern 1a to be detected is stored in advance, and the reference pattern data is output based on correction addresses X and Y.

The pattern data from the reference pattern setting section 18 is then given to the mismatching dot detection circuit 9.

As shown in FIG. 5, this mismatching dot detection circuit 9 combines pattern data A of the basic pattern setting section 18 and image data B of the wiring pattern 1a provided via the time correction circuit 8.
This is to detect the mismatch state in dot units corresponding to the address of the image data, and the detection of these mismatches is performed within the effective area C set in one effective area setting section.

FIG. 6 shows a specific circuit diagram of the mismatched dot detection circuit 9. As shown in FIG. In this case, the image data B of the wiring pattern 1a is supplied to one input terminal of the exclusive OR circuit 91, and the image data B of the AND circuit 9 is supplied via the inverter 92.
3 and the second input terminal of the AND circuit 94, and the pattern data A of the reference pattern setting section 18 is supplied to the other input terminal of the exclusive OR circuit 91. The signal is applied to the third input terminal of the AND circuit 94 via the third input terminal and the inverter 95. And exclusive OR circuit 91
The output of the AND circuit 96 is applied to one input terminal of the AND circuit 96, and the output of the AND circuit 93 is applied to the other input terminal of the AND circuit 96.
．． The setting output of the effective area setting section 19 is applied to the first input terminal of the effective area setting section 94 . The set output of this effective area setting section 19 is designed to generate an rHJ level output for the effective area. As a result, as shown in FIG. 5, mismatched dots between pattern data A and image data B are detected within the effective area C set in the effective area setting section 19, and both pattern data A and image data B exist. In the dot part, none of the AND circuits 93, 94, and 96 generates an rHJ level output, and only pattern data A exists.
Based on the rHJ level output of the exclusive OR circuit 91, the AND circuit 96 generates an rHJ level output indicating a mismatched dot, and the AND circuit 93 generates an rHJ level output indicating a narrow dot. If there is an rHJ level output from the exclusive OR circuit 91, the AND circuit 96 outputs rHJ indicating that there is a mismatch dot.
At the same time, the AND circuit 94 generates an rHJ level output representing a thick dot.

The output from the mismatched dot detection circuit 9 is given as a write signal to the mismatched dot position storage section 20, thin dot position storage section 21, and thick dot position storage section 22, respectively.

Further, correction addresses X and Y output from the X address correction section 16 and the y address correction section 17 are connected to the mismatching dot position storage section 20, thin dot position storage section 21, and thick dot position storage section 22 via a data bus. Ru. Therefore, each time the AND circuit 96 outputs the mismatched dot signal, the correction address X1Y indicating the position of the mismatched dot at that time is stored in the mismatched dot position storage section 20.

Similarly, each time a thin dot signal is output from the AND circuit 93, a correction address XSY indicating the position of the thin dot at that time is stored in the thin dot position storage section 21, and a thick dot signal is output from the AND circuit 94. each time, a correction address X indicating the position of the thick dot at that time,
Y is stored in the large tot position storage section 22.

In this case, the mismatched dot position storage section 20, thin dot position storage section 21, and thick dot position storage section 22 are controlled by an image processing processor DSP (not shown) and accessed at high speed. In addition, these mismatched dot position storage units 20
, thin dot position storage section 21, thick dot position storage section 2
The output of No. 2 is given to the pass/fail judgment section 23. The pass/fail judgment unit 23 calculates the distribution of mismatched dots, thin dots, and thick dots from the position information sent from each position storage unit 20 to 22, and determines whether the dots exist within a preset tolerance range. The quality of the wiring pattern 1a is determined by the following. The judgment result of the acceptability judgment section 23 is given to the display section 24 and displayed there.

On the other hand, the output from the mismatching dot detection circuit 9 is also given to the counter 25. This counter 25 counts the number of mismatched dots, the number of thin dots, and the number of thick dots.
The data is stored in the thin dot number storage section 27 and the thick dot number storage section 28, and is then provided to the image processing processor DSP. As a result, the number of corrected addresses X1Y as position information stored in each position memory #20 to #22 is grasped, so when reading this position information to the quality determination unit 23, the memory units 26 to 28 When the image processing processor DSP has read out the number stored in the storage units 20 to 22, the image processing processor DSP can finish accessing each of the storage units 20 to 22, thereby making the processing more efficient.

Next, the operation of the embodiment configured as described above will be explained.

First, the wiring pattern 1 to be inspected by the correction camera 3
An image of position mark 1b corresponding to point a is captured. The video output captured by the correction camera 3 is processed by a preprocessing unit 10.
After being pre-processed and converted into digital data by the A/D converter 11, the data is converted into parallel data by the S/P converter 12 and provided to the positional deviation calculator 13. Then, this positional deviation calculation section 13 calculates the position coordinate data X of the mark position setting section 14. , yo, the positional shift is calculated. In this case, position coordinate data X as shown in FIG. ,yo
The positional deviation of the reference position mark 1b with respect to the X direction is the positional deviation 1. , a positional deviation Ω in the y direction, and a positional deviation θ in the rotational direction. Positional deviation calculation unit 13
Then, the positional deviation of these position marks 1b in the X direction Ω8y
The positional deviation p in the direction and the positional deviation θ in the rotational direction are calculated, and further COSθ, -5inθ, (Ω, cosθ-
g, sinθ), sinθ, (R, sinθ+0.co
sθ) is calculated and written to the register 141.

The output of the register 141 is specifically given to the X address correction section 16 and the y address correction section 17 together with the address data X1y of the scanning address generation section 15 which is output in synchronization with the external synchronization signals φ and T, and here the above-mentioned Address correction using affine transformation is performed, and the correction address
Y is required. In this case, in the X address correction section 16, as shown in FIG. 4(a), the scanning address creation section 15
<cos θ is given to the multiplier 161 based on the address data
Here, COS θX is calculated, and then, based on the address y of the scanning address generation unit 15 and the positional deviation amount θ in the rotational direction (-sin θ is given to the multiplier 162, -5 in θ
y is calculated, the outputs of these multipliers 161 and 162 are added by an adder 163, and this addition result and (Ω, cosθ−Ω
sin θ) by the adder 164, the corrected address X of the above-mentioned equation (3) is obtained.

Furthermore, regarding the y-address correction section 17, FIG.
), the address y of the scanning address generation unit 15
Based on the amount of positional deviation θ in the rotational direction from the register 141, <sinθ is given to the multiplier 171, where sin θX is obtained, and then the address y of the scanning address generation unit 15 and the amount of positional deviation θ in the rotational direction are Based on this, (COSθ is given to the multiplier 172, cosθy is obtained here, the outputs of these multipliers 171 and 172 are added together in the adder 173, and this addition result and (N, stnθ+l, cosθ) are added to the adder 17.
By adding 4, the corrected address Y of the above-mentioned equation (4) can be obtained.

Then, the correction addresses XSY of the X address correction section 16 and the y address correction section 17 are given to the reference pattern setting section 18, and the reference pattern data corresponding to these correction addresses X and Y are read out from the reference pattern setting section 18. , are output to the mismatching dot detection circuit 9.

On the other hand, wiring pattern 1 captured by video camera 2
The video output of a is an external synchronization signal φST as image data.
output in sync with In this case, external synchronization signal φST
The image data output by the scan address generator 1
This corresponds to the address xSV output from 5 in synchronization with the external synchronization signal φ'iT. This image data is preprocessed by a preprocessing section 6, converted into digital data by an A/D conversion section 7, and then delayed by a time correction circuit 8 for the time necessary for address correction processing and sent to a mismatching dot detection circuit 9. given to.

In this state, the mismatching dot detection circuit 9 detects the reference pattern data A of the pattern setting section 18 and the time correction circuit 8.
A mismatch state with the image data B from the image data B is detected in units of dots corresponding to the addresses x and y of the image data B.

Here, assuming that the reference pattern data A of the reference pattern setting section 18 and the image data B of the time correction circuit 8 are located in the effective area C in a relationship as shown in FIG. In the part shown in FIG. 6, an rHJ level output is given from the effective area setting section 19, an rHJ level input is given as reference pattern data A, and an rHJ level input is given as image data B.
Starting with the exclusive OR circuit 91, AND circuits 93, 94 .
All outputs of 96 remain at the "L" level, and no mismatched dots are detected. Next, in the part indicated by D2 in the figure, in FIG.
A level output is given, an "L" level input is given as reference pattern data A, and an rHJ level input is given as image data B. Then, from the exclusive OR circuit 91, r
Since the HJ level output is generated, the AND circuit 96 goes to the rHJ level and a mismatch dot detection output is generated, and at the same time, the AND circuit 94 goes to the rHJ level and a detection output indicating a thick dot is generated.

Then, each time a signal is output from these AND circuits 96 and 94, the correction address X%Y, which is the position information of the mismatched dot and the thick dot, is stored in the mismatched 11 dot position storage section 20 and the thick dot position storage section 22. Become so. Further, in a portion indicated by D3 in FIG. 6, an rHJ level output is given from the effective area setting section 19 in FIG. 6, an rHJ level input is given as reference pattern data A, and an rLJ level human power is given as image data B. Then, the exclusive OR circuit 91 generates an rHJ level output, so the AND circuit 96 goes to the rHJ level and a mismatch dot detection output is generated, and at the same time, the AND circuit 93 goes to the rHJ level and detects a thin dot. Output is generated.

Then, each time a signal is output from the AND circuits 96 and 93, a correction address XSY, which is the position information of the mismatched dot and the narrowed dot, is stored in the mismatched 11 dot position storage section 20 and the narrowed dot position storage section 21. Become so.

Then, in the acceptability judgment section 23, the fat dot storage section 2
The quality of the wiring pattern is determined based on the distribution of thick dots sent out from the thin dot position storage section 21 and the distribution of thin dots sent out from the thin dot position storage section 21, and the results are displayed on the display section 24.
will be displayed. The number of mismatched dots, the number of thin dots, and the number of thick dots are also counted by the counter 25, and the count values are stored in the number of mismatched dots storage section 26, the number of thin dots storage section 27, and the number of thick dots storage section 28. Therefore, when reading the positional information from each of the storage units 26 to 28, the reading process of the mismatched positional information can be completed when the count value stored in each of the storage units 26 to 28 is reached, which is efficient. Become.

Therefore, if there is a deviation in this way, the position of the mark attached to the wiring pattern is read, and the positional deviation of this mark position with respect to the preset reference mark position is calculated as the positional deviation in the X direction, the positional deviation in the X direction, 1y, and The positional deviation θ in the rotational direction is determined, and the correction address XSY is calculated using affine transformation using the positional deviation data and the scanning address X1y corresponding to the address of the image data of the wiring pattern as the correction amount parameter.
, Y reads out the reference pattern data from the reference pattern setting section, and determines the mismatch between the reference pattern data and the image data of the wiring pattern, and determines whether the wiring pattern is good or bad from this mismatch state. It is possible to more efficiently determine the acceptability of wiring patterns that are sequentially captured in real time.

Moreover, due to the deviation of the mark position from the reference mark position,
The elements of positional deviation in the direction 1), positional deviation g in the X direction, and positional deviation θ in the rotational direction are taken into consideration, and based on these, the correction address X,
Since Y is obtained by affine transformation, it is possible to judge with high precision whether there is a mismatch between the reference pattern data and the image data of the wiring pattern. Also, compared to the conventional one, XS!
/, θ table and a drive mechanism for this table are not required, making it advantageous in terms of cost.

It should be noted that the present invention is not limited to the above-mentioned embodiments, but can be implemented with appropriate modifications without changing the gist.

For example, in the above embodiment, two cameras were used, but one
It can also be done with just a stand. In this case, the position mark is first read, sinθ, CO8θ, and gxlMy are calculated, and real-time processing is performed at the same time as the wiring pattern is read. In this case, after the A/D converter 7,
It is necessary to provide a switching element for switching the connection between the time correction circuit 8 and the S/P converter 12. In addition, although the above description describes the case where it is applied to the visual inspection of wiring patterns,
It is also possible to apply to other pattern inspections.

[Effects of the Invention] The present invention corrects the address of the reference pattern stored in advance corresponding to the pattern to be inspected based on the positional deviation data calculated with respect to the reference mark position set in advance regarding the mark position. Since the mismatch between the image data of the pattern to be inspected and the reference pattern read out based on the corrected address is determined, the pattern to be inspected can be inspected accurately and efficiently in real time. Furthermore, since an X5Ys θ table or the like is not used, the inspection speed can be increased, and a table driving mechanism is not required, so the price can be reduced. In addition, various factors such as positional deviation in the X direction, positional deviation in the Since the corrected address for the part is obtained, it is possible to judge with high precision whether or not the image data of the pattern to be inspected and the reference pattern match, and highly reliable pattern appearance inspection can be realized.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the circuit configuration of an embodiment of the present invention, FIG. 2 is a configuration diagram for explaining the mounting state of a video camera and a correction camera used in the embodiment, and FIG.
The figure is a diagram for explaining the calculation results of the position mark position deviation calculation unit of the same embodiment, and FIG. 4 is an X used in the same embodiment.
A circuit diagram showing the address correction section and the X address correction section, FIG. 5 is a diagram for explaining the operation of the mismatching dot detection circuit used in the same embodiment, and FIG. 6 is a circuit diagram showing the mismatching dot detection circuit. It is. 1... Circuit board, 1a... Wiring pattern, 1b...
・Position mark, 2... Video camera, 3... Correction camera, 8... Time correction circuit, 9... Unmatched dot detection circuit, 13... Position deviation calculation unit, 14... Mark Position setting section, 15...Scanning address creation section, 16...
-X address correction section, 17...X address correction section, 1
8... Reference pattern setting section, 19... Effective area setting section, 20... Unmatched dot position storage section, 21...
Thin dot position storage unit, 2 2... Thick dot position storage unit, 3... Pass/fail judgment unit, 24... Display unit, 5... Counter.

Claims (5)

[Claims] (1) Detect the position mark attached corresponding to the pattern to be inspected, calculate the positional deviation data of the detected mark position with respect to the preset reference mark position, and calculate the positional deviation data of the detected mark position with respect to the preset reference mark position. A pattern inspection method characterized by correcting the address of a reference pattern stored in correspondence with the inspection pattern, and determining a mismatch between the reference pattern data read out based on the correction address and the image data of the pattern to be inspected. . (2) A claim characterized in that the positional deviation data calculated with respect to a preset reference mark position regarding the mark position is the amount of positional deviation in each of the x, y directions, and rotational direction with respect to the positional coordinate data of the reference mark. The pattern inspection method according to item 1. (3) The correction address uses affine transformation with the positional deviation amount in the x, y direction and rotational direction with respect to the position coordinate data of the reference mark at the mark position and the address corresponding to the image data of the pattern to be inspected as correction amount parameters. 3. The pattern inspection method according to claim 2, wherein (4) Disagreement judgment between the standard pattern data and the image data of the pattern to be inspected is determined when the standard pattern data is “present” and the image data is “absent”, and when the standard pattern data is “absent”.
4. The pattern inspection method according to claim 1, wherein the pattern inspection method is performed separately when the image data is "present." (5) Position mark detection means for detecting a position mark attached corresponding to the pattern to be inspected, and positional deviation data for the mark position detected by the position mark detection means with respect to a preset reference mark position is calculated. a positional deviation calculating means; an address correcting means for correcting the address of a reference pattern stored in advance in correspondence with the pattern to be inspected based on the positional deviation data calculated by the positional deviation calculating means; and the pattern to be inspected. an image data output means for outputting image data from the image data output means; and a mismatch determination means for determining a mismatch between the image data output from the image data output means and the reference pattern data read out based on the address corrected by the address correction means. A pattern inspection device characterized by comprising: