JP2006284447A - Defect inspection device and defect inspection method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To cope with detection/classification of a defect on different patterns photographed at different places of an inspection object, and to improve accuracy of an automatic classification result. <P>SOLUTION: When performing detection/automatic classification of a defect of the inspection object 1 having a repeated pattern such as a liquid crystal display device or a semiconductor wafer, the repeated pattern is divided beforehand into a plurality domains, and databases 71, 72, 73 for automatic classification differentiated by a domain on the repeated pattern to which a defect portion detected from an inspection image belongs are created. Then, the defect of the inspection object 1 is classified by using a database which is different corresponding to the domain to which the detected defect belongs. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a defect inspection apparatus and a defect inspection method, and more particularly to a defect inspection apparatus and a defect inspection method for detecting a pattern formed on an inspection target substrate such as a flat panel display and performing a defect inspection.

  Conventionally, a semiconductor device and a flat panel display are produced by forming a fine device pattern on a semiconductor wafer or a glass substrate, respectively. When such a device pattern is formed, dust or the like may adhere to the semiconductor wafer or the glass substrate or may be damaged. A semiconductor wafer or glass substrate in which such a defect has occurred becomes a defective device, and yield is reduced.

  Therefore, in order to stabilize the yield of the production line at a high level, a defect inspection device is used to detect defects caused by dust, scratches, etc. at an early stage, and the cause can be determined, which is effective for production equipment and processes. It is preferable to take appropriate measures. For example, in the case of a flat panel display, if a detected defect type can be repaired, it is repaired and passed to the next process.

  An effective method for determining the cause of the defect is to first classify the defect by examining what the defect is when the defect is discovered. Here, the defect inspection apparatus for examining what a defect is is like an optical microscope, and the defect is magnified to identify what the defect is.

  As a method for automatically classifying defects on a semiconductor wafer, there is a method for extracting defects by comparing a defect image taken with an imaging device and a reference image, and comparing and classifying the defect features with those registered in a database. Widely adopted. In a flat panel display or the like, defects can be automatically classified by the same method.

Here, an example of a conventional defect automatic classification method will be described with reference to FIGS.
FIG. 19 is a process chart for creating a defect classification database in a conventional defect inspection apparatus. In the defect classification database creation step, the defect image (inspected image) and the reference image captured by the imaging device 201 are stored in the defect image memory 202 and the reference image memory 203, respectively. The defect extraction unit 204 compares the defect image with the reference image, and after extracting only the image of the defective part from the defect image, the feature extraction unit 205 quantifies the feature amount such as the size and color of the defect as defect information. .
The digitized defect information is once stored in the pre-classification data memory 206 and then classified by the operator. The operator classifies each defect based on experience, and assigns a classification code to each classified group. The characteristic information of each group to which the classification code is assigned is stored in the classification result memory 212 as information of classifications 1 to N. Then, the feature information of the classifications 1 to N obtained by eliminating the redundancy from the information of the classifications 1 to N stored in the classification result memory 212 by the data selection unit 213 is stored in the database memory 218 as a database.

FIG. 20 is a process diagram of the classification execution process in the conventional defect inspection apparatus described above.
In the classification execution process, as in the database creation process, the defect image and the reference image captured by the imaging device 201 are stored in the defect image memory 202 and the reference image memory 203, respectively. The defect extraction unit 204 compares the defect image with the reference image, and after extracting only the image of the defective part from the defect image, the feature extraction unit 205 quantifies the feature amount such as the size and color of the defect as defect information. .
Then, the extracted feature information of the defect is compared with the feature information of the classifications 1 to N included in the database stored in the database memory 218 by the comparison / classification code assigning unit 219, and the code of the classification with the matching feature is obtained. As a result, the classification result is output from the classification result output unit 220. Thereby, the operator can know the kind of defect.

In addition, an image defect determination processing device has been proposed that divides dissimilarity image data representing a difference between an image to be inspected (defect image) and a reference image into small regions and performs defect determination processing with different threshold values (for example, , See Patent Document 1).
The image defect determination processing apparatus is provided with three types of defect determination processing, namely, a difference pixel calculation process, a maximum difference calculation process, and a total difference calculation process, and sets a threshold value according to the nature of each defect. Therefore, it is possible to classify defects and easily perform defect determination according to the purpose. For example, it is possible to inspect stains on printed matter, LSI pattern defects, stains on the surface of industrial products, coating unevenness, and the like.
JP-A-5-108800

However, since the one described in Patent Document 1 simply defines the area using the difference image data, the inspection image always needs to photograph the same place of the inspection object. There was a problem that it was not possible to deal with defect classification on different patterns taken at different locations.
In a defect inspection at a high magnification called a so-called review inspection, a place where an image to be inspected is taken is not fixed, and a wiring pattern as a base is different every time. For this reason, the image defect discrimination processing apparatus based on the small area address method of the difference degree image data described in Patent Document 1 cannot cope with it.

  The present invention has been made in view of such points, and it is intended to cope with detection and classification of defects on different patterns photographed at different locations of an inspection object, and to improve the accuracy of automatic classification results. To do.

  In order to solve the above-mentioned problems, the present invention first performs an inspection image when extracting and classifying defects by comparing an inspection image obtained by imaging an inspection object having a repetitive pattern with a reference image having no defect. And the reference image are extracted, and the image and position of the defective part are extracted. Next, the feature amount of the extracted defective part image is quantified to generate feature information. Detect which region of the repeated pattern image the position belongs to, and for each region of the repeated pattern image divided into a predetermined number, classify the defects occurring in that region for each similar defect Save it as a database for defect classification, and select the database to be used for defect classification from the database stored for each area based on the area information of the detected defect part. Finally, with reference to the selected database, based on the feature information of the defective portion, characterized by applying the classification code of the defect corresponding with respect to the defect site.

According to the above-described configuration, the inside of the repetitive pattern, which is the minimum unit of the design pattern of the inspection object, is divided into regions having different defect distributions, and defect classification is performed using a different defect classification database for each region. , Classification errors can be reduced.
In addition, since the object to be divided into different areas is a repeated pattern image, it is possible to cope with detection / classification of defects on different patterns photographed at different places on the inspection object.

  According to the present invention, it is possible to cope with detection / classification of defects on different patterns photographed at different places on the inspection object, and to improve the accuracy of the automatic classification result.

Examples of the best mode for carrying out the present invention will be described below, but the present invention is not limited to the following examples.
In this example, the case of performing a defect inspection of a design pattern formed on a glass substrate of a flat panel display will be described. However, the inspection object is not limited to this example. For example, a semiconductor wafer, a photomask, a magnetic disk The present invention can be applied to a substrate in which a predetermined pattern is formed on an inspection target substrate.

First, FIG. 1 is a block diagram showing an example of a configuration of a suitable defect inspection apparatus to which the present invention is applied.
A defect inspection apparatus 100 shown in FIG. 1 performs a classification inspection of defects such as a glass substrate placed as an inspection object on the transfer stage 2. In this defect inspection apparatus 100, the transfer stage 2 positions the placed glass substrate 1 at a designated coordinate position. The imaging device 10 images the inspection object, and images the inspection object on the transfer stage 2, that is, the surface of the glass substrate 1. The imaging output by the imaging device 10 is supplied as an image file to the defect image memory 21 or the defect extraction reference image memory 22.

The defect image memory 21 stores a defect image (inspected image) including an image of a defective part of the glass substrate 1 obtained as an imaging output by the imaging device 10.
When the defect classification database is created and the defect classification is executed, the glass substrate 1 on the transfer stage 2 is moved to the coordinate position of the defect based on the coordinate position data of the defect detected in the defect inspection in the production line of the glass substrate 1. Position to. At that position, defect image data including an image of a defective portion obtained by imaging the surface of the glass substrate 1 by the imaging device 10 is written in the defect image memory 21.

The defect extraction reference image memory 22 stores a reference image having no defect to be compared with the defect image.
When the defect classification is performed, the imaging is performed in a state where the glass substrate 1 is positioned at the same position of a similar pattern having no defect (a position based on the coordinate position data of the defect detected in the defect inspection or a position corresponding thereto). An image obtained by imaging the surface of the glass substrate 1 by the apparatus 10 is written in the defect extraction reference image memory 22 as defect extraction reference image data.

  Writing and reading of each image data to / from the defect image memory 21 and the defect extraction reference image memory 22 described above is controlled by a data processing control unit (not shown).

An example of the reference image (defect-free image) is shown in FIG.
As shown in FIG. 2, a device pattern on the surface of the glass substrate 1 is formed by aligning pixel patterns 110 as unit pixels according to a certain rule.

3 to 7 show examples of defect images including defects detected at the time of defect inspection.
FIG. 3 is an example in which a pattern disconnection defect 121 exists on a part of a device pattern, for example, a signal line pattern. FIG. 4 is an example in which a pattern protrusion defect 122 exists on the signal line pattern of the device pattern. FIG. 5 is an example in which a dust defect 123 exists in a part of the device pattern. FIG. 6 shows an example in which a capacitor (holding capacitor element) formation defect 124 is included in a part of the device pattern. FIG. 7 is a high magnification display of the capacitor formation defect 124 of FIG.
In addition, the said defect image is an example, Comprising: The defect which arises in a flat panel display is not restricted to this example.

  The defect extraction unit 30 extracts an image (difference image) and a position of a defective part by comparing an inspection image obtained by imaging an inspection object with a reference image having no defect. Specifically, the defect image and the defect extraction reference are compared by comparing the defect image data stored in the defect image memory 21 with the defect extraction reference image data stored in the defect extraction reference image memory 22. The difference between the images is extracted as an image of the defective part of the glass substrate 1 imaged by the imaging device 10. The defect part image extracted by the defect extraction unit 30 is supplied to the feature extraction unit 50.

  The feature extraction unit 50 extracts and quantifies information such as size, color, contrast, and shape as feature amounts from the defect site image extracted by the defect extraction unit 30. Then, the digitized information is supplied to the comparison / classification code adding unit 80.

  On the other hand, the defect inspection apparatus 100 of this example includes a pattern matching unit 40 that detects which position on the glass substrate 1 the defect portion corresponds to. There are two detection methods: pattern matching of a defect image with a reference image stored in advance as a target, or conversely, pattern matching of a reference image with a defect image as a target.

When targeting the former reference image, for example, an objective lens having a lower magnification than that used for defect inspection is mounted in a state where the photographing location of the glass substrate 1 is previously positioned at a position free from defects. An image obtained by enlarging the image size of the image taken by the imaging device 10 in accordance with the magnification ratio with the objective lens used for defect inspection is written in the pattern matching reference image memory 23.
The pattern matching unit 40 performs pattern matching of the defect image data stored in the defect image memory 21 using the reference image stored in the pattern matching reference image memory 23 as a target, and detects the position of the defective part.

Alternatively, as another method for targeting a reference image, for example, an objective lens having a lower magnification than that used when defect inspection is performed in a state where the glass substrate 1 is previously positioned at a position free from defects over a wide range is mounted. An image photographed by the imaging device 10 is written in the pattern matching reference image memory 23.
The pattern matching unit 40 targets the reference image stored in the pattern matching reference image memory 23 as a target, and uses the defect image data stored in the defect image memory 21 as a ratio of the magnification with the objective lens used when the reference image is captured. Then, pattern matching is performed after conversion into a defect image reduced in accordance with the position, and the position of the defective portion is detected. The position information of the defective part is supplied to the area code assigning unit 60.
The above-described method of targeting the reference image is useful when the magnification of the objective lens at the time of defect inspection is relatively high with respect to the pixel size, for example, when the imaging range of the defect image is only a part of the pixel ( (See FIG. 7 described later.)

  When targeting the latter defect image, for example, the above-mentioned objective lens having the same magnification as that used for defect inspection is mounted in a state where the photographing location of the glass substrate 1 is previously positioned at a position free of defects. An image obtained by cutting out only a pattern corresponding to one pixel, which is the minimum unit of the repetitive pattern, from the image captured by the imaging device 10 is written in the pattern matching reference image memory 23.

  The pattern matching unit 40 performs pattern matching on the minimum repeated pattern image stored in the pattern matching reference image memory 23 using the defect image stored in the defect image memory 21 as a target, and detects the position of the defective portion. The position information of the defective part is supplied to the area code assigning unit 60.

  The above-described method of targeting a defect image is useful when the magnification of the objective lens at the time of defect inspection is relatively low with respect to the pixel size, that is, when a large number of pixels are captured in the defect image. (See FIGS. 3 to 6 described later.)

  The area code assigning section 60 that constitutes the area detecting section together with the pattern matching section 40 belongs to which area of the area definition image the position of the defective part detected by the pattern matching section 40 is read from the area definition image memory 24. Is detected and a region code is generated. This area code is supplied to the area code output unit 91 and the database selection unit 70. Further, the region definition image is an image obtained by defining the result of the operator resolving the repeated pattern into several regions in advance and converting it into the same size as the pattern matching reference image, and storing it in the region definition image memory 24. Has been.

  Note that the pattern matching unit 40 may divide the inspection image captured by the imaging apparatus 10 into a plurality of pieces and perform pattern matching using the divided inspection images.

  The region code assigning unit 60 examines which region of the region definition image read from the region definition image memory 24 corresponds to the position of the defective part detected by the pattern matching unit 40, and sets the region. A corresponding area code is assigned.

  The database selection unit 70 selects a defect classification database corresponding to the region code supplied from the region code provision unit 60 and supplies the defect classification database to the comparison / classification code provision unit 80. For example, when a region definition image is divided into three regions and a defect is present in region 1, the region code indicating the region 1 is supplied from the region code adding unit 60 to the database selecting unit 70, and the database selecting unit 70 Upon receipt of the region code, the region 1 database 71 is selected and supplied to the comparison / classification code assigning unit 80.

  The database for each area is prepared in advance by the operator by preparing the position information data of defects that may or may only occur in that area for each area. According to the defect classification database creation process as shown, databases are created for the number of existing areas.

For example, in this example, the defect image (inspected image) and the reference image captured by the imaging device 10 are stored in the defect image memory 21 and the reference image memory 22, respectively. The defect extraction unit 30 compares the defect image with the reference image, and after extracting only the image of the defective part from the defect image, the feature extraction unit 50 quantifies the feature amount such as the size and color of the defect as defect information. .
The digitized defect information is once stored in a pre-classification data memory (not shown) and then classified by the operator. The operator classifies each defect based on experience, and assigns a classification code to each classified group. The feature information of each group given the classification code is stored in the classification result memory as information of classifications 1 to N. Then, the feature information of the classifications 1 to N obtained by eliminating the redundancy from the information of the classifications 1 to N stored in the classification result memory by the data selection unit (not shown) is used as a database of defects generated in the area. Stored in database memory.

  For example, when there are three types of defects of category a, category b, and category c in region 1, feature information of category a, category b, and category c excluding redundancy is stored in region 1 database 71. . When there are three types of defects of category a, category e, and category f in region 2, feature information of category a, category e, and category f from which redundancy is eliminated is stored in region 2 database 72. Further, when there are two types of defects of category a and category g in region 3, feature information of category a and category g from which redundancy is eliminated is stored in region 3 database 73.

  At least the memory storing each area database and each reference image memory (storage unit) is a non-volatile memory such as a flash memory.

  The comparison / classification code assigning unit 80 functioning as a classification execution unit compares the feature information of the defect image digitized by the feature extraction unit 50 with the feature information of the defect included in the database selected by the database selection unit 70. , Classify codes with matching features. For example, when the database selection unit 70 selects the database 71, the feature information from the feature extraction unit 50 is compared with the feature information of the classifications a, b, and c included in the database 71, and the classification codes with the matching features are compared. Is granted. The classification code assigned in this way is sent to the classification code output unit 92.

  Next, defect inspection processing by the defect inspection apparatus 100 having the above-described configuration will be described with reference to the flowcharts of FIGS.

  In FIG. 8, first, an image to be inspected, for example, a glass substrate 1 of a flat panel display is photographed by the imaging device 10 to obtain an inspected image (defect image) and stored in the defect image memory (step S1).

  Next, in the defect extraction unit 30, the defect extraction reference image corresponding to the inspection image stored in the defect extraction reference image memory 22 is compared with the inspection image, and a defect site is extracted (step). S2).

  Subsequently, the feature extraction unit 50 extracts the feature amount of the defective part from the information obtained from the extracted image of the defective part (step S3).

  Then, the comparison / classification code assigning unit 80 selects a database to be used at the time of defect classification in accordance with the position of the defective part with respect to the unit pixel pattern (step S4). The process of step S4 may be performed at any time between step S1 and step S5.

Here, the database selection process for defect classification in the process of step S4 will be described with reference to FIG.
First, the pattern matching unit 40 performs pattern matching between the inspection image captured by the imaging device 10 and the pattern matching reference image stored in the pattern matching reference image memory 23 (step S11).

FIG. 10 is an enlarged view of the basic unit pixel pattern 110.
In this example, the pixel pattern 110 is a schematic configuration of one pixel pattern of a glass substrate from which, for example, an active matrix liquid crystal display color filter is removed, and is largely divided into a signal line 111, a scanning line 112, a storage capacitor element ( Capacitor) 113 and pixel electrode 114. As an example of the pattern matching reference image, an image such as the pixel pattern 110 is stored in the pattern matching reference image memory 23. Note that the structure of the pixel pattern of this example is a schematic representation and is not limited to this example.

  Next, the region code adding unit 60 determines which region of the region definition image stored in the region definition image memory 24 the position of the defective part extracted by pattern matching belongs, and generates a corresponding region code. It is given (step S12).

  FIG. 11 is a diagram illustrating an example (1) of a region definition image that is decomposed into four regions, for example, corresponding to the pixel pattern 110 of FIG. In the region definition image 130 shown in FIG. 11, the first region 131 is near the signal line 111, the second region 132 is near the scanning line 112, the third region 133 is near the capacitor 113, and the fourth region is displayed. A region 134 corresponds to a defect generated near the pixel electrode 114.

  FIG. 12 is a diagram for explaining the area detection of the pattern disconnection defect (see FIG. 3). In this case, pattern matching is performed using the image of the one-pixel pattern 110 in FIG. 10 as the pattern matching reference image. In the pattern matching process, it is possible to detect that the defect position belongs to the first region 131 by superimposing the region definition image 130 of FIG. 11 on the matched region and the pixel pattern 110.

  FIG. 13 is a diagram for explaining the region detection of a capacitor formation defect (see FIG. 7) at high magnification. In this case, it can be detected from the region definition image 130 that the capacitor formation defect 124 is in the third region 133.

  FIG. 14 is a table showing an example of defect distribution classified for each defect in the region definition image 130 of FIG.

FIG. 15 is a table showing defect distribution examples in which the defect distribution example shown in FIG. 14 is classified by region.
By creating a defect classification database for each region, the database selection unit 70 can select an appropriate database according to the position where the defect occurs.

  Then, the database selection unit 70 selects a database corresponding to the assigned region code from the database for each region in which the feature information of defects occurring in the region is stored (step S13), and the database selection process Ends.

  According to the example of FIG. 15, when the area definition image 130 is used, since only the pattern defect and dust defect exist mainly in the first area 131, the pattern defect and dust are used as the database for the first area 131. By using a database of defects, it is possible to prevent erroneous classification into other defects, for example, pattern protruding defects.

  After the process of step S3 of FIG. 9 is complete | finished, it returns to the flowchart of FIG. Then, the comparison / classification code providing unit 80 compares the feature information of the defect image supplied from the feature extraction unit 50 with the feature information included in the database selected by the database selection unit 70, and the position where the defect is detected. Depending on the type of defect, the type of defect is classified (step S5). A classification code is assigned to the classified defect and output from the classification code output unit, so that the operator can confirm the position and type of the defect.

According to the above-described embodiment, by dividing the minimum pixel pattern of the repetitive pattern into regions having different defect distributions and performing defect classification using a different defect classification database for each region, the classification result error Can be reduced. Moreover, the present invention can be applied to a case where the defect image is enlarged to include only a part of the pixels.
If only a specific defect needs to be detected (for example, when repair work is performed), the defect inspection process can be speeded up by automatically classifying only the defect that occurs in the area where the defect exists. is there

In addition to the region definition image 130 described above, the pixel pattern 110 may be decomposed as shown in FIG.
An area definition image 140 of the example (2) of the area definition image shown in FIG. 16 is obtained by decomposing the image pattern 110 into three areas, and corresponds to the vicinity of the wiring pattern such as the signal line 111 and the scanning line 112. The region is divided into a first region 141, a second region 142 corresponding to the vicinity of the capacitor 113, and a third region 143 corresponding to the vicinity of the pixel electrode 114.

  FIG. 17 is a table showing an example of defect distribution classified for each defect in the region definition image 140 of FIG. FIG. 18 is a table showing defect distribution examples obtained by classifying the defect distribution examples shown in FIG. 15 into regions. The area definition image can be decomposed into various other areas.

  The present invention is not limited to the above-described embodiments, and various modifications and changes can be made without departing from the scope of the present invention.

It is a block diagram which shows the structural example of the defect inspection apparatus which concerns on one embodiment of this invention. It is a figure which shows an example of a reference image (defect-free image). It is a figure which shows an example of a pattern disconnection defect. It is a figure which shows an example of a pattern protrusion defect. It is a figure which shows an example of a dust defect. It is a figure which shows a capacitor formation defect defect. It is an enlarged view of the principal part of FIG. It is a flowchart which shows the defect inspection process which concerns on the example of 1 embodiment of this invention. It is a flowchart which shows the database selection process for defect classification based on one embodiment of this invention. It is a figure which shows an example of 1 pixel pattern. It is the figure which showed the example (1) of the area | region definition image which concerns on one embodiment of this invention. It is explanatory drawing of the example (in the case of a pattern disconnection defect) of area | region code provision which concerns on the example of 1 embodiment of this invention. It is explanatory drawing of the example (in the case of the capacitor formation defect defect at the time of high magnification) of the area | region code provision which concerns on one embodiment of this invention. It is a table | surface which shows the defect distribution at the time of using the area | region definition image of FIG. 12 is a table showing defect distributions by region in the region definition image of FIG. 11. It is the figure which showed the example (2) of the area | region definition image which concerns on one embodiment of this invention. It is a table | surface which shows the defect distribution at the time of using the area | region definition image of FIG. It is a table | surface which shows the defect distribution according to area | region in the area | region definition image of FIG. It is a creation process figure of the database for defect classification in the conventional defect inspection apparatus. It is a classification execution process figure in the conventional defect inspection apparatus.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Glass substrate (inspection object), 10 ... Imaging device, 21 ... Defect image memory, 22 ... Defect extraction reference image memory, 23 ... Pattern matching reference image memory, 24 ... Area definition image memory, 30 ... Defect extraction , 40 ... pattern matching unit, 50 ... feature extraction unit, 60 ... region code addition unit, 70 ... database selection unit, 71, 72, 73 ... database, 80 ... comparison / classification code addition unit, 91 ... region code output unit , 92 ... Classification code output unit, 100 ... Defect inspection device, 110 ... Pixel pattern, 130, 140 ... Area definition image, 131-134, 141-143 ... Area

Claims (6)

A defect inspection apparatus for extracting and classifying defects by comparing a test image obtained by imaging an inspection object consisting of a repetitive pattern and a reference image without defects,
A defect extraction unit that compares the image to be inspected with the reference image and extracts an image and a position of a defective part;
A feature extraction unit that generates feature information by quantifying the feature amount of the image of the defect site extracted by the defect extraction unit;
An area detection unit for detecting which area of the repeated pattern image the position of the defective part belongs to;
For each area of the repetitive pattern image divided into a predetermined number, a storage unit that classifies defects occurring in the area for each similar defect and stores it as a defect classification database for each area;
A database selection unit that selects a database to be used for defect classification from the storage unit based on the region information of the defect site detected by the region detection unit;
A classification execution unit that refers to the database selected by the database unit and assigns a classification code of the corresponding defect to the defect site based on the feature information of the defect site supplied from the defect extraction unit; A defect inspection apparatus characterized by that. The area detection unit performs pattern matching between the repetitive pattern image stored in advance and the inspection image in the vicinity of the defective portion of the inspection image, and calculates a positional relationship of the defective portion with respect to the repetitive pattern image. The defect inspection apparatus according to claim 1, wherein a region to which the defective part belongs is detected. When the imaging range of the inspected image is a part of the repetitive pattern, the inspected ratio is set so that the imaging ratio of a part of the repetitive pattern becomes the same as that of the reference image with a reference image captured with a lower imaging magnification as a target. The defect inspection apparatus according to claim 1, wherein pattern matching is performed after the image is reduced. When the imaging range of the image to be inspected is a part of the repetitive pattern, pattern matching of the image to be inspected is performed using an image obtained by enlarging the reference image captured at a lower imaging magnification. The defect inspection apparatus according to claim 1 or 2. The defect inspection apparatus according to claim 2, wherein the inspection image is divided into a plurality of pieces, and pattern matching is performed using the divided inspection images. A defect inspection method for extracting and classifying defects by comparing an inspection image obtained by imaging an inspection object consisting of a repetitive pattern and a reference image without defects,
Comparing the image to be inspected with the reference image and extracting an image and position of a defective part;
Generating feature information by quantifying the feature amount of the image of the extracted defect site;
In parallel with the generation process of the feature information, detecting which region of the repeated pattern image the position of the defective part belongs to;
For each area of the repetitive pattern image divided into a predetermined number, the defects generated in the area are classified for each similar defect and stored as a defect classification database for each area;
Selecting a database to be used for defect classification from a database stored for each region based on the region information of the detected defect site;
And a step of referencing the selected database and assigning a classification code of the corresponding defect to the defective part based on the characteristic information of the defective part.

Images