JPH08201037A - Pattern inspection device and method - Google Patents

Abstract

PURPOSE: To detect a difference part between two patterns by the rocking comparison for each local region without using a large-scale hardware and without requiring a long time. CONSTITUTION: The binary image data of an inspection target pattern are stored as object data and mask data (S10 and S12), data DP of each inspection block are successively extracted from the object data (S14), rocking is made and range is set with a position corresponding to a current inspection block in a mask pattern for each extraction, and data DM of each mask block which is successively shifted by one picture element two-dimensionally within the range are successively extracted (S18). Then, when all picture elements are '1' or '0' for both data DP and DM, a current inspection block is judged to be a matched part without rocking comparison (S22 and S32). Otherwise, rocking comparison (S24-S30) is performed, thus judging whether the inspection block is a matched part or a different part (S32 and S34).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern inspection apparatus and method used for comparing a first school and a second school to detect a difference in a calibration step of a printing process.

[0002]

2. Description of the Related Art At the proofreading stage of a printing process, a proof or a plate-making film is compared with a first school and a second school to detect a difference, and whether or not a correction-instructed portion is correctly corrected or not. There are no changes other than the above,
The plate inspection work is done to confirm. As a plate inspection device for performing such work, the images obtained by shooting the film of the first school and the re-school with a TV camera are stored in the memory, and the stored images of the first school and the re-school are compared and collated. Japanese Patent Publication No. 4-61345 discloses a method in which a difference between the two is displayed in a blinking manner (hereinafter referred to as "conventional example 1").

When the images of the first school and the re-school are compared and collated in the prior art example 1, it is necessary to align the positions of the first school and the re-school. The above publication describes the following two methods for this alignment. (1) Method using pin mechanism (2) Method to align the position of the re-school film while alternately displaying the first-school image stored in the memory and the re-school image currently captured by the TV camera

On the other hand, as a method of comparing and collating two image patterns, by comparing the two image patterns for each local area while shifting them within a predetermined range, it is possible to absorb the positional deviation and the positioning error of the object. A method of matching two patterns (hereinafter referred to as “shaking comparison method”) is known, and a pattern defect inspection apparatus (hereinafter referred to as “conventional example 2”) as an example of the shaking comparison method is disclosed in Japanese Patent Publication No.
-21769 gazette. This pattern defect inspection apparatus detects a defect by comparing a two-dimensional object pattern of a binary pixel signal with a two-dimensional master pattern for each local area having a predetermined size. One local area of the object pattern is detected. On the other hand, as shown in FIG. 5, an area (hereinafter referred to as a “shaking range”) in which the periphery is expanded two-dimensionally by a predetermined number of pixels centering on the local area M0 corresponding to the position in the master pattern is used. An area expanded by 3 pixels vertically and horizontally is set as a shaking range), and a master pattern of a plurality of local regions M11 to M77 which is sequentially shifted two-dimensionally by one pixel within the shaking range is set. The object pattern of one local area is compared with the pattern matching method using the inspection window WM. At this time, in order to reduce the time required to inspect the entire pattern, pattern matching between the object pattern of one local area and the master pattern of the plurality of local areas M11 to M77 is performed by parallel processing.

[0005]

When the method (1) is used for the alignment between the first school and the re-school in the above-mentioned conventional example 1, a pin hole is provided for an inspection object having no pin hole. Must be newly established. Further, when the method of (2) is adopted, the operator moves the inspection object (revision film) by hand while looking at the screen of the monitor, so that there is a limit to the accuracy of alignment. . On the other hand, it is conceivable to separately provide a drive mechanism to move the inspection object, but in this case, there is a problem that the cost is increased and the apparatus is enlarged. Furthermore, when the inspection object is stretchable like paper, it is not possible to make all the parts of the inspection object match by any of the methods (1) and (2).

On the other hand, in the conventional example 2, the above-mentioned drawbacks in the conventional example 1 can be solved, but the pattern matching of the object pattern of one local area and the master pattern of the plurality of local areas M11 to M77 is performed in parallel. The hardware for the pattern matching becomes large in scale, and the cost is increased. Further, in the plate inspection device, an object to be stretched like paper is regarded as an inspection object, or even an inspection object whose expansion and contraction can be ignored may cause an inclination between two images of the inspection object. As a result, expansion and contraction are accumulated to increase the positional deviation, or the positional deviation at the end portion is increased due to the inclination of the image (see FIG. 6). In order to deal with these in the shake comparison, the shake range must be increased. When the shaking range becomes large, the hardware becomes large in scale when performing parallel processing as in the second conventional example. On the other hand, in the case where the shake comparison is sequentially performed in order to avoid increasing the scale of the hardware, the processing time becomes longer as the shake range increases.

The present invention has been made to solve such a problem, and an object of the present invention is to use a large-scale hardware without requiring a long processing time. It is an object of the present invention to provide a pattern inspection apparatus and method for detecting a difference between two image patterns by comparing each local area based on the shake comparison method.

[0008]

A first pattern inspection apparatus according to the present invention made to solve the above problems,
In a pattern inspection apparatus for collating two inspection target patterns and detecting a difference between the two inspection target patterns, a) by two-dimensionally scanning the two inspection target patterns and reading each pixel , An image input means for generating two binary image data composed of two kinds of values of "1" and "0", and b) two binary image data generated by the image input means, respectively, as an object pattern. Image storage means for storing as object data and master data representing a master pattern; and c) object extraction means for sequentially extracting from the object data image data corresponding to each block in units of a block consisting of a plurality of pixels in the object pattern. d) Image data for one block is extracted by the object extracting means. Object counting means for determining the number of pixels of "1" and the number of pixels of "0" in each block, and e) the image in the master pattern for the image data extracted in block units by the object extracting means. An extension block is set which is two-dimensionally expanded to a periphery by a predetermined number of pixels with a block positionally corresponding to a block of data as a central block, and each block is sequentially shifted one-dimensionally by two pixels in the extension block. Master extraction means for sequentially extracting corresponding image data from the master data, and f) based on the number of "1" pixels and the number of "0" pixels obtained by the object counting means for one block of the object pattern. , Ratio of pixels of "1" or "0"
Is larger than a predetermined value in the vicinity of 100%, and if it is larger than the predetermined value, it is determined that the block of the object pattern is not a different portion, and it is not extracted by the object extraction means. Image data for one block is extracted as image data for the next block,
When the value is less than the predetermined value, the first control means for causing the master extraction means to sequentially extract the image data of each block in the extended block of the master pattern corresponding to the block of the object pattern, and g) by the object extraction means The image data extracted in block units is compared with each image data sequentially extracted in block units from the extension block by the master extraction unit corresponding to the image data, and the image data represented by the two image data to be compared is compared. Collating means for sequentially determining whether two patterns match, and h) if the collating means determines that the two patterns match for any block in the extension block, the extension block by the master extracting means. The extraction of the image data from the inside is stopped and the object extracting means Unextracted as image data of the next block 1
Second control means for extracting image data for blocks,
It is characterized by having.

In a second pattern inspection apparatus according to the present invention, in the first pattern inspection apparatus, the number of "1" pixels in the block is determined based on the image data for one block extracted by the master extraction means. And master counting means for determining the number of pixels of "0", and the first control means is "1" in one block of the object pattern.
When the ratio of the pixels of "0" or the ratio of the pixels of "0" is larger than the predetermined value, i) the image data of each block in the extended block of the master pattern corresponding to the block of the object pattern is sequentially transmitted to the master extraction unit. Ii) based on the number of pixels of "1" and the number of pixels of "0" obtained by the master counting means for one block in the extension block, the ratio of pixels of "1" or "0" It is determined by the master extraction means whether the block of the object pattern is not a different portion only when the ratio of the pixels of "is larger than a predetermined value in the vicinity of 100%. Stopping the extraction of the image data from within the extended block, and causing the object extraction means to extract the unextracted image data of one block as the image data of the next block. It is characterized.

According to the first pattern inspection method of the present invention, two patterns to be inspected are two-dimensionally scanned and read out for each pixel, so that two patterns of "1" and "0" are formed. One binary image data is generated, and the two binary image data are stored as object data representing an object pattern and master data representing a master pattern, respectively, and each block is made up of a block composed of a plurality of pixels in the object pattern as a unit. The image data corresponding to the image data is sequentially extracted from the object data, and the image data extracted in block units from the object data is two-dimensionally set with the block positionally corresponding to the block of the image data in the master pattern as the central block. Set an expansion block that expands to the periphery by a specified number of pixels and The image data corresponding to each block which is two-dimensionally shifted by one pixel in the lock is sequentially extracted from the master data, and the image data extracted in block units from the object data is expanded according to the image data. By comparing with each image data sequentially extracted from within the block in block units, it is sequentially determined whether or not two patterns represented by the two image data to be compared coincide with each other. In a pattern inspection method for detecting a different portion in two inspection target patterns, a) every time one block of image data in an object pattern is extracted from the object data, the ratio of pixels of "1" based on the image data or Check whether the ratio of "0" pixels is larger than a predetermined value near 100%, b) If it is larger than the value, it is determined that the block in the object pattern is not a different portion, and one block of unextracted image data is extracted from the object data as image data of the next block in the object pattern, and c). If the predetermined value or less, the image data of each block in the extended block of the master pattern corresponding to the block in the object pattern is sequentially extracted from the master data as the image data of the comparison,
It is characterized by:

A second pattern inspection method according to the present invention is the same as the first pattern inspection method, wherein a ratio of "1" pixels or a ratio of "0" pixels in one block of the object pattern is smaller than the predetermined value. If i is also large, i) the image data of each block in the extension block of the master pattern corresponding to the block of the object pattern is sequentially extracted, and ii) "1" based on the image data of one block in the extension block. The percentage of "pixels" or the percentage of "0" pixels is 100
% It is checked whether or not it is larger than a predetermined value in the vicinity, and only when it is larger than the predetermined value, it is determined that the block of the object pattern is not a different portion, and image data extraction from the extension block is performed. It is characterized in that the image data for one block that has not been extracted as the image data for the next block in the object pattern is extracted from the object data while being stopped.

[0012]

[Operation of First Pattern Inspection Apparatus] According to the first pattern inspection apparatus of the present invention, object data and master data, which are binary image data obtained by reading two patterns to be inspected, form an image. The difference between the two patterns to be inspected is detected in the following manner based on these data stored in the storage means.

First, the object extraction means sequentially extracts image data from the object data in block units. For each extraction, the object counting means determines the number of pixels of "1" and the number of pixels of "0" in one block of the object pattern based on the extracted image data, and the first control means determines that block. The ratio of "1" pixels or the ratio of "0" pixels in
It is checked whether it is larger than a predetermined value near 0%.

When the ratio of pixels of "1" or the ratio of pixels of "0" is larger than a predetermined value, it can be judged that the block is a blank part or a solid part. And since there are few cases where additional characters and patterns are inserted in the blank part and solid part in the first school which is the target of plate inspection work in the re-school,
Regarding the blank portion or the solid portion, it can be generally considered that the object pattern and the master pattern match. Therefore, when the ratio of the pixels of "1" or the ratio of the pixels of "0" is larger than a predetermined value, it is determined that the block is a matching portion, and the image data of the next block in the object pattern is the object extracting means. Extracted by.

When the ratio of pixels of "1" or the ratio of pixels of "0" is less than a predetermined value, it is determined that the block is neither a blank part nor a solid part, and the object pattern is determined for the block based on the shake comparison method. And the master pattern match.

To this end, first, for the image data extracted in block units by the object extracting means,
By the master extraction means, an extension block whose central block is a block of the master pattern that positionally corresponds to the block of the object pattern related to the extraction is set as the master pattern, and within this extension block, one is two-dimensionally set.
The image data corresponding to each block sequentially shifted pixel by pixel is sequentially extracted from the master data.

The image data sequentially extracted from the master data as each block in the extension block in this manner is compared with the image data extracted from the object data as one block of the object pattern corresponding to the extension block. Are compared, and it is sequentially determined whether or not the two patterns represented by the two image data match. Thus, the process of collating both patterns in block units while correcting the positional deviation between the object pattern and the master pattern within the range of the extended block, that is, the shaking comparison in block units is sequentially performed.

If it is determined as a result of comparison by the collating means that the current block in the object pattern is the matching portion, the extraction of image data by the master extracting means for the current extended block is completed. After that, the image data of the next block is extracted by the object extracting means in order to determine whether or not the next block of the object pattern matches the master pattern. On the other hand, as a result of the comparison by the collating means, if it is determined that all blocks sequentially shifted by one pixel in the extended block do not match, the current block in the object pattern is determined as a different portion, and the object extracting means The image data of the block is extracted.

As described above, it is determined for each block of the object pattern whether or not it matches the master pattern, and when there is no undetermined block of matching / mismatching in the object pattern, the operation of this pattern inspection apparatus is completed. .

[Operation of Second Pattern Inspecting Device] According to the second pattern inspecting device of the present invention, the object data and the master data are stored in the image storing means as in the first pattern inspecting device. The extraction means and the object counting means operate. Then, the number of pixels of "1" and the number of pixels of "0" are obtained based on the image data of one block extracted from the object data, and as a result, the ratio of the pixels of "1" in the block or " When the ratio of 0 "pixels is equal to or less than the predetermined value, the master extraction unit and the collation unit sequentially perform the shaking comparison for the block.

However, unlike the first pattern inspection apparatus, the master extraction means operates even when the ratio of "1" pixels or the ratio of "0" pixels is larger than a predetermined value, and the master counting means operates. , Based on the image data for one block extracted from the master data,
The number of pixels of "1" and the number of pixels of "0" are obtained, and only when the ratio of pixels of "1" or the ratio of pixels of "0" in the block is also larger than a predetermined value, that is, Only when both the block in the object pattern and the one block in the extended block in the corresponding master pattern are blank portions or solid portions,
It is determined that the block in the object pattern is the matching portion, and the extraction of the image data by the master extracting means for the extension block at the present time is completed. After that, the image data of the next block is extracted by the object extracting means in order to determine whether or not the next block of the object pattern matches the master pattern.

[Operation of Pattern Inspection Method] The operations of the first and second pattern inspection methods according to the present invention are as follows.
Similar to the operation of the first and second pattern inspection apparatuses, it is checked for each block of the object pattern whether or not it is a blank part or a solid part, and the blocks that are neither the blank part nor the solid part are sequentially examined. Shake to make a comparison. Thereby, the difference between the two inspection target patterns is detected in block units.

[0023]

【Example】

[Hardware Configuration of Embodiment] FIG. 2 is a diagram showing a hardware configuration of a pattern inspection system which is an embodiment (hereinafter referred to as “embodiment 1”) of the present invention. This pattern inspection system includes a host computer 20 to which a mouse 22 as a pointing device is connected, a scanner 12 and a printer 14, and uses a proof or a plate-making film as a first school and a re-school as an inspection object 10. It is used for plate inspection work in the proofreading stage of the printing process.

In this pattern inspection system, the scanner 12 reads the patterns of the first school and the second school which are the inspection object 10 and generates image data thereof. This image data is transferred to the host computer 20. The image data may be generated by reading the pattern of the inspection object 10 with a television camera instead of the scanner 12.

The host computer 20 is equipped with a CPU, a memory, a CRT display, an I / O port, etc., and stores the image data transferred from the scanner 12 in two.
After the binarization, the binary image data for the first school is stored as the object data, and the binary image data for the re-school is stored as the master data in the internal memories. Then, based on these master data and object data and the input operation by the mouse 22, the processing described later is performed to detect the difference between the first school and the re-school. The data indicating this difference is transferred to the printer 14 as the inspection result.

The printer 14 is a host computer 20.
Based on the inspection result data transferred from, the difference between the first school and the re-school is output. Instead of the printer, a plotter, a CRT display or the like may output the difference between the first school and the second school.

[Difference Detection Processing] In the present embodiment, the processing for detecting the difference between the first school and the re-school (hereinafter referred to as “difference detection processing”) is performed by the CPU based on a predetermined program in the host computer 20. It is performed by operation. The principle of this difference detection processing is basically the same as the principle of the defect detection processing of the conventional example 2 described above, and the comparison is performed by shaking each local area (hereinafter referred to as "inspection block"). The details of the difference detection process of this embodiment will be described below with reference to the flowchart of FIG.

First, in step S10, data obtained by reading an uncorrected pattern, which is the first school of a proof or a plate-making film, is binarized and then stored in a memory as object data. Similarly, in step S12, after the data obtained by reading the corrected pattern that is the proof or the reprint of the plate-making film is binarized,
Stored in memory as master data. In this embodiment,
The positions at which the first school and the second school are placed in the scanner 12 are determined by the pin mechanism, whereby the master pattern represented by the master data and the object pattern represented by the object data are roughly aligned. It is in a state. When the pin mechanism as the alignment means is not used, the alignment may be performed as follows, for example, after reading the patterns of the first school and the re-school as described above. That is, the master pattern and the object pattern are displayed in different colors on the CRT display of the host computer, and the operator looks at this display and uses the mouse 2
A rough alignment may be performed by manipulating 2 to change the relative positions of both patterns.

After the master data and the object data are obtained as described above, in step S14, the data D corresponding to one inspection block is extracted from the object data.
P is extracted (hereinafter, the inspection block from which the data DP is extracted is referred to as "current inspection block", and the extracted data DP
Is called "object inspection data"). Here, the size of the inspection block is selected to be about the minimum character size in the first school and the second school to be inspected.

In the next step S16, the number of pixels of "1" in the extracted object inspection data DP and "
The number of 0 "pixels is calculated.

Then, in step S18, a block of a master pattern of the same size (hereinafter referred to as "master reference block") that corresponds in position to the current inspection block of the object pattern is obtained, and the master reference block is used as a central block for predetermined pixels. Set the area widened by the number as the shaking range, and within the shaking range, 2
One of a plurality of blocks having the same size (hereinafter, referred to as “master block”) that are dimensionally shifted one pixel at a time is selected as a current master block. Since the starting point of this selection in the present embodiment is the master reference block, and the present execution time is the first execution point of step S18, the master reference block is selected as the current master block here.
Then, the data for the current master block (hereinafter referred to as "master inspection data") DM is extracted from the master data.

In the next step S20, the number of pixels of "1" and the number of pixels of "0" in the extracted master inspection data DM are obtained.

In this way, the object inspection data D
As a result of obtaining the number of pixels of "1" and the number of pixels of "0" for P and the master inspection data DM, if all the pixels are "1" or all the pixels are "0" for both data DP and DM, The current object block and the current master block are both blank parts or solid parts, and it can be determined that there is no difference between the object pattern and the master pattern for both blocks. On the other hand, the patterns of the first school and the re-school which are the inspection object 10 are as shown in FIG. 3, for example, and there are many blanks in the peripheral portion of the inspection object 10 and the solid portion in the pattern part etc. There are also many. Then, as mentioned above, the size of the inspection block is selected so that it will be about the minimum character size in the first school and the re-school of the inspection target, so there are blank blocks that do not contain characters between line breaks and lines between sentences. Many occur. Therefore, in order to shorten the processing time by utilizing this point, in step S22, it is determined whether all the pixels of the object inspection data DP and the master inspection data DM are "1" or all the pixels are "0". When it is determined that all the pixels are "1" or all the pixels are "0", the process proceeds to step S32, and it is determined that the current inspection block is the coincident portion. This means that no changes have been made to the first school in the area corresponding to the current inspection block at the re-school. After execution of step S32, the process proceeds to step S36 to check whether or not the next inspection block of the object pattern matches the master pattern.

In step S22, all pixels are "1".
Or, if it is not determined that all pixels are "0",
In step S24, the object inspection data DP
And the two patterns represented by the master inspection data DM are compared and collated. Then, in step S26, it is determined whether or not these two patterns have a certain difference or more. For example, similar to the second conventional example, it is sufficient to compare the above two patterns by a pattern matching method using an inspection window and determine whether there is a difference. That is, an inspection window including a plurality of pixels (for example, 2 × 2)
Pixel-sized window) WP is scanned within the current inspection block, and the same-sized inspection window WM is scanned within the current master block while being positioned corresponding to the inspection window WP, and the inspection window WP is detected at any scanning position. , WM, it may be determined that there is a difference between the two patterns represented by the two inspection data when the values of corresponding pixels included in WM do not match, and it is determined that there is no difference in other cases. According to the pattern matching method using such an inspection window, there is an advantage that it is possible to prevent erroneous determination due to a quantization error at the time of reading the first school and the second school.

If it is determined in step S26 that there is a difference, another master shifted by one pixel in an order of gradually moving away from the master reference block, which is the starting point of the master block selection, within the shaking range. The blocks are sequentially selected as the current master block, and the above processing (steps S18 to S26) is repeated. For this reason, first, in step S28, it is determined whether or not the comparison and collation has been performed for all the master blocks which are two-dimensionally sequentially shifted by one pixel within the shaking range, that is, whether or not the master block shifting has been completed. If it is determined that the shift of the master block is not completed, step S30
Go to. In step S30, the master block to be compared and collated next is selected as the current master block from within the range to be shaken, and the current master block data is extracted as master inspection data from the master data. The start address of is changed by one pixel.
Then, the process returns to step S18 to newly extract the master inspection data DM based on the changed start address. Then, using the master inspection data DM, the current master block and the current inspection block are
It is determined whether it is a blank portion or a solid portion (step S22), and if it is determined that it is neither a blank portion nor a solid portion, the object pattern and the master pattern are compared and collated (step S24). Similarly, as long as it is determined that the current inspection block and the current master block are neither blank nor solid (see step S22) and there is a difference between the two patterns of the current master block and the current inspection block (step S22) (step S22). S26), step S28
→ S30 → S18 → S20 → S22 → S24 → S26 →
The loop of S28 is repeatedly executed. as a result,
If it is determined that there is a difference in all the master blocks that are two-dimensionally shifted one pixel at a time within the shaking range,
In step S34, it is determined that the current inspection block is a different portion. This means that at the re-school, changes were made to the first school within the area corresponding to the current inspection block. After execution of step S34, the process proceeds to step S36 to check whether or not the next inspection block of the object pattern matches the master pattern.

In the above process, steps S20 and S22 are executed only for the master inspection data DM which is first extracted within each shaking range, and the master inspection data extracted after the second time within each shaking range. For the data DM, the process may proceed to step S24 after step S18.

When it is determined in step S26 that there is no difference between the two patterns of the current master block and the current inspection block, the process proceeds to step S32, and it is determined that the current inspection block is the matching portion. This means that in the re-school, there is no change from the first school in the area corresponding to the current inspection block. After execution of step S32, it is checked whether or not the inspection block next to the object pattern matches the master pattern.

For this purpose, first, in step S36, it is determined whether or not all the inspection blocks of the object pattern have been checked for a match / mismatch with the master pattern, that is, whether or not the inspection block has ended.
If the inspection block is not completed, the process proceeds to step S38. In step S38, the next inspection block to be checked for a match / mismatch with the master pattern is selected as the current inspection block, and the object inspection data corresponding to the current inspection block is extracted from the object data. Calculate the start address. here,
The inspection block for which the match / mismatch is to be checked is selected, for example, in the scanning order when the scanner 12 reads the refresh pattern. That is, the inspection blocks are sequentially selected in the main scanning direction at the time of reading, and every time the selection of one inspection block row corresponding to the main scanning line is completed, the position of the inspection block to be selected is one inspection in the sub scanning direction. Move by blocks.

When the start address is calculated in step S38, the process returns to step S14, new object inspection data DP is extracted based on this start address, and the object inspection data DP is used in the same manner as described above. Check if the object pattern matches the master pattern for the current test block. Similarly, as long as there is an unprocessed inspection block in the object pattern, steps S36 → S38 → S14 → ...
The loop of ...-> S36 is repeatedly executed, and when it is determined in step S36 that the inspection block is completed, the difference detection process of this embodiment is completed.

The inspection block determined to have a difference by the difference detection processing (see step S34) corresponds to the difference between the first school and the re-school, and this difference is the printer 14.
In a different color from both the object pattern and the master pattern. The worker looks at such output results and confirms whether the portion instructed to correct for the first school has been correctly corrected, and whether or not there is a changed portion other than the portion instructed to correct.

[Effects of Embodiment] According to the above-mentioned Embodiment 1,
Since the shake comparison for each inspection block is sequentially performed for the object pattern indicating the first school and the master pattern indicating the re-school (steps S18 to S30), the shake comparison is performed by parallel processing. No large-scale hardware like is needed. Further, according to the first embodiment, if both the inspection block and the master block corresponding to the inspection block are blank portions or solid portions, it is determined that the inspection blocks are coincident portions without shaking and comparison ( Steps S22 and S3
2) The time required for the difference detection process can be shortened.

[Other Embodiments] A pattern inspection system which is another embodiment of the present invention (hereinafter referred to as "Embodiment 2") will be described below. This pattern inspection system has the same hardware configuration as that of the first embodiment (see FIG. 2), but the content of the difference detection processing is different from that of the first embodiment. The difference detection processing in this embodiment will be described below.

In the normal proofreading, it is rare that additional characters or patterns are inserted in the blank part or the solid part in the first school in the re-school. Therefore, in the present embodiment, if the inspection block of the object pattern indicating the first school is a blank portion or a solid portion, it is determined that the first school and the re-school are the same for the inspection block, and the master inspection data is not extracted. Therefore, the process shifts to the process for the next inspection block without performing shaking comparison for that inspection block.

FIG. 4 is a flow chart showing the difference detection processing of this embodiment. In this flowchart, as in the case of the flowchart in FIG. 1, two binary image data obtained by reading a first school showing a pattern before correction and a second school showing a pattern after correction are respectively converted into object data representing an object pattern. And as master data representing the master pattern (step S
50, S52), and then, it is checked for each inspection block of the object pattern whether or not it matches the master pattern.

For this purpose, first, in step S54, data of one inspection block is extracted from the object data as object inspection data DP, and then,
In step S56, the number of pixels of "1" and the number of pixels of "0" in the object inspection data DP are obtained.

In the next step S58, it is determined whether or not all the pixels of the extracted object inspection data DP are "1" or all the pixels are "0". Here, all pixels are "1"
Alternatively, if it is determined that all the pixels are "0", the current inspection block is a blank part or a solid part.
Then, it is determined that the current inspection block is the matching portion.
Then, in order to check whether the next inspection block of the object pattern matches the master pattern,
It proceeds to step S74.

If it is not determined in step S58 that all pixels are "1" or all pixels are "0", the number of pixels of "1" in the master inspection data DM is different from the first embodiment. Without performing the processing (steps S20 and S22 in FIG. 1) of determining whether all pixels are "1" or all pixels are "0" by obtaining the number of pixels of "0", the current inspection block Are shaken for comparison (steps S60 to S68). The specific procedure of this shaking comparison is the same as that of the first embodiment, and is two-dimensionally set within the shaking range.
If it is determined that there is a difference in all master blocks that are sequentially shifted pixel by pixel, the process proceeds to step S72, and it is determined that the current inspection block is a different portion. Then, in order to check whether the next inspection block of the object pattern matches the master pattern, step S74 is performed.
Go to.

When it is determined in step S64 that there is no difference between the two patterns of the current master block and the current inspection block, the process proceeds to step S70, and it is determined that the current inspection block is the matching portion. After that, the process proceeds to step S74 to check whether or not the next inspection block of the object pattern matches the master pattern.

Hereinafter, as long as there is an unprocessed inspection block in the object pattern, steps S74 → S76 → S5.
The loop of 4 → ... → S74 is repeatedly executed, and when it is determined in step S74 that the inspection block is completed, the difference detection process of this embodiment is completed.

As described above, according to the present embodiment, if the inspection block is a blank portion or a solid portion, the inspection block is judged to be a coincident portion without shaking and comparison (step). S58, S70),
The time required for the difference detection process can be shortened. Further, as in the first embodiment, since the shaking process for each inspection block of the object pattern is sequentially performed, no large-scale hardware is required.

[Modification] In the first and second embodiments described above, when all the pixels are "1" or all the pixels are "0" in the data of one block extracted from the object data or the master data, Although the block is assumed to be a blank part or a solid part (see steps S22 and S58), "1" is set even if all pixels are "1" or all pixels are not "0".
When the ratio of the pixels of “0” or the ratio of the pixels of “0” is close to 100%, the block may be regarded as a blank part or a solid part. By doing so, it is possible to prevent a block that is originally a blank portion or a solid portion from being erroneously recognized as a blank portion or a solid portion due to a quantization error when reading the first school or the re-school.

[0052]

As described above, according to the present invention, since the shake comparison for each block of the object pattern is sequentially performed, the difference between the two patterns can be achieved by a smaller hardware than the conventional shake comparison. Can be detected.

Further, in the first pattern inspection apparatus and method according to the present invention, when almost all pixels of one block in the object pattern are "1" or substantially all pixels are "0", the second pattern inspection apparatus according to the present invention is provided. In addition to this, in the pattern inspection apparatus and method, when substantially all pixels are "1" or almost all pixels are "0" in one block of the master pattern corresponding to the position, the block of the object pattern is a blank part or a solid pattern. It is shaken as a part and comparison is omitted. Thereby, the time required to inspect the entire two inspection target patterns can be shortened.

[Brief description of drawings]

FIG. 1 is a flowchart showing a difference detection process in a pattern inspection system that is an embodiment of the present invention.

FIG. 2 is a diagram showing a hardware configuration of a pattern inspection system that is an embodiment of the present invention.

FIG. 3 is a diagram showing an example of a pattern of an inspection object in a plate inspection work.

FIG. 4 is a flowchart showing a difference detection process in a pattern inspection system which is another embodiment of the present invention.

FIG. 5 is a view conceptually showing a shake comparison method.

FIG. 6 is a diagram showing a positional shift that accompanies an inclination between an object pattern and a master pattern.

[Explanation of symbols]

 10 ... Object to be inspected (first school and re-school) 12 ... Scanner 14 ... Printer 20 ... Host computer 22 ... Mouse

Claims (4)

[Claims] 1. A pattern inspection apparatus for collating two inspection object patterns and detecting a difference between the two inspection object patterns, wherein: a) two-dimensionally scanning the two inspection object patterns Image input means for generating two binary image data composed of two kinds of values of "1" and "0" by reading each of them, and b) two binary image data generated by the image input means. Image storage means for storing object data representing an object pattern and master data representing a master pattern, respectively, and c) image data corresponding to each block is sequentially extracted from the object data in units of a block composed of a plurality of pixels in the object pattern. Object extraction means, and d) one block of image data by the object extraction means. Object counting means for determining the number of pixels of "1" and the number of pixels of "0" in each block each time is extracted, and e) a master for the image data extracted in block units by the object extracting means. In the pattern, an extension block is set that is two-dimensionally expanded to the periphery by a block that corresponds to the block of the image data as a central block, and two-dimensionally sequentially shifted by one pixel in the extension block. Master extraction means for sequentially extracting the image data corresponding to each block from the master data; and f) the number of "1" pixels and "0" pixels obtained by the object counting means for one block of the object pattern. Based on the number of pixels, the ratio of pixels of "1" or "0"
Is larger than a predetermined value in the vicinity of 100%, and if it is larger than the predetermined value, it is determined that the block of the object pattern is not a different portion, and it is not extracted by the object extraction means. Image data for one block is extracted as image data for the next block,
When the value is less than the predetermined value, the first control means for causing the master extraction means to sequentially extract the image data of each block in the extended block of the master pattern corresponding to the block of the object pattern, and g) by the object extraction means The image data extracted in block units is compared with each image data sequentially extracted in block units from the extension block by the master extraction unit corresponding to the image data, and the image data represented by the two image data to be compared is compared. Collating means for sequentially determining whether two patterns match, and h) if the collating means determines that the two patterns match for any block in the extension block, the extension block by the master extracting means. The extraction of the image data from the inside is stopped and the object extracting means Unextracted as image data of the next block 1
Second control means for extracting image data for blocks,
A pattern inspection apparatus comprising: 2. The pattern inspection apparatus according to claim 1, wherein the number of pixels of "1" and the number of pixels of "0" in the block are based on the image data of one block extracted by the master extracting unit. Further comprising a master counting means for determining, i) the object pattern, when the ratio of the pixels of "1" or the ratio of pixels of "0" in one block of the object pattern is larger than the predetermined value. The image data of each block in the extended block of the master pattern corresponding to the block is sequentially extracted by the master extracting means, and ii) "1" of one block in the extended block obtained by the master counting means. Based on the number of pixels and the number of pixels of "0", it is checked whether the ratio of the pixels of "1" or the ratio of the pixels of "0" is larger than a predetermined value near 100%, and the predetermined value is determined. Only when is larger, it is determined that the block of the object pattern is not a different part, the extraction of the image data from the extension block by the master extraction unit is stopped, and the image data of the next block is sent to the object extraction unit. The pattern inspection apparatus is characterized in that the image data of one block that has not been extracted as is extracted. 3. Two-dimensional image data consisting of two kinds of values of "1" and "0" is generated by scanning two patterns to be inspected two-dimensionally and reading for each pixel, The two binary image data are respectively stored as object data representing an object pattern and master data representing a master pattern, and image data corresponding to each block is sequentially arranged from the object data in units of a block composed of a plurality of pixels in the object pattern. An extended block in which image data extracted and extracted in block units from object data is two-dimensionally expanded to the periphery by a predetermined number of pixels with a block corresponding to a block of the image data in the master pattern as a central block. Is set to sequentially shift two-dimensionally by one pixel in the extension block. Image data corresponding to each block are sequentially extracted from the master data, and image data extracted in block units from the object data are sequentially extracted in block units from the extension block corresponding to the image data. A pattern inspection for detecting a different portion in the two inspection target patterns on a block-by-block basis by a process of sequentially comparing data and two patterns represented by two image data to be compared to determine whether or not they match. In the method: a) Every time one block of image data in the object pattern is extracted from the object data, the ratio of "1" pixels or the ratio of "0" pixels is close to 100% based on the image data. Check if it is larger than a predetermined value, and b) If it is larger than the predetermined value, the object It is determined that the block in the turn is not a different portion, and image data for one block that has not yet been extracted as image data for the next block in the object pattern is extracted from the object data. C) If the image data is less than the predetermined value , Image data of each block in the extended block of the master pattern corresponding to the block in the object pattern is sequentially extracted from the master data as the image data of the comparison target,
A pattern inspection method characterized by the above. 4. The pattern inspection method according to claim 3, wherein when the ratio of the pixels of "1" or the ratio of pixels of "0" in one block of the object pattern is larger than the predetermined value, i) the object The image data of each block in the extension block of the master pattern corresponding to the block of the pattern is sequentially extracted, and ii) the ratio of pixels of "1" or "0" based on the image data of one block in the extension block. "The pixel ratio is 100
% It is checked whether or not it is larger than a predetermined value in the vicinity, and only when it is larger than the predetermined value, it is determined that the block of the object pattern is not a different portion, and image data extraction from the extension block is performed. A pattern inspection method, characterized in that the image data for one block that has not yet been extracted as the image data for the next block in the object pattern is extracted from the object data while being stopped.