JPH1173513A - Device and method for pattern inspection - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a pattern inspection method and the device that can simultaneously perform a inspection of a pattern superimposed upon a wiring substrate, a print material or the like in a multiplexed way. SOLUTION: A pattern inspection device consists of a line sensor 10, a picture detection part 11, a detected picture storage part 12, a decision part 13, a normal/defective condition decision output part 14, a reference data storage part 15, a reference data automatic generation part 16, an illumination part 17, an illumination/image pickup control part 18, an NC driving part 19 and an NC table 20. The picture detection part 11 AD-converts an image pickup signal of a substrate 21 which is like a reference substrate or an inspection substrate and detects a reference pattern picture or an inspection pattern picture. The reference data automatic generation part 16 automatically generates reference data from the reference pattern picture in accordance with a reference data generation condition parameter. The decision part 13 compares inspection picture data stored in the detected picture storage part 12 with the reference data stored in the reference data storage part 15, detects a defect of the pattern and performs normal/defective condition decision.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern inspection method for imaging a pattern formed on a wiring board, a printed matter, or the like, and comparing the image with reference data created in advance from a reference pattern to detect a pattern defect. And its device.

[0002]

2. Description of the Related Art FIG. 20 shows an example of a wiring board.

Usually, a wiring board is formed by a first step in which a copper foil circuit pattern 2 and a through-hole 3 and a via 4 for forming conduction with a circuit pattern on the lower side of the substrate 1 are formed. In some cases, a gold plating or the like is added to the pad portion 5 for connecting the IC, and then a solder resist 6 is printed on the pad portion 5 and silk printing of the characters 7 and the marks 8 is overlaid thereon. It is manufactured through a number of processes.

[0006] Reference numeral 6E denotes an end of the solder resist 6 (hatched portion).

In these steps, it is necessary to overlap the positions with reference to the circuit pattern, and this is similar to overlapping the plates of the multi-printed material.

In these manufacturing processes, there are various patterns in the course of circuit formation, such as an etching resist pattern and a plating resist pattern. On the other hand, an exposure film and a screen printing plate for forming these patterns are also used. is there. Further, the through hole 3 and the via (via hole) 4 can be regarded as a kind of pattern.

In the formation of these patterns, if any part of the printed wiring board has a defect in the middle of the process, the printed wiring board becomes defective. At important points, the appearance inspection of the above-mentioned various patterns is performed by sampling inspection or total inspection.

The inspection items of the first stage inspection include disconnection, short circuit, chipping, protrusion, pinhole, residual copper, thickening, thinning, deviation of through hole, etc. of the circuit pattern 2.

The inspection items at the second stage include misalignment of printing of solder resist 6, blurring and blurring, misalignment of silk printing (characters 7 and marks 8), blurring and blurring, and pad 5 for IC connection. In the inspection of the first stage circuit pattern such as chipping, protrusion, dirt, unevenness, foreign matter adhesion, etc., an automatic inspection is performed after partial circuit etching.
Inspection performance is still inadequate, such as excessive detection and the need for confirmation after inspection. Further, it cannot be used for various patterns in the various steps described above, and is insufficient in function, performance and cost.

In the second-stage inspection, almost all inspections are performed as a final inspection, and most of them are directly visually inspected or visually inspected using a magnifying glass of about three times. Pad part 5
In order to detect fine defects of about 5 μm to 20 μm, visual inspections using a microscope of 10 to 40 times magnification are increasing.

In addition, the first-stage inspection is often omitted or insufficient in terms of manufacturing cost and the like.
The inspection at the stage also complements the inspection of the circuit pattern 2 at the first stage as a final inspection. In such a case, the visual inspection of the circuit pattern 2 is also performed at the inspection at the second stage.

Recently, semiconductor IC package circuits (PG
A, QFP, BGA, etc.), diversification of wiring boards, high density and miniaturization of circuits have progressed, and circuit pattern width / interval has become 10
0 μm / 100 μm to 50 μm / 50 μm, and the number of circuit defect failures has increased significantly with this. Therefore, it is necessary to enhance the inspection in the first stage and to inspect the circuit pattern 2 in the second stage. Is increasing.

A method of detecting a defect in a pattern by imaging a pattern and comparing the image with reference data prepared in advance from a reference pattern is disclosed in Japanese Patent Application No. 7-3 of the present applicant.
No. 34944. In this conventional method, a plurality of standard substrates are imaged to generate reference data.

[0014]

Defects A such as chipping, projections, dirt, etc. of the pads of the pad portion 5 on the substrate 1 shown in FIG. Defects B such as chips, pinholes, residual copper, etc., are not of a size that is a problem with respect to the circuit pattern width, but are good. Further, the defect C is peeling of the solder resist 6 on the circuit pattern 2 and is small but defective. It is necessary to determine the type of the pattern in which such a defect occurs and the location of the pattern.

In the above-described automatic inspection method / apparatus of the second stage, there is an individual inspection of the solder resist 6 or the silk prints 7 and 8 or a fine inspection of the pad portion 5, but the overall circuit pattern There is no practical one that can be automatically inspected including up to 2.

In the method described in Japanese Patent Application No. 7-334944, there is no detailed setting for a multiplexed pattern, and a high-precision inspection cannot be performed with this alone.

Although there have been various approaches to the automatic inspection method / apparatus of the first stage, there are many excessive detections, and it is necessary to improve the accuracy of inspection determination. Improvement is desired.

Accordingly, these inspections are performed by visual inspection by a person, but have many problems such as low efficiency, nervous fatigue, variation in inspection criteria, and oversight of defects.

Factors that make it difficult to automate these problems are, firstly, in the manufacture of wiring boards, even in non-defective products, misregistration occurs between circuit patterns, solder resists, silk printing, and the like. Second, the inspection of solder resist, silk printing, pad portions, and the like has a large difference in tolerances such as defect size and brightness fluctuation level, and the inspection is difficult at the same time.

In addition, not only the printed circuit board but also the printed material which is printed in a multiplex manner has a difference in plate displacement and a difference in the allowable amount of defects depending on the type and size of the pattern.
Many are considered to be the same issues as above.

For example, in a printed matter, if there is a bright point in a wide area with a dark color, it is conspicuous and judged to be defective. When there is a bright point, or when there is a bright point in a wide area of the intermediate color, it is difficult to notice, so it may be desirable to judge it as a good product. Further, for example, there is a case where the shape and position of a thin line of a specific color in an identification mark or the like have an important meaning, and there is a case where it is desired to inspect more strictly than other portions.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and an object of the present invention is to provide a pattern inspection method and apparatus capable of simultaneously inspecting multiplexed patterns such as wiring boards and printed matter.

[0023]

In order to achieve the above object, according to the first aspect of the present invention, a plurality of pattern types are classified in advance according to a pixel value range of reference image data obtained from a reference pattern. For each pixel of the reference image data, the value of the pixel and a predetermined neighboring pixel are determined by setting the reference data generation condition parameter for each pixel, and then reading the reference image data obtained from the reference pattern. Compared with the pixel value range, if it is within the pixel value range of a single pattern type, it is determined to be a uniform part of the pattern type, and reference data is generated based on the reference data generation condition parameter of the pattern type. If it is not within the pixel value range of one pattern type, the pattern type inside and outside the step portion is determined as a step portion, and the pattern type inside and / or outside the step portion is determined. Reference data is generated based on a quasi-data generation condition parameter, and detected image data obtained by imaging the inspection pattern is compared with the reference data to detect a defect in the inspection pattern.

According to a second aspect of the present invention, in the first aspect of the present invention, a range of neighboring pixels for uniform portion discrimination is set for each of the plurality of pattern types, and the pixel and the predetermined Is compared with the pixel value range of the plurality of pattern types, and if the pixel value is within the pixel value range of the single pattern type, the range of the uniform portion determination neighboring pixels of the pattern type is read and the uniformity is determined. When the range of the neighborhood pixel for the part determination is equal to or less than the range of the predetermined neighborhood pixel, it is determined to be a uniform part of the pattern type, and when the range of the neighborhood pixel for the uniform part determination is larger than the range of the predetermined neighborhood pixel, The value of the neighboring pixel in the large range is compared with the pixel value range of the pattern type, and if the pixel value is within the pixel value range, it is determined to be a uniform portion of the pattern type. Determined as Te, and generating a respective reference data.

According to a third aspect of the present invention, in the first or second aspect, the pixel value is luminance.

According to a fourth aspect of the present invention, in the first or second aspect, the pixel value is a color component value or a color information value such as chromaticity or saturation.

According to a fifth aspect of the present invention, in the first or second aspect of the present invention, the imaging method is a method of detecting a height of an inspection pattern to form an image, and setting the pixel value to a height value. Features.

According to a sixth aspect of the present invention, in the first or second aspect, the pixel value is a non-linear conversion value of the detection data.

According to a seventh aspect of the present invention, in any one of the first to sixth aspects, the reference image data is obtained by imaging a reference pattern.

According to an eighth aspect of the present invention, there is provided a pattern inspection method according to any one of the first to sixth aspects, wherein the reference image data is created from CAD data of the reference pattern. In a ninth aspect of the present invention, in any one of the first to seventh aspects of the present invention, the position of each image data obtained by imaging a plurality of reference patterns is aligned,
The reference image data is generated by performing an average value operation, an intermediate value operation, or an intermediate average value operation between image data for each pixel.

According to a tenth aspect of the present invention, in any one of the first to seventh aspects, the position of each image data obtained by imaging a plurality of reference patterns is aligned, and the position of each image data is set for each partial area. Is performed, and standard image data composed of a combination of image data having high correlation is selected, and the standard image data is used as reference image data of the partial area.

According to an eleventh aspect of the present invention, in the ninth or tenth aspect of the present invention, the method of aligning each of the image data is performed by using a method of determining the amount of positional shift between the image data for each partial area with a resolution of less than a pixel. It is characterized in that a method of measuring and performing interpolation calculation of neighboring pixel values in accordance with the measured amount of positional deviation to obtain each image data after alignment is obtained.

According to a twelfth aspect of the present invention, in any one of the first to eleventh aspects, regarding the uniform portion, a non-defective range of the uniform portion pixel value set in the reference data generation condition parameter of the pattern type. With reference to the reference data, the non-defective range of each uniform portion pixel value set in the reference data generation condition parameter of the pattern type inside and / or outside the step portion and the position shift allowable range of the pattern edge Therefore, the maximum allowable range of the pixel value is obtained for each pixel in the step portion and the neighboring pixels, and is used as reference data.

According to a thirteenth aspect of the present invention, based on the twelfth aspect, the inner and / or outer pattern in the direction orthogonal to the pattern edge is based on the reference data generation condition parameter of the pattern type inside and / or outside the step portion. The width of the pattern is measured, and a position shift allowable range of the pattern edge is determined from the measured pattern width.

According to a fourteenth aspect, in the thirteenth aspect, the inner and / or outer pattern in the direction orthogonal to the pattern edge is based on the reference data generation condition parameter of the pattern type inside and / or outside the step portion. The width is measured, and among these pattern widths, those having a pattern width in a specific range of a predetermined pattern type determine a position shift allowable range of the pattern edge based on a design pattern width. .

According to a fifteenth aspect, in any one of the first to fourteenth aspects, the reference data includes a pixel value upper limit reference value and a pixel value lower limit reference value. Each pixel value is the pixel value upper reference value,
It is characterized in that it is compared with a pixel value lower limit reference value to detect a pattern defect.

According to a sixteenth aspect of the present invention, in any one of the first to fifteenth aspects, the position of the inspection pattern is adjusted based on the pattern position of a specific pattern type of the reference pattern, and the detected image data is referenced. It is characterized in that it is compared with data and a pattern defect is detected.

According to a seventeenth aspect of the present invention, in any one of the first to sixteenth aspects, a predetermined neighboring pixel corresponding to each pixel of the reference image data is a window pixel centered on the pixel. In the case of the stepped portion, the pixel in the orthogonal direction and the pixel in the oblique direction of the window pixel array are compared, and if the difference between the pixel values is smaller than a predetermined value, the pixel in the orthogonal direction is preferentially determined as the edge-side pixel. It is characterized by doing.

According to an eighteenth aspect of the present invention, in any one of the first to seventeenth aspects, the reference data generation condition parameter comprises a mode value, a uniform portion mask upper limit value, and a uniform portion mask lower limit value. Uniform portion mask mode or step portion mask mode composed of mode value, step portion upper limit value, step portion lower limit value, or step portion composed of mode value, pixel value upper limit allowable value, pixel value lower limit allowable value It is characterized by including one of the shift modes.

According to a nineteenth aspect, in the eighteenth aspect of the present invention, the uniform portion mask mode includes setting the uniform portion mask upper limit value and the uniform portion mask lower limit value to the upper limit reference value of the reference data regardless of the previous reference data. Overlap setting in which the upper limit of uniform portion mask and the lower limit of uniform portion mask are updated to the upper limit reference value and lower limit reference value of the reference data only when the forced setting mode is set to the lower limit reference value or when the allowable range of the previous reference data is expanded. It is characterized by comprising a mode.

According to a twentieth aspect of the present invention, in the invention of the eighteenth aspect, the stepped portion mask mode sets the stepped portion mask upper limit value and the stepped portion mask lower limit value to the upper limit reference value of the reference data regardless of the previous reference data. Set to the lower reference value 1
It is characterized by comprising a pixel forced setting mode or a predetermined pixel forced setting mode in which an upper limit reference value and a lower limit reference value of reference data of a predetermined pixel on the center side of the pixel are also set.

According to a twenty-first aspect, in the eighteenth aspect, the step shift mode shifts an edge position of the step by one or more first predetermined pixels in a plus direction or a minus direction. A mode in which a second predetermined image smaller than 1 is shifted in the plus direction or the minus direction by a unit of sub-pixel,
When the edge position is shifted by a first predetermined pixel or a second predetermined pixel in a direction orthogonal to the edge line, an upper limit correction value or a minimum pixel value change obtained by adding an upper limit allowable value to a maximum pixel value change value of the pixel. The lower correction value obtained by subtracting the lower allowable value from the value updates the upper reference value with the upper correction value or updates the lower reference value with the lower correction value only when the allowable range of the previous reference data is expanded. And

According to a twenty-second aspect of the present invention, in any one of the first to twenty-first aspects, the reference image data is divided into a plurality of inspection areas, and a reference data generation condition parameter is provided for each inspection area. The reference data is generated based on the reference data generation condition parameter for each inspection area with respect to the reference image data of the area.

According to a twenty-third aspect of the present invention, in any one of the first to twenty-second aspects, there is provided a step of creating a plurality of the reference image data under different conditions. The first reference data is generated based on the first reference data generation condition parameter, and the second and subsequent reference image data are modified based on the second and subsequent reference data generation condition parameters based on the previous reference data. , The reference data is sequentially updated.

According to a twenty-fourth aspect of the present invention, in the twenty-third aspect of the present invention, a plurality of reference patterns having variations are used as the different conditions.

According to a twenty-fifth aspect, in the twenty-third aspect, different illumination conditions are used as the different conditions.

According to a twenty-sixth aspect of the present invention, in the twenty-third aspect of the present invention, a different color component or a color detection condition such as chromaticity is used as the different condition.

According to a twenty-seventh aspect of the present invention, in the twenty-third aspect of the present invention, different detection conditions based on light conversion characteristics of the object to be detected, such as polarization characteristics or fluorescence characteristics, are used as the different conditions.

According to a twenty-eighth aspect of the present invention, in any one of the twenty-third to twenty-seventh aspects, the positions of the second and subsequent reference patterns are determined based on the pattern position of a specific pattern type of the first reference pattern. In addition, the reference data is sequentially updated.

According to a twenty-ninth aspect, in any one of the first to twenty-eighth aspects, a specific area number is added to the reference data for a specific area, and the detected image data is compared with the reference data. When detecting a defect in the pattern, if a specific area number is added to the reference data, the pattern is inspected according to the determination condition of the specific area number.

According to a thirtieth aspect of the present invention, in the invention of the twenty-ninth aspect, the condition for judging the specific area number is a variation width of a pixel value, a variation width of a high-frequency component of a pixel value, or a variation width of a low-frequency component of a pixel value. In the specific area, these re-determinations are performed to detect pattern defects.

According to a thirty-first aspect of the present invention, in the thirty-ninth aspect, the condition for determining the specific area number is set as a matching mode with a good determination pattern, and the detected image data is compared with the reference data in the specific area. If it is determined to be a defect, the re-determination is performed by comparing it with a previously determined good determination pattern, and if it is determined to be good, it is excluded from the defect.

According to a thirty-second aspect of the present invention, in the thirty-ninth aspect, the determination condition of the specific area number is set as a position correction comparison determination mode, and the detected image data is compared with the reference data in the specific area to determine a defect. When the determination is made, the positional relationship between the detected image data and the reference data is shifted by a predetermined amount, and the comparison determination is performed again.

According to a thirty-third aspect, in the thirty-ninth aspect, the determination condition of the specific area number is set as a position correction comparison determination mode, and the detected image data is compared with the reference data in the specific area to determine a defect. If it is determined that the positional deviation of the specific object in the detected image is detected, the reference data is shifted to and set in a position correction area corresponding to the positional deviation, and the position correction area is excluded from the specific area. In the partial area, the reference data of the contour of the specific area is shifted and set, and the detected image data in the specific area and the position correction area are compared with the shifted reference data, and If it is determined that the defect is determined, it is excluded from the defect.

According to a thirty-fourth aspect of the present invention, there is provided a method according to any one of the twenty-ninth to thirty-third aspects, wherein the specific area is set based on a uniform portion or a step portion of a predetermined pattern type. Set a specific area number in the reference data generation condition parameter of the pattern type,
When the reference data is generated, the specific area number is added to the reference data of an area based on the uniform portion or the step portion of the pattern type or an area obtained by enlarging or reducing this area by a predetermined distance.

According to a thirty-fifth aspect of the present invention, in any one of the twenty-ninth to thirty-third aspects, the teaching pattern of the specific area is imaged, and the position of the specific area is taught to correspond to the specific area. It is characterized in that a predetermined value is set in the reference data or a specific area number is added to the reference data.

According to a thirty-sixth aspect of the present invention, in any one of the thirty-ninth to thirty-third aspects, the reference image data is read, a partial pattern having a predetermined characteristic is searched for, and the detected position of the partial pattern is used as a reference. To determine a specific area,
It is characterized in that a predetermined value is set in the reference data corresponding to the specific area, or a specific area number is added to the reference data.

According to a thirty-seventh aspect, in the thirty-sixth aspect, the predetermined feature is any one of a pattern shape, luminance, chromaticity, distribution and combination thereof.

According to a thirty-eighth aspect of the present invention, an image detecting means for imaging an inspection pattern or a reference pattern, an illumination / imaging control means for controlling the illumination and the imaging, and storing the sensed detected image data or the reference image data. Detection image storage means, reference data automatic generation means for generating reference data, reference data storage means for storing the generated reference data, and determination means for detecting the pattern defect by comparing the detected image data with the reference data; The reference data automatic generation means classifies in advance into a plurality of pattern types according to a pixel value range of reference image data obtained from a reference pattern, and sets a reference data generation condition parameter for each pattern type. Next, for each pixel of the reference image data obtained from the reference pattern, the value of the pixel and a predetermined neighboring pixel If it is within the pixel value range of a single pattern type compared to another pixel value range, it is determined to be a uniform part of the pattern type, and reference data is generated based on the reference data generation condition parameter of the pattern type, and If it is not within the pixel value range of a single pattern type, the pattern types inside and outside the step portion are determined as the step portion, and the reference data generation condition parameters of the pattern types inside and / or outside the step portion are determined based on the reference data generation condition parameters. It is a means for generating data.

According to a thirty-ninth aspect of the present invention, in the thirty-eighth aspect of the present invention, the position of each image data which is obtained by imaging a plurality of reference patterns and is stored is calculated for each pixel or partial area. It is characterized by comprising reference image data creating means for obtaining reference image data.

According to a fortieth aspect, in the thirty-eighth or thirty-ninth aspect, a specific area number is added to the reference data for a specific area, and the detected image data is compared with the reference data to determine the pattern. When detecting a defect, if a specific area number is added to the reference data, a specific area determining means for inspecting a pattern according to the determination condition of the specific area number is provided.

[0062]

BEST MODE FOR CARRYING OUT THE INVENTION

(Embodiment) FIG. 1 shows the configuration of a pattern inspection apparatus using the pattern inspection method of the present invention,
0, image detection unit 11, detected image storage unit 12, determination unit 1
3, a pass / fail judgment output unit 14, a reference data storage unit 15, a reference data automatic generation unit 16, an illumination unit 17, an illumination / imaging control unit 18, an NC drive unit 19, and an NC table 20.

The NC drive section 19 is for moving a substrate 21 such as a reference substrate or an inspection substrate set on the NC table 20 in the Y-axis direction in synchronization with the line sensor 10. Further, an XYθ axis fine movement mechanism for positioning the substrate 21 is included as necessary.

The line sensor 10 is driven by the N-axis
The substrate 21 such as a reference substrate or an inspection substrate on the C table 20 is imaged.

The illuminating section 17 illuminates to obtain a good detected image.

The illumination / imaging control unit 18 controls the switching of one or a plurality of illuminations, the resolution of imaging, the switching control of the visual field, and the like.
In conjunction with the image detecting unit 11, the image sensing unit 11 controls imaging under various lighting / imaging conditions.

The image detecting section 11 converts an image signal of a substrate 21 such as a reference substrate or an inspection substrate into an analog signal to detect a reference pattern image or an inspection pattern image.

The detected image storage section 12 stores the detected reference image data or detected image data.

The reference data automatic generation section 16 automatically generates reference data from a reference pattern image in accordance with a reference data generation condition parameter described later.

The reference data storage section 15 stores automatically generated reference data.

The determination section 13 compares the inspection image data in the detected image storage section 12 with the reference data in the reference data storage section 15, detects a pattern defect, and makes a pass / fail judgment.

The pass / fail judgment output unit 14 outputs pass / fail judgment data, defect position data, defect type data, defect image data, and other inspection result information.

FIG. 2 shows an example of a reference data creating method of the pattern inspection method of the present invention.

The method of creating the reference data shown in the figure is as follows. The reference image data 32 is acquired by imaging 31 of the reference pattern 30, and the reference data generation condition parameters set in advance are read out while reading the reference image data 32 for each pixel. In this method, reference data 35 for each pixel is generated 34 in accordance with the reference data 33. Alternatively, a plurality of reference patterns 30 can be imaged, and the reference data 35 corresponding to the variation of the reference patterns 30 can be sequentially updated.

In this case, there is a possibility that a pattern shift may occur. However, by sequentially overlapping the first reference pattern with reference to a pattern position of a specific pattern type and generating reference data, at least a specific pattern The amount of variation can be taught without lowering the accuracy of the comparative inspection.

Also, as shown in FIG.
Using the AD data 41 and 42, CAD data 4 of each pattern divided into each process such as circuit, hole processing, resist printing, etc.
The reference image creation processing 43 may be performed by conversion, combination, synthesis, or the like of 1, 42, and the target reference image data 32 may be created by combining the pixel value levels 44 such as the luminance level of the detected image.

The CAD data includes similar data such as CAM data for manufacturing a substrate.

As another example, it is difficult to manufacture a reference pattern that is a perfect non-defective product when a reference pattern is imaged to obtain reference image data. In some cases, it is difficult to perform the operation, and a method of creating reference image data in such a case will be described.

FIG. 4 shows the configuration of the reference image data creating section. In the pattern inspection apparatus shown in FIG. 1 and the method of creating reference data shown in FIG. 2, this section corresponds to a section for creating reference image data from a plurality of reference patterns. I do.

In the first method, a plurality of reference patterns 30 ₁ to 30 _N are respectively imaged in FIG. 4 (31), and each reference pattern image is stored (45). (46), and an average value calculation, an intermediate value operation, or an intermediate average value operation between the respective image data is performed for each pixel (47). Is created. For example, if there are three or more reference patterns and an intermediate value is selected for each pixel, appropriate reference image data can be created relatively easily. The reference pattern is 4
If the number is equal to or more than one and the average value of the intermediate data excluding the data at both ends is selected for each pixel, appropriate reference image data can be created with higher accuracy.

[0081] The second method, in Figure 4 as in the first method, by imaging a plurality of reference patterns 30 ₁ to 30 _N, respectively (31), stores the respective reference pattern image (45) Then, align the position of each reference pattern (4
6) Perform correlation calculation between the image data for each partial area (47), select standard image data from a combination of image data with high correlation, and set it as reference image data (32) for the partial area. . In this method, the reference image data in the partial area is easily obtained smoothly for the processing for each partial area.

In the third method, a plurality of reference patterns are respectively imaged and each of the reference pattern images is stored. Then, the positional deviation between the respective reference pattern images for each partial region is determined by sub-scanning using a normalized correlation method or the like. Measurement is performed in pixel units (units smaller than pixels), and interpolation of neighboring pixels is performed according to the measured amount of positional deviation to obtain each reference pattern image after alignment,
From these, optimal reference image data is created in the same manner as in the first and second methods.

Although the present embodiment describes the pixel value of the detected image data as luminance, the present invention relates to inspection of a color pattern in which the pixel value is a color component value or a color information value such as chromaticity or chroma, and the like. It can also be applied to the inspection of a three-dimensional pattern in which the height is detected, an image is formed, and the pixel value is a height value.

The pixel value is a value obtained by nonlinearly converting the detected data, the luminance information, the color information, and the like. Alternatively, a value obtained by combining height information may be used.

FIG. 5 shows an example of conversion of detection data. The horizontal axis shows the detection data, and the vertical axis shows the conversion data. a is a normal linear transformation. b is a non-linear conversion in which the gain of the low pixel value portion is increased and the gain of the high pixel value portion is decreased, and when the circuit pattern under the solder resist is detected brightly for pattern inspection, or when the solder resist is peeled, blurred, or damaged. It is suitable for detecting a defect with a minute change in brightness such as adhesion of a foreign substance. c is a non-linear conversion in which the gain of the low pixel portion is reduced and the gain of the high pixel value portion is increased, and is suitable when a delicate surface condition inspection such as peeling, scratching, stains, dirt, and foreign matter adhesion of the pad portion is important. . d is a non-linear conversion in which the gain of a plurality of sections is further increased, and is suitable for detecting pixel values of various patterns in a well-balanced manner and detecting each defect with high accuracy. Although these non-linear conversions may be ordinary gamma correction conversions, the above-described polygonal line conversions are easier to manage the conversion formula of each section and are easier to use.

Next, a method of comparing and inspecting the inspection image data with the reference data in the judgment section 13 of FIG. 1 will be described.

The reference data in the reference data storage section 15 mainly includes a luminance upper limit reference value and a luminance lower limit reference value corresponding to each pixel, and a specific region number may be added.

As shown in FIG. 6, the judgment unit 13 includes a primary judgment unit 13A and a secondary judgment unit (not shown).

In the comparative inspection, first, the inspection pattern image is picked up by the imaging unit 48 composed of the line sensor 10, the image detection unit 11, and the detection image storage unit 12, and the position is matched with the reference pattern image. In the primary determination unit 13A, the luminance value of each pixel of the inspection image data is
If the reference data stored in the reference data storage unit 15 for each pixel is within the range of the upper luminance reference value and the lower luminance reference value, it is determined to be non-defective. Pixels exceeding this range are regarded as defective candidate pixels. To detect.

For the pixels to which the specific area numbers have been added, pass / fail judgment processing by the specific area number position judging section 49 and judgment processing of the specific area numbers by the specific area judging sections 50 ₁ to 50 _N are also performed to judge pass / fail.

After making the above-described primary determination, the secondary determination unit determines the center of the pixel determined as a defect candidate in the spatial filtering process or the primary determination process shown in Japanese Patent Application No. 7-334944, for example. If the defective candidate pixel is determined to be non-defective by a secondary determination method of determining a defect / non-defect based on the detection pattern data of the peripheral pixel group, the pixel is excluded from defects.

Further, the defect candidate pixels are grouped with neighboring defect candidate pixels and converted into defect point data (defect point position data, defect point size data, defect point type data, etc.), and then pass / fail judgment output is performed. Output from the unit 14.

Here, the positioning is performed based on a specific pattern type serving as a position reference such as a circuit pattern, and a hole shift, a solder resist shift, and a silk printing shift are detected.

Next, the method of generating the reference data will be described in detail with reference to the wiring board 21 shown in FIG.

First, as shown in FIG.
In addition to the circuit pattern 2, the pattern of the solder resist 6 and the silk pattern (characters 7.
There are different parts such as marks 8) and through holes 3, lands 9, IC connection pads 5, vias 4 and the like, and these are determined by the luminance level of the detected image and stratified. This is called material identification.

6E indicates an end of the solder resist 6 (hatched portion).

FIG. 7B shows a luminance profile of the X line. As described above, the pattern type is determined for all the pixels based on the luminance value range on the vertical axis.

Each pattern includes a broad large pattern and a narrow pattern with a narrow width, and various patterns such as a pattern central portion (luminance uniform portion (same as a luminance flat portion)) and a pattern edge portion (luminance step portion). Although there are parts, the distance from the edge is measured by searching for neighboring pixels for each pixel, and these various pattern parts are determined.

A method of automatically generating reference data including a luminance upper limit reference value, a luminance lower limit reference value, and the like based on reference data generation condition parameters set in advance by a combination of the above-described material determination and pattern region determination will be described.

First, one or more inspection areas are set, and reference data generation condition parameters are set for each inspection area.

In FIG. 7A, the inspection area is divided into an inspection area A excluding the pad and an inspection area B including the pad in order to more strictly inspect the IC connection pad section 5.

Tables 1 and 2 show reference data generation condition parameter setting tables.

[0103]

[Table 1]

[0104]

[Table 2] Table 1 is a setting table of the inspection area except the IC connection pad.
Table 2 is a setting table of the inspection area B of the IC connection pad.
For each inspection area, the target pattern is set to a luminance level (0 to 1)
(00%), the layers are arranged in descending order of priority, the upper limit luminance value ap and the lower limit luminance value bp, which are the luminance ranges of the respective layers, are set, and the reference data generation condition parameters are set for each layer. The reference data generation condition parameters include a uniform portion mask mode MCp (0, 1, 2), a uniform portion mask upper limit value MChp, a uniform portion mask lower limit value MClp, an edge mask mode MEp (0, 1, 2, 3), and an edge. Mask upper limit value MEhp, edge mask lower limit value MElp, two-pixel shift mode S2p (0, 1, 2), one-pixel shift mode S1
p (0,1,2), sub-pixel shift mode S0p (0,
0.1 to 1.0 pixel), upper limit allowable value αp, lower limit allowable value β
p, a specific area number Zp (1, 2, 3) and the like. Each mode set value 0 indicates mode off. For the specific area number, a specific area determination parameter is set as a separate table for each number. Table 3 shows the specific area determination parameters.
As shown in (1), it is composed of a specific area number, a brightness variation determination mode, a variation width determination reference value, and the like. The brightness variation determination mode indicates a determination for each of high and low frequency components.

[0105]

[Table 3] In the high-frequency determination mode, the determination method of the specific region is to compare and determine the magnitude of the difference value between the pixel values in the specific region with the fluctuation width reference value, and to determine the average of 3 × 3 pixel values in the medium-frequency determination mode. The magnitude of the variation value in the specific area is compared with the variation width reference value, and in the low frequency determination mode, the magnitude of the variation value in the specific area of the average of 7 × 7 pixel values is compared with the variation width reference value. Has been determined.

The reference data automatic generation method will be described in more detail.

As shown in FIG. 8A, the reference image data is sequentially read out for each of the inspection areas a, b, and c obtained by dividing the entire inspection area of the reference image data 32, and the luminance level of each pixel and neighboring pixels is determined. Then, the reference data generation condition parameters are compared with the reference data generation condition parameters of the inspection area, and the reference data 35 including the luminance upper limit reference value and the luminance lower limit reference value is generated for each pixel as shown in FIG.

The reference data 35 can be updated a plurality of times from the plurality of reference image data 32. In this case, the reference data 35 is aligned with the first reference pattern image and compared with the previously created reference data 35. The part that needs to be updated is determined, and the reference data 35 is updated. FIG. 8B shows R _ij (B
FIG. 8C shows a window of 3 × 3 pixels, and FIG. 8D shows reference data corresponding to R _ij (Bc0).

In the flowchart of FIG. 9, the pixel R
_ij (Bc0) is read out and window pixels W0 to Wk of 3 × 3 pixels are read out, and the maximum brightness W
kmax and the minimum luminance value Wkmin are obtained.

Next, the central pixel luminance value (Bc0 = W0) is compared with a luminance value range (bp to ap) for each pattern type.
A pattern type p that satisfies bp ≦ Bc0 ≦ ap is obtained.

In this pattern type p, Wk
If min ≧ bp and Wkmax ≦ ap, the pixel is determined to be a uniform luminance portion. If this is not satisfied, it is determined to be a luminance step.

As shown in the flowchart of FIG.
In the case of the uniform brightness portion, the mask mode MCp of the uniform brightness portion is set.
Check, if mask mode MCp = 2, forced setting
As a mode, regardless of the previous allowable range,
The upper limit set value MChp is set to the luminance upper limit reference value U._ijSet to
The uniform portion mask lower limit set value MClp is changed to the brightness lower limit reference value L. _ij
Set to.

If the mask mode MCp = 1,
Only in the direction to increase the previous allowable range as the overlap setting mode
Update. That is, the upper and lower limit correction values are defined as U = MChp, L
= MChp and the uniform portion mask upper limit set value (upper limit correction value
U) is the previous luminance upper limit reference value U_ijOnly when greater than
The luminance upper limit reference value U is added to the partial mask upper limit set value MChp._ijTo
Updated and the uniform portion mask lower limit set value (lower limit correction value L) is
Luminance lower limit reference value L _ijUniform mass only when smaller
Luminance lower limit reference value L_ijUpdate
You.

The uniform portion mask mode is off (MCp = 0)
In the shift mode S2p = 0 and S1p = 0, the upper limit correction value U is set to the luminance value Bc0 of the center pixel + the upper limit allowable value αp, and the lower limit correction value L is set to the luminance value Bc0 of the center pixel.
A lower limit allowable value βp, and based on the upper and lower limit correction values U and L, when the upper limit correction value U is larger than the previous luminance upper limit reference value Uij, the luminance upper limit reference value U _ij is defined as U _ij = U. update, updating and = L only luminance lower reference value L _ij when the upper limit value correction value L becomes smaller than the previous brightness lower reference value L _ij. When the uniform portion mask mode is off (MCp = 0) and the shift mode S2p ≠ 0 or S1p ≠ 0,
The upper and lower limit correction values are U = Wkmax + αp, L = Wkmin−βp
Only when the upper limit correction value U becomes larger than the previous upper limit luminance value, the upper luminance reference value U _ij is calculated as U _ij = Wkmax + α.
p, and only when the lower-limit correction value L is smaller than the previous lower-limit luminance reference value, the luminance lower-limit reference value L _{ij is changed} to L _ij = Bc0.
Update to -βp.

As described above, in the shift mode, the upper and lower limit luminance values are updated only when the previous allowable range is expanded.

Next, if a specific area number of the pattern type is set for the uniform portion, the area number Zp is set to the specific area number of the reference data.

At this time, depending on the specific area number, in order to remove a relatively large change in the luminance of the peripheral portion in the uniform portion and to perform a more stable determination, a 5 × 5 pixel including the outer peripheral pixels of the 3 × 3 window is used. Maximum brightness W2k of window pixel
max and the minimum luminance value W2kmin, and W2km ≦ b
If p or W2kmax ≧ ap, it is excluded from the uniform portion and no specific region number is added.

Here, another example of the determination of the uniform luminance portion will be described. For example, since the pattern edge position varies greatly depending on the pattern type such as the silk print pattern and the solder resist pattern, it is desirable to increase the range for determining the luminance uniform portion so as not to include a portion close to the pattern edge portion.

In such a case, the range M of the uniform portion determination neighboring pixels is set in advance for each of a plurality of pattern types.
First, assuming that a predetermined neighboring pixel to be used first is a 3 × 3 pixel window, 3 × 3 window pixels W0 to Wk centered on the pixel R _ij (Bc0) are read out for each pixel, and A maximum luminance value Wkmax and a minimum luminance value Wkmin are obtained.

Next, the central pixel luminance value (Bc0 = W0) is compared with a luminance value range (bp to ap) for each pattern type.
A pattern type p that satisfies bp ≦ Bc0 ≦ ap is obtained. Then, in this pattern type p, Wkmin ≧ b
If p, Wkmax ≤ ap, the range M of the uniform area discriminating pixel of the pattern type is read. If M is 3 or less, the area is determined to be a uniform area. The maximum luminance value and the minimum luminance value of the window pixel are obtained. If these values are also within the above-mentioned luminance value range (bp to ap), the pixel is determined to be a luminance uniform part. If these are not satisfied, it is determined to be a luminance step portion.

Next, a description will be given of a case where it is determined that the pixel is a luminance step portion.

FIG. 11 is a flowchart of the reference data automatic generation of the luminance step portion.

First, for the luminance step portion pixel R _ij (Bc0), it is determined whether the inner pattern is low to high and the edge direction is determined from eight directions.

FIGS. 12A to 12H show the center pixel (Bc).
0), the edge-side pixels Be1 in different edge directions (eight directions).
And the arrangement of the center pixels Bc1...

FIGS. 13A to 13D show the state of the step portion. The vertical axis indicates the luminance level, and the horizontal axis indicates continuous pixels in the direction orthogonal to the edge line.

The steps 1 and 2 perform the lower and upper limit generation processing of the inner pattern, and the steps 3 and 4 perform the upper and lower limit generation processing of the inner pattern. This determination is made based on whether the inner pattern luminance, that is, the central pattern luminance is lower or higher than the edge pattern luminance. The steps 2 and 4 have an intermediate luminance level, but determine whether the level is low or high.

In the case of the internal pattern low-order upper / lower limit data generation processing, first, the edge pixel having the maximum luminance value is determined from the window pixels, and the edge direction is determined. At this time, in order to increase the edge direction discrimination accuracy, the weight of the orthogonal pixels is set slightly larger than that of the oblique pixels. An appropriate weight is about 10 to 20% of (maximum luminance value-minimum luminance value) of this window pixel.

When the edge direction is determined, the stepped portions are divided into the central pixels Bc0 to Bc5, Bc1 ',.
Bc2 ′, Bc2 ″ and edge-side pixels Be1, Be2
Is determined.

Next, the inner pattern type is determined from the minimum luminance values of Bc0, Bc1, and Bc2, and the outer pattern type is determined from the maximum luminance values of Be1 and Be2. Then, the inner pattern type and the outer pattern type are compared, and the one with the higher priority is determined as the upper pattern type.

The edge mask mode MEp of the upper pattern type p is checked, and the predetermined pixel compulsory setting mode E is set.
In the pixel forced setting mode, when MEp = 1, the edge mask upper limit value MEhp is set to the upper limit reference value U _ij of the center pixel Bc0, and the edge mask lower limit value is set to the center pixel Bc0.
_Is set to the lower-limit reference value L1 _ij .

Edge mask mode 1 when MEp = 2
In addition, the upper limit value of the edge mask is set to the center pixels Bc1, Bc.
The upper limit reference values Uc1 and Uc1 'are set to 1', and the lower limit values of the edge mask are set to the lower limit reference values Lc1 and Lc1 'of the center pixels Bc1 and Bc1'. When MEp = 3, in addition to the edge mask mode 2, the upper limit value of the edge mask is set to the upper limit reference value Uc2 of the center side pixels Bc2, Bc2 ′, and Bc2 ″.
Uc2 ', Uc2 ", and the lower limit value of the edge mask is set to the lower limit reference value Lc of the center pixels Bc2, Bc2', Bc2".
2, Lc2 'and Lc2 ".

The edge mask mode is off (MEp = 0)
In the case of, the shift modes S2, S1, and S0 are checked.

If the inner pattern is low, the processing shown in the flowchart of FIG. 14 is performed.

When the shift mode S2p = 2 or S1p = 2 or S1p = 1 and the inner pattern is continuous for three pixels, the upper limit correction value U of the center pixel Bc0 (U = edge side pixel luminance B
When e1 + the outer pattern upper limit allowable value αe) expands the previous allowable range, the upper limit reference value Uij is changed to the upper limit correction value U by (U _ij =
Be1 + αe), and the lower limit correction value L of the center pixel Bc0 (L = center pixel luminance Bc1−inner pattern lower limit allowable value βc) is the lower limit reference value L when the previous allowable range is expanded.
_ij is updated with the lower limit correction value L as (L _ij = Bc1−βc). Here, three consecutive pixels are determined to be continuous if Bc2 and Bc3 are in the inner pattern. In addition to the above, when the shift mode S2p = 2 or S2p = 1 and the inner pattern is continuous for 5 pixels, the upper limit correction value U (U = edge side pixel luminance Be2 + outer pattern upper limit allowable value αe) of the center pixel Bc0 is previously allowed. To widen the range, the upper limit reference value U _ij is updated with the upper limit correction value U to (U _ij = Be2 + αe), and the center pixel B
If c0 of the lower limit correction value L (L = center side pixel luminance Bc2- inner pattern and lower tolerances .beta.c) extending the previous tolerance, the lower limit reference value L _ij at the lower limit correction value L (L _ij = Bc2-
βc).

Here, it is determined that five consecutive pixels are continuous if Bc2 to Bc5 are within the inner pattern.

After updating L _ij = Bc2−βc, Uc1
If <Be1 + α, the upper limit value of the center pixel is Uc1 =
It is updated to Be1 + αe. And after this update or U
Unless c1 <Be1 + α, Uc1 ′ <Be1 + αe
Is determined, and if Uc1 ′ <Be1 + αe, the upper limit value of the center pixel is updated to Uc1 ′ = Be1 + αe, and the process returns. If Uc1 '<Be1 + αe, the process returns. When S2p = 0 when the shift mode S2p is checked, the process also returns. By the way, the shift mode S0p is checked when the shift mode S1p = 1 and the three pixels are discontinuous or when S2p = 0 or 1 and S1p = 0. S0p = s is a numerical value from 0 to 1 in subpixel units.

When s ≠ 0, the maximum luminance value Bmax and the minimum luminance value Bmin of the central pixel Bc0 when the edge position is shifted by ± s pixels as shown in FIGS. 15A and 15B are calculated, and The value obtained by adding the upper limit allowable value of the outer pattern to the maximum luminance value is U, the value obtained by subtracting the lower limit allowable value of the inner pattern from the minimum luminance value is L, and the upper limit value of the central pixel is (U = Bmax + α).
e) and lower the center pixel lower limit value to (L = Bmin
-Βc). When U _ij <U, U _ij =
U is updated, and when L _ij > L, L _ij = L is updated, and other than these are not updated. After performing such correction and update processing, the process returns.

When the shift mode S0p is checked, s =
If the value is 0, the routine immediately returns without performing the correction and update processing.

FIG. 15A shows a case where the edge position is shifted within one pixel, and FIG. 15B shows a case where the edge position is shifted to an adjacent pixel.

When the edge direction is an oblique direction, s is not proportional to the pixel luminance change, but the pixel luminance value is obtained from the edge position for the pixel in the lookup table.

FIG. 16 is a flowchart for the case where the inner pattern is high. The upper and lower limit data is generated in the same way as in the case where the inner pattern is low.

That is, shift mode S2p = 2 or S1
When p = 2 or S1p = 1 and the inner pattern is three pixels in a row,
The upper limit correction value of the center pixel Bc0 (U = center pixel brightness)
Degree Bc1 + internal pattern upper limit allowable value αc) is the last allowable range
If you want to increase the upper limit reference value U _ijWith the upper limit correction value U (U_ij
= Bc1 + αc) and the lower limit of the center pixel Bc0
Correction value L (L = edge side pixel luminance Be1-below the outer pattern
If the limit allowable value βe) extends the previous allowable range, the lower limit is used.
Value L_ijWith the lower limit correction value L (L_ij= Be1-βe) and updated
I do. Here, Bc2 and Bc3 are the inner patterns for three consecutive pixels.
If it is within the number, it is determined to be continuous. In addition to the above,
Mode S2p = 2 or S2p = 1 and the inner pattern consists of 5 pixels
In the case of the continuation, the upper limit correction value U of the center pixel Bc0 (U = center
Side pixel luminance Bc2 + internal pattern upper limit allowable value αc)
If you want to extend the allowable range,_ijIs the upper limit correction value
In U_ij= Bc2 + αc) and the center pixel B
c0 lower limit correction value L (L = edge-side pixel luminance Be2-out
When the pattern lower limit value βe) expands the previous allowable range
Is the lower reference value L_ijWith the lower limit correction value L (L_ij= Be2-β
e) is updated.

Here, it is determined that five consecutive pixels are continuous if Bc2 to Bc5 are within the inner pattern.

After updating Lij = Be2-βe, Lc1
> Be1-βe, the lower limit value of the center pixel is Lc1.
= Be1-βe. And after this update or
If not Lc1> Be1-βe, Lc1 ′> Be1-
βe is determined, and if Lc1 ′> Be1-βe, the lower limit value of the center pixel is updated to Lc1 ′ = Be1-βe, and the routine returns. If Lc1 '> Be1-βe, the process returns. When S2p = 0 when the shift mode S2p is checked, the process also returns.

By the way, when the shift mode is S1p = 1 and three pixels are not continuous, or when S2p = 0 or 1, S1
When p = 0, the shift mode S0p is checked. S0p = s is a subpixel unit (a unit smaller than a pixel)
Is a numerical value from 0 to 1. When s ≠ 0, the edge position is ±
The maximum luminance value Bmax and the minimum luminance value Bmin of the central pixel Bc0 when shifted by s pixels are calculated, and the value obtained by adding the upper limit allowable value of the inner pattern to the maximum luminance value is U, and the value of the outer pattern from the minimum luminance value is U. The value obtained by subtracting the lower limit is L, the upper limit of the central pixel is corrected to (U = Bmax + αc), and the upper limit of the central pixel is corrected to (L = Bmin−βe).

When U _ij <U, U _ij = U is updated. When L _ij > L, L _ij = L is updated. After performing such correction and update processing, the process returns. Also, in the shift mode S0p check, s = 0.
In this case, the routine immediately returns without performing the correction and update processing.

When the edge direction is an oblique direction, s is not proportional to the pixel luminance change, but the pixel luminance value is obtained from the edge position for the pixel in the lookup table.

Next, FIG. 17 shows an example of reference data generation by the above-described reference data automatic generation method. (A) of FIG.
Is an example in which a one-pixel forced shift is uniformly performed in the shift mode S1p = 2. However, a target in which a large pattern, a small width pattern, a through hole with a large displacement, and the like are mixed cannot generate an appropriate upper and lower reference value for luminance. . In other words, through hole and general pattern inspection is too severe inspection,
Conversely, the possibility of overlooking a defect in a thin circuit pattern portion increases. In the figure, the through-hole / luminance level, the circuit pattern / luminance level, the substrate / luminance level,
Indicates an allowable luminance range (upper / lower reference value).

FIG. 17B shows that the shift mode is set to S2p
= 1, S1p = 1, S0p = 0.5, the through-hole portion is discriminated by layer, and the center mask setting and 2
In the example in which the pixel edge mask is set, appropriate luminance upper and lower limit reference values are generated for the above-described pattern portion. In the figure, large pattern edge / two pixel shift, medium width pattern edge / one pixel shift, small width pattern edge / s = 0.5 pixel shift, respectively, MS1 is a through hole / center mask, MS2 is a through hole Shows an edge two-pixel mask.

As described above, by setting the reference data generation condition parameters, appropriate reference data can be generated in accordance with the inspection standard of the target pattern.

In the above example, in the generation of the reference data of the stepped portion, the pattern type on the side that determines the allowable range of the edge position deviation of the stepped portion is set by giving a priority to each pattern type, and appropriate reference data is generated. Although it was made possible, as another example, the priority order is not provided, and among the reference data generation condition parameters of the step type inside / outside pattern type,
You may select and determine the one in which the allowable range of the reference data becomes larger.

It is also possible to set in advance which pattern type to select depending on the combination of the pattern types inside and outside the stepped portion.

In the step portion, the non-defective range of the luminance of the step portion is obtained by matching the non-defective range of the uniform portion luminance of each pattern type and the allowable range of the positional deviation of the pattern edge position of the step portion. The shift allowable range is determined by the distance between the pattern edge and the adjacent pattern, that is, the inner pattern width or the outer pattern width. This is because usually, a small positional deviation allowable range is set for a thin pattern, and a large positional deviation allowable range is set for a thick pattern. Here, a pattern width of a stable size may not be obtained for a considerably thin pattern due to a variation in a manufacturing process. In such a case, if the pattern width is within a specific range of the pattern type, the reference image data is corrected based on the pattern width in the pattern design, and the reference data is generated, whereby optimum reference data can be obtained.

In the case where the different pattern types have the same luminance level in the detected image of the reference pattern, stratification cannot be performed only by the luminance level. In this case, however, the reference data generation method is as follows. In addition, there is a method of setting the reference data generation condition parameter for each inspection area, and it is possible to generate reference data suitable for each different pattern type even at the same luminance level. The inspection area B of the pad section 5 in FIG. 7 is also this example. As another method, the image is divided into a plurality of times, and the image is first taken under different lighting conditions such as switching of the lighting direction so that the pattern type can be distinguished, and one reference data of the corresponding portion is generated first. ,
Next, there is a method of capturing an image under the illumination conditions at the time of inspection, updating, generating, and superimposing the reference data of the remaining portion. As an example of this, if the luminance difference between the through-hole and the substrate is small in the illumination during inspection, first, only the through-hole is detected as dark by oblique illumination, and the reference data for the through-hole is generated first. Then, there is a method of generating the reference data of the remaining portion by the illumination at the time of inspection.

As another method, a method may be used in which color imaging is performed and detection conditions for different color components and chromaticity are used instead of the different illumination conditions. As an example of this, there is discrimination when the luminance levels are similar, such as a copper pattern, a gold plating pattern, a solder plating pattern, and a silk pattern.

As described above, there is a possibility that a pattern misalignment may occur when a plurality of images are taken, but at least the first reference pattern is at least specified by generating reference data by sequentially superimposing the pattern positions of a specific pattern type as a reference. For the pattern (1), the accuracy of the pattern comparison inspection is not reduced.

Here, another example of the condition for determining a specific area will be described.

In the case of defect detection, pseudo defects such as dust, rust, and acceptable defects may be excessively detected. However, among these pseudo defects, there is a type in which a shape and a pattern of luminance variation can be specified. Those which can be specified as pseudo defects in this way should be treated as good as possible.

According to this method, the judgment condition of a specific area is set as a collation mode with a good judgment pattern, a specific area number is added to the reference data for a specific area, and the primary judgment is made in the primary inspection of the area. When it is determined to be a defect candidate, the determination pattern is re-determined by comparing it with a previously specified pseudo-defective pattern, that is, a good determination pattern,
If it is determined to be good, it is excluded from the defects. For example, FIG.
For a defect B of 0 or the like, a pattern of an allowable limit is registered as a good judgment pattern, and can be excluded from defects in the collation mode.

As another example, there is a case where it is desired to allow a certain amount of positional deviation of the through-hole (or via) 3 shown in FIG. In this case, if the allowable amount of misalignment is increased as the reference data, the defect detection performance such as oversight of the pinhole defect C is reduced.

As a countermeasure, as shown in Table 3, a specific area number is set for the corresponding portion, and this determination condition is set as a position correction comparison determination mode. Then, in the primary judgment of the inspection of the area, as shown in FIG.
When it is determined to be 2, the positional deviation between the detected image data and the reference data in this area is corrected and compared again, and the part determined to be non-defective is excluded from defects.

That is, in the case of FIG.
Is shifted to the position correction area Y and the comparison determination is performed again to obtain a through hole (or via).
3 (defective candidate a2) is determined to be non-defective, and pinhole defect C (defect candidate a1) is determined to be defective. Here, the reference data of the partial area Z excluding the position correction area Y from the specific area X disappears, but as shown in FIGS. 18B to 18C, the reference data of the contour of the specific area X (luminance allowable range Wa). Is shifted and set, the inspection of the partial area Z can be performed normally. By setting a limit on the amount of position correction, a large through-hole deviation (or via deviation) can be determined to be defective. As described above, it is possible to allow the displacement of the through hole (or via) 3 without lowering the defect detection performance.

In FIG. 18A, 9 is a land, P is the position of a through hole in the reference pattern, and La is FIG. 18B.
The lines from which the inspection image / brightness level Lb of (c) is extracted are shown.

The above method can be applied to silk printing misalignment and the like. In the pattern inspection, alignment is performed with reference to a circuit pattern. Therefore, the positional deviation of silk characters, silk marks, and the like becomes large. However, when the positional deviation is allowed, the detection accuracy of faint silk or blurring defects is reduced.

As shown in FIG. 19 (a), if a specific area X 'is set in the printed portion of the silk 7 and the inspection is performed in the above-described position correction comparison / determination mode, the deviation of the silk printing can be achieved without lowering the defect detection accuracy. Is acceptable. At this time, there is an unstable area where the edge of the circuit pattern 2 is exposed at the mark where the silk printing is displaced. However, as shown in FIG. 19B, the upper and lower limit reference values (luminance allowable range Wb) are changed to those shown in FIG. ), Stable inspection can be performed without erroneous detection of circuit edges. Note that La ′ in FIG.
The lines from which the inspection image / brightness level Lb 'in (c) is extracted are shown. Lc indicates a reference image / luminance level.

Here, when the through holes (or vias) 3 and the silk 7 are set as the specific area, the specific area is insufficient only with the uniform luminance portion, but the luminance uniform portion and the luminance step portion of this pattern type are insufficient. If they are combined, it can be appropriately assigned as a specific area. Further, the area may be enlarged by a predetermined number of pixels around the luminance uniformity portion. Alternatively, a specific area may be set inside a predetermined pixel with reference to the outer periphery of the land.

In this case, there is also a method of directly inputting a specific area while displaying a specified area on a monitor image by operating a keyboard or a mouse. However, a teaching pattern of this part is created from CAD data or the like, and the created pattern is created. The position of a specific area is taught by imaging a teaching pattern, and predetermined upper and lower reference values may be set in reference data corresponding to the specific area, or a specific area number may be added. Good.

Further, this part is registered as a partial pattern having predetermined characteristics, and the reference image data is read out.
Searching for the partial pattern having the predetermined feature, determining a specific region based on the detected position of the partial pattern, and setting predetermined upper and lower reference values in the reference data corresponding to the specific region Alternatively, a specific area number may be added.

The predetermined features include a pattern shape,
The characteristics of the luminance and chromaticity or their distribution and combination are effective. As a matching method, for example, a partial pattern may be compared and searched by a pattern matching method such as a normalized correlation method.

Although the above embodiment has been described with reference to a wiring board, the present invention can be applied to general printed matters, processed goods, assemblies, and the like as described below. For example, when an aesthetic appearance is required for a printed matter, an inspection can be performed as required by changing an inspection standard for each color of the multicolor printing pattern or each line width / site.

When the shape and position of a thin small portion such as an identification mark is important as information, it can be distinguished from other portions and strictly inspected.

Further, for example, when a predetermined shape / position is important for the function of the substrate in a substrate or the like which has been processed in a large number of predetermined shapes, the portion is inspected more rigorously than the other portions, and at the same time, the other portions are subjected to ordinary processing. Inspection can be done.

Further, when inspecting the position and shape of a component obtained by assembling the component on a substrate, the inspection can be performed by changing the standard for each predetermined component.

Further, when performing an appearance inspection of a defect such as a scratch or a stain on an object having a pattern or a pattern, the inspection can be performed while distinguishing the defect from the pattern or the pattern.

Further, for example, a pattern composed of a plurality of patterns, such as a life test, an accelerated test, or a product monitoring, whose pattern changes temporally and environmentally, can be inspected separately for each pattern.

As described above, when a plurality of patterns are mixed or overlapped and an inspection standard is to be changed for each pattern type or site and inspection is to be performed, if the respective patterns can be detected separately, the present invention can be applied. Possible and very effective.

[0177]

According to the first aspect of the present invention, a plurality of pattern types are classified in advance according to a pixel value range of reference image data obtained from a reference pattern, and a reference data generation condition parameter is set for each pattern type. Each time, the reference image data obtained from the reference pattern is read, and the value of the pixel and a predetermined neighboring pixel for each pixel of the reference image data are compared with the pixel value ranges of the plurality of pattern types, and a single pattern If it is within the pixel value range of the type, it is determined to be a uniform part of the pattern type, and reference data is generated based on the reference data generation condition parameter of the pattern type. For example, the pattern type inside and outside the step portion is determined as the step portion, and the inside and / or outside the step portion are determined.
Or, based on the reference data generation condition parameters of the outer pattern type, the reference data is generated, and the detected image data obtained by imaging the inspection pattern is compared with the reference data to detect a defect in the inspection pattern. Inspection of multiple overlapping patterns and inspection of patterns with different standards for each part can be performed at the same time, and practical automatic inspection can be performed, which can automate pattern inspection, save labor, improve reliability, and improve productivity. effective.

According to a thirty-eighth aspect of the present invention, there is provided an image detecting means for imaging an inspection pattern or a reference pattern, an illumination / imaging control means for controlling the illumination and the imaging, and storing the detected image data or the reference image data. Detection image storage means, reference data automatic generation means for generating reference data, reference data storage means for storing the generated reference data, and determination means for detecting the pattern defect by comparing the detected image data with the reference data; The reference data automatic generation means, in advance, stratified into a plurality of pattern types by the pixel value range of the reference image data obtained from the reference pattern,
A reference data generation condition parameter is set for each pattern type, and then, for each pixel of the reference image data obtained from the reference pattern, the values of the pixel and predetermined neighboring pixels are defined as the pixel value ranges of the plurality of pattern types. If the comparison result is within the pixel value range of a single pattern type, it is determined as a uniform portion of the pattern type and reference data is generated based on a reference data generation condition parameter of the pattern type. If the pixel type is not within the pixel value range, the pattern type inside and outside the step portion is determined as the step portion, and the reference data is generated based on the reference data generation condition parameter of the pattern type inside and / or outside the step portion. Therefore, a pattern inspection apparatus utilizing the features of the first aspect can be realized.

According to a second aspect of the present invention, in the first aspect of the present invention, the range of the uniform portion determining neighboring pixel is set for each of the plurality of pattern types, and the pixel and the predetermined Is compared with the pixel value range of the plurality of pattern types, and if the pixel value is within the pixel value range of the single pattern type, the range of the uniform portion determination neighboring pixels of the pattern type is read and the uniformity is determined. When the range of the neighborhood pixel for the part determination is equal to or less than the range of the predetermined neighborhood pixel, it is determined to be a uniform part of the pattern type, and when the range of the neighborhood pixel for the uniform part determination is larger than the range of the predetermined neighborhood pixel, The value of the neighboring pixel in the large range is compared with the pixel value range of the pattern type, and if the pixel value is within the pixel value range, it is determined to be a uniform portion of the pattern type. Is determined as , Because it produces a respective reference data, reduces the excess detection such as near the pattern edges, there is an effect that accurate inspection is possible.

According to a third aspect of the present invention, in the first or second aspect of the present invention, since the pixel value is used as the luminance, the above-described operation and effect can be obtained using the luminance as the pixel value.

According to a fourth aspect of the present invention, in the first or second aspect of the present invention, the pixel value is a color component value or a color information value such as chromaticity or saturation, so that it is difficult to discriminate it with black and white luminance. Even when it is difficult to perform the inspection, there is an effect that the inspection can be performed with high accuracy by the color characteristics.

A fifth aspect of the present invention is the method according to the first or second aspect, wherein the imaging method detects a height of an inspection pattern to form an image, and the pixel value is a height value. In the case of inspecting a pattern by detecting the height, there is an effect that the inspection can be similarly performed with the feature of the height.

According to a sixth aspect of the present invention, in the first or second aspect of the present invention, the pixel value is used as a non-linear conversion value of the detection data. Since the tone can be appropriately secured, there is an effect that the inspection of each inspection item of each pattern type can be performed with high accuracy.

The invention of claim 7 is the invention of claims 1 to 6
In any one of the inventions, since the reference image data is obtained by imaging the reference pattern, teaching to the inspection device (teaching of the inspection standard) is performed in accordance with the pattern manufacturing process such as expansion and contraction and thickening of the pattern. The effect is that the practical inspection can be easily performed.

The invention of claim 8 is the first to seventh aspects of the present invention.
In any of the inventions described above, the reference image data is created and obtained from the CAD data of the reference pattern.
Since the data is matched with the D data, there is an effect that a highly accurate inspection can be performed based on a product standard. The invention of claim 9 is
8. The image processing apparatus according to claim 1, wherein the position of each image data obtained by imaging a plurality of reference patterns is adjusted, and an average value operation, an intermediate value operation, or an intermediate value operation between the image data is performed for each pixel. Since the average image calculation is performed to generate the reference image data, the invention of claim 10
8. The image processing apparatus according to claim 1, wherein a position of each image data obtained by imaging a plurality of reference patterns is aligned, a correlation operation between the image data is performed for each partial region, and an image having a high correlation is obtained. Since the standard image data consisting of a combination of data is selected and the standard image data is used as the reference image data of the partial area, the invention according to claim 11 is the invention according to claim 8 or claim 9 The alignment method of each image data is performed by measuring the amount of displacement between the image data for each partial area at a resolution of a unit smaller than a pixel, and performing interpolation calculation of neighboring pixel values according to the measured amount of displacement. Since it is a method of obtaining each subsequent image data,
Even when it is difficult to obtain a complete reference pattern, an optimal reference pattern can be obtained by calculation from a plurality of reference patterns, so that teaching work using the reference pattern is simplified and the inspection accuracy is improved.

According to a twelfth aspect of the present invention, in any one of the first to eleventh aspects, the uniform portion is
The non-defective range of the pixel value of the uniform portion set in the reference data generation condition parameter of the pattern type is used as reference data, and the step portion is set in the reference data generation condition parameter of the pattern type inside and / or outside the step portion. The maximum allowable range is determined for each pixel in the stepped portion and the vicinity thereof from the non-defective range of each uniform portion pixel value and the position shift allowable range of the pattern edge, and the obtained maximum allowable range is used as reference data. According to the twelfth aspect of the present invention, based on the reference data generation condition parameters of the pattern type inside and / or outside the step portion, the pattern width in the direction perpendicular to the pattern edge is measured and measured. Since the position shift allowable range of the pattern edge is determined from the pattern width, the reference data of the stepped portion can be optimized, and Detection is reduced, the effect of improving inspection accuracy.

According to a fourteenth aspect of the present invention, in accordance with the thirteenth aspect, the inner and / or outer pattern in the direction orthogonal to the pattern edge is based on a reference data generation condition parameter of the pattern type inside and / or outside the step portion. The width of the pattern is measured, and among these pattern widths, the pattern width in a specific range of a predetermined pattern type is determined based on the designed pattern width, so that the position shift allowable range of the pattern edge is determined. Even when a considerably thin pattern cannot be formed with a predetermined dimension, the dimension can be applied in design, so that the reference data of the step portion can be optimized, excessive detection is reduced, and the inspection accuracy is improved.

The invention according to claim 15 is the invention according to claims 1 to 1.
4. The invention according to claim 4, wherein the reference data includes a pixel value upper reference value and a pixel value lower reference value, and for each pixel of the detected image data, the pixel value is calculated as the pixel value upper reference value and the pixel value lower reference value. And a defect of the pattern is detected, so that the inspection can be performed at high speed by a simple inspection algorithm.

According to a sixteenth aspect of the present invention, in any one of the first to fifteenth aspects, the position of the inspection pattern is adjusted based on the pattern position of a specific pattern type of the reference pattern, and the detected image data is referenced. Since the pattern defect is detected by comparing the data with the data, a highly accurate inspection can be performed even if the inspection pattern is misaligned.

In a seventeenth aspect of the present invention, in any one of the first to sixteenth aspects, a predetermined neighboring pixel corresponding to each pixel of the reference image data is a window pixel centered on the pixel. In the case of the stepped portion, the pixel in the orthogonal direction and the pixel in the oblique direction of the window pixel array are compared, and if this luminance difference is smaller than a predetermined value, the pixel in the orthogonal direction is preferentially determined as the edge-side pixel. In addition, the direction of the edge can be determined without being affected by the variation in the luminance of the pattern edge.

According to an eighteenth aspect of the present invention, in any one of the first to seventeenth aspects, the reference data generation condition parameters include a mode value, a uniform portion mask upper limit value, and a uniform portion mask lower limit value. Uniform portion mask mode or mode value, step portion mask upper limit value, step portion mask mode or mode value composed of step portion mask lower limit value,
2. A step shift mode comprising a pixel value upper limit allowable value and a pixel value lower limit allowable value.
According to a ninth aspect, in the invention of the eighteenth aspect, the uniform portion mask mode sets the uniform portion mask upper limit value and the uniform portion mask lower limit value to the upper limit reference value and the lower limit reference value of the reference data regardless of the previous reference data. It consists of a forced setting mode, or an overlap setting mode in which the upper limit value of the uniform portion mask and the lower limit value of the uniform portion mask are updated to the upper limit reference value and the lower limit reference value of the reference data only when the allowable range of the previous reference data is expanded. According to a twentieth aspect of the present invention, in the invention according to the eighteenth aspect, the step portion mask mode sets the step portion mask upper limit value and the step portion mask lower limit value to the upper limit reference value of the reference data and the lower limit reference value regardless of the previous reference data. 1 pixel compulsory setting mode for setting values, or a predetermined pixel compulsory setting mode for setting also the upper reference value and lower reference value of the reference data of a predetermined pixel at the center side of the pixel. Since consisting, furthermore claim 21
The invention according to claim 18, wherein in the step shift mode, the edge position of the step is shifted by one or more first predetermined pixels in a plus or minus direction, or in a sub-pixel unit. And a mode in which the second predetermined image smaller than 1 is shifted in the plus direction or the minus direction, and the edge position is shifted by the first predetermined pixel or the second predetermined pixel in the direction orthogonal to the edge line. An upper limit correction value obtained by adding the upper limit allowable value to the pixel value change maximum value of the pixel or a lower limit correction value obtained by subtracting the lower limit allowance value from the minimum pixel value change value becomes the upper limit reference value only when the allowable range of the previous reference data is expanded. Is updated with the upper-limit correction value or the lower-limit reference value is updated with the lower-limit correction value. In the above, the reference image data is divided into a plurality of inspection areas, a reference data generation condition parameter is provided for each inspection area, and the reference image data of each inspection area is compared with the reference data based on the reference data generation condition parameter for each inspection area. Is generated, it is possible to combine various types of reference data generation modes, and it is possible to generate appropriate reference data suitable for each pattern and its part.

A twenty-third aspect of the present invention is the invention according to any one of the first to twenty-second aspects, further comprising the step of creating a plurality of the reference image data under different conditions. The first reference data is generated based on the first reference data generation condition parameter, and the second and subsequent reference image data are modified based on the second and subsequent reference data generation condition parameters based on the previous reference data. , And the reference data are sequentially updated.
In the invention of claim 23, since a plurality of reference patterns having variations are used as the different conditions, the invention of claim 25 further uses different illumination conditions as the different conditions in the invention of claim 23. According to a twenty-sixth aspect of the present invention, in the twenty-third aspect, a color detection condition such as a different color component or chromaticity is used as the different condition.
In the invention of the third aspect, since different detection conditions depending on the light conversion characteristics of the detection target such as polarization characteristics or fluorescence characteristics are used as the different conditions, it is easy to input the variation of the reference substrate and to set the discrimination of the pattern easily. There is an effect that data can be sequentially updated to achieve optimization.

According to a twenty-eighth aspect of the present invention, in any one of the twenty-third to twenty-third aspects, the positions of the second and subsequent reference patterns are determined based on the pattern position of a specific pattern type of the first reference pattern. In addition, since the reference data is sequentially updated, good reference data can be created even if there is a deviation between the reference patterns, and there is an effect that a highly accurate inspection can be performed.

In particular, the invention of claim 29 is the invention according to any one of claims 1 to 28, wherein a specific area number is added to the reference data for a specific area, and the detected image data is used as the reference data. If a specific area number is added to the reference data when comparing and detecting a defect of the pattern, the pattern is inspected according to the determination condition of the specific area number. In the specific region, the determination condition of the specific region number is defined as a pixel value variation width, a pixel value high-frequency component variation width, or a pixel value low-frequency component variation width. Is detected, so that the specific area can be inspected in more detail.

According to a thirty-first aspect, in the thirty-ninth aspect, the condition for determining the specific area number is set as a collation mode with a good determination pattern, and the detected image data is compared with the reference data in the specific area. If it is determined that the defect is a defect, the pattern is compared with a previously determined good determination pattern and re-determined. If it is determined that the defect is good, the defect is excluded from the defect. In the invention of the twenty-ninth aspect, when the determination condition of the specific area number is set as a position correction comparison determination mode, and when the detected image data is determined to be defective by comparing the detected image data with the reference data in the specific area, the detected image data and the The comparison determination is performed again by shifting the positional relationship of the reference data by a predetermined amount, and when it is determined that the reference data is good, it is excluded from the defect, so that there is an effect that excess detection can be reduced.

According to a thirty-third aspect, in the thirty-ninth aspect, the condition for determining the specific area number is set as a position correction comparison determination mode, and the detected image data is compared with the reference data in the specific area. If determined to be defective,
The position shift of the specific object in the detected image is detected, the reference data is shifted to a position correction area corresponding to the position shift and set, and the specific area is excluded from the specific area except the position correction area. The reference data of the outline portion of the area is shifted and set, and the detected image data in the specific area and the position correction area are compared with the shifted reference data. , So
There is an effect that excess detection of a large displacement such as a through hole, a via, or a silk print pattern can be reduced and the inspection can be performed with high accuracy.

According to a thirty-fourth aspect of the present invention, in any one of the twenty-first to thirty-third aspects, the specific area is
This is a method of setting a uniform portion or a step portion of a predetermined pattern type as a reference, wherein a specific area number is set in a reference data generation condition parameter of the pattern type, and the uniform portion of the pattern type is generated when the reference data is generated. Alternatively, the specific area number is added to the reference data of an area based on the step portion or an area obtained by enlarging or reducing this area by a predetermined distance. 33. In any one of the thirty-third inventions, the teaching pattern of the specific area is imaged, and the position of the specific area is taught,
Since a predetermined value is set in the reference data corresponding to the specific area, or a specific area number is added to the reference data, the invention of claim 36 is further provided in claims 29 to 3
In any one of the inventions, the reference image data is read, a partial pattern having a predetermined characteristic is searched, a specific area is determined based on the detected position of the partial pattern, and the reference data corresponding to the specific area is determined. Since a predetermined value is set or a specific area number is added to the reference data, the invention according to claim 37 is the invention according to claim 36, wherein the predetermined feature is a pattern shape, luminance, chromaticity or the like. Since it is one of these distributions and combination states,
There is an effect that the setting of the specific area can be easily performed.

In a thirty-ninth aspect of the present invention, in the thirty-eighth aspect of the present invention, the position of each image data obtained by imaging and storing a plurality of reference patterns is adjusted, and the operation between each image data is performed for each pixel or partial area. Since reference image data generating means for obtaining reference image data is provided, even when it is difficult to obtain a complete reference pattern, an optimal reference pattern can be obtained by calculation from a plurality of reference patterns, so teaching work using the reference pattern is simplified, This has the effect of improving inspection accuracy.

In a fortieth aspect of the present invention, in the thirty-eighth or thirty-ninth aspect, a specific region number is added to the reference data for a specific region, and the detected image data is compared with the reference data to determine the pattern. When a defect is detected, if a specific area number is added to the reference data, a specific area determining means for inspecting a pattern according to the determination condition of the specific area number is provided, so that a detailed inspection can be performed for a specific area. In addition, there is an effect that excess detection can be reduced and inspection accuracy can be improved.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a pattern inspection apparatus using a method according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram of a method for creating reference data according to the embodiment.

FIG. 3 is an explanatory diagram of another method of creating reference data according to the embodiment.

FIG. 4 is an explanatory diagram of a reference image data creation example according to the embodiment.

FIG. 5 is an explanatory diagram of a detection data conversion method according to the embodiment.

FIG. 6 is a configuration diagram of a main part corresponding to the inspection of a specific region of the above.

FIG. 7A is an explanatory diagram of a detection image of a pattern to-be-inspected in the above. (B) is an explanatory view of a luminance profile of the detection image of the above.

FIG. 8 is an explanatory diagram of a method for automatically generating reference data according to the embodiment.

FIG. 9 is a flowchart of automatic generation of reference data according to the embodiment.

FIG. 10 is a flowchart of generation of reference data of a luminance uniform part according to the third embodiment.

FIG. 11 is a flowchart of generation of reference data of a luminance step portion according to the third embodiment.

FIG. 12 is a diagram showing an example of the arrangement of an edge pixel and a center pixel in the edge direction of the central pixel in the above.

FIG. 13 is an explanatory diagram of a state of a luminance step portion according to the third embodiment.

FIG. 14 is a flowchart of generation of shift mode reference data of a lower order of the inner pattern according to the first embodiment.

FIG. 15 is an explanatory diagram of generation of reference data of a step portion according to the embodiment.

FIG. 16 is a flowchart of generation of shift mode reference data of a higher order in the inner pattern.

FIG. 17 is an explanatory diagram of a generation example of a luminance upper limit reference value according to the embodiment.

FIG. 18 is an explanatory diagram of an inspection example of the through hole portion of the above.

FIG. 19 is an explanatory diagram of an inspection example of a silk-printed portion of the above.

FIG. 20 is an explanatory diagram of an example of an image of a wiring board used for pattern inspection.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Line sensor 11 Image detection part 12 Detected image storage part 13 Judgment part 14 Pass / fail judgment output part 15 Reference data storage part 16 Reference data automatic generation part 17 Lighting part 18 Lighting / imaging control part 19 NC drive part 20 NC table 21 Substrate

Claims (40)

[Claims] 1. A method according to claim 1, wherein a plurality of pattern types are classified in advance according to a pixel value range of reference image data obtained from the reference pattern, and a reference data generation condition parameter is set for each pattern type. The obtained reference image data is read out, and for each pixel of the reference image data, the value of the pixel and a predetermined neighboring pixel are compared with the pixel value ranges of the plurality of pattern types, and within the pixel value range of a single pattern type. If there is, it is determined as a uniform portion of the pattern type, and reference data is generated based on the reference data generation condition parameter of the pattern type. And the outer pattern type is determined, and reference data is generated based on the reference data generation condition parameters of the inner and / or outer pattern types of the step portion, Pattern inspection method characterized by detecting a defect of the inspection pattern by comparing the detected image data obtained by imaging the 査 pattern and the reference data. 2. A range of neighboring pixels for uniform portion determination is set for each of the plurality of pattern types, and for each pixel of the reference image data, the value of the pixel and a predetermined neighboring pixel is determined for each of the plurality of pattern types. Compared with the pixel value range, if within the pixel value range of a single pattern type, read the range of the uniform portion determining neighboring pixels of the pattern type, and the range of the uniform portion determining neighboring pixels is the predetermined neighborhood. If it is equal to or smaller than the range of pixels, it is determined to be a uniform portion of the pattern type. The pixel data is compared with the pixel value range of the type, and if it is within the pixel value range, it is determined to be a uniform portion of the pattern type. Specially Pattern inspection method of claim 1 wherein. 3. The pattern inspection method according to claim 1, wherein said pixel value is luminance. 4. The pattern inspection method according to claim 1, wherein said pixel value is a color component value or a color information value such as chromaticity or saturation. 5. The pattern inspection method according to claim 1, wherein the imaging method is a method of detecting a height of an inspection pattern and forming an image, and the pixel value is a height value. 6. The pattern inspection method according to claim 1, wherein the pixel value is a non-linear conversion value of the detection data. 7. The pattern inspection method according to claim 1, wherein the reference image data is obtained by imaging a reference pattern. 8. The pattern inspection method according to claim 1, wherein the reference image data is created from CAD data of a reference pattern. 9. The method according to claim 8, wherein positions of respective image data obtained by imaging a plurality of reference patterns are aligned, and an average value operation, an intermediate value operation, or an intermediate average value operation between the respective image data is performed for each pixel. The pattern inspection method according to claim 1, wherein reference image data is generated. 10. A standard image comprising a combination of highly correlated image data, wherein the positions of respective image data obtained by imaging a plurality of reference patterns are aligned, and a correlation operation between the respective image data is performed for each partial area. 8. The pattern inspection method according to claim 1, wherein data is selected, and the standard image data is used as reference image data of the partial area. 11. A method of positioning each image data,
A method of measuring the amount of misregistration between image data for each partial area at a resolution of less than a pixel and performing interpolation calculation of neighboring pixel values according to the measured amount of misalignment to obtain each image data after alignment. The pattern inspection method according to claim 9 or 10, wherein: 12. For the uniform portion, the non-defective range of the pixel value of the uniform portion set in the reference data generation condition parameter of the pattern type is used as reference data, and for the step portion, inside and / or outside the step portion. From the non-defective range of each uniform portion pixel value and the permissible range of the position of the pattern edge set in the reference data generation condition parameter of the pattern type, the maximum permissible range of the pixel value for each pixel in the step portion and the vicinity thereof The pattern inspection method according to any one of claims 1 to 11, wherein the pattern inspection method is used as reference data. 13. A method for measuring a pattern width inside and / or outside in a direction orthogonal to a pattern edge based on reference data generation condition parameters of a pattern type inside and / or outside a step portion, and based on the measured pattern width. 13. The pattern inspection method according to claim 12, wherein an allowable position shift range of the edge is determined. 14. A method for measuring a pattern width inside and / or outside in a direction orthogonal to a pattern edge based on reference data generation condition parameters of a pattern type inside and / or outside a step portion, and among these pattern widths, 14. The pattern inspection method according to claim 13, wherein for a pattern width in a specific range of a predetermined pattern type, a position shift allowable range of the pattern edge is determined based on a design pattern width. 15. The reference data includes a pixel value upper reference value and a pixel value lower reference value, and for each pixel of the detected image data, compares each pixel value with the pixel value upper reference value and the pixel value lower reference value. 15. The pattern inspection method according to claim 1, wherein a defect of the pattern is detected. 16. A test pattern position is adjusted based on a pattern position of a specific pattern type of a reference pattern,
2. The method according to claim 1, wherein the detected image data is compared with reference data to detect a pattern defect.
5. The pattern inspection method according to any one of 5. 17. A predetermined neighboring pixel corresponding to each pixel of the reference image data is defined as a window pixel centered on the pixel, and in the case of the stepped portion, a pixel in an orthogonal direction of the window pixel array and a pixel in an oblique direction are arranged. 17. The pixel according to claim 1, wherein the pixels are compared, and if the difference between the pixel values is smaller than a predetermined value, the pixel in the orthogonal direction is preferentially determined as the edge-side pixel. Pattern inspection method. 18. The reference data generation condition parameter may be:
A uniform portion mask mode composed of a mode value, a uniform portion upper limit value, and a uniform portion lower limit value, or a step portion mask mode composed of a mode value, a step portion upper limit value, and a step portion lower limit value, or a mode value 18. The pattern inspection method according to claim 1, further comprising one of a step shift mode including a pixel value upper limit allowable value and a pixel value lower limit allowable value. 19. A forced setting mode in which the uniform portion mask mode sets the uniform portion mask upper limit value and the uniform portion mask lower limit value to the upper limit reference value and the lower limit reference value of the reference data regardless of the previous reference data. 2. An overlap setting mode for updating and setting the upper limit value of the uniform portion mask and the lower limit value of the uniform portion mask to the upper limit reference value and the lower limit reference value of the reference data only when the allowable range of the reference data is expanded.
8. The pattern inspection method according to 8. 20. The step mask mode is a one-pixel forced setting mode in which a step mask upper limit value and a step mask lower limit value are set to an upper reference value and a lower reference value of reference data regardless of previous reference data. 19. The pattern inspection method according to claim 18, further comprising a predetermined pixel compulsory setting mode for setting an upper reference value and a lower reference value of reference data of a predetermined pixel on the center side of the pixel. 21. The step shift mode wherein the edge position of the step is shifted by one or more first predetermined pixels in the plus or minus direction, or the second is smaller than 1 in subpixel units. In the mode in which the predetermined image is shifted in the plus direction or the minus direction, and the pixel value of the pixel changes when the edge position is shifted by the first predetermined pixel or the second predetermined pixel in the direction orthogonal to the edge line. The upper limit correction value obtained by adding the upper limit allowable value to the maximum value or the lower limit correction value obtained by subtracting the lower limit allowable value from the pixel value variation minimum value, the upper limit reference value is set to the upper limit correction value only when the allowable range of the previous reference data is expanded. 2. The method according to claim 1, wherein the lower limit reference value is updated with the lower limit correction value.
8. The pattern inspection method according to 8. 22. The reference image data is divided into a plurality of inspection areas, a reference data generation condition parameter is provided for each inspection area, and a reference data generation condition parameter for each inspection area is provided for the reference image data of each inspection area. 22. The pattern inspection method according to claim 1, wherein reference data is generated based on the reference data. 23. A method for producing a plurality of said reference image data under different conditions.
The first reference data is generated based on the first reference data generation condition parameter, and the second and subsequent reference image data are modified based on the second and subsequent reference data generation condition parameters based on the previous reference data. 23. The pattern inspection method according to claim 1, wherein the reference data is sequentially updated. 24. A method according to claim 23, wherein a plurality of reference patterns having variations are used as said different conditions.
The pattern inspection method described. 25. The pattern inspection method according to claim 23, wherein different illumination conditions are used as said different conditions. 26. The pattern inspection method according to claim 23, wherein different color components or color detection conditions such as chromaticity are used as said different conditions. 27. A pattern inspection method according to claim 23, wherein different detection conditions based on light conversion characteristics of a detection target such as polarization characteristics or fluorescence characteristics are used as said different conditions. 28. The method according to claim 23, wherein the reference data is sequentially updated by adjusting the positions of the second and subsequent reference patterns based on the pattern position of a specific pattern type of the first reference pattern. Item 28. The pattern inspection method according to any one of Items 27 to 27. 29. A specific area number is added to the reference data for a specific area, and when the detected image data is compared with the reference data to detect a pattern defect, the specific area number is added to the reference data. The pattern inspection method according to any one of claims 1 to 28, wherein a pattern is inspected according to the determination condition of the specific area number. 30. The determination condition of the specific area number is defined as a pixel value variation width, a pixel value high frequency component variation width, or a pixel value low frequency component variation width. 30. The pattern inspection method according to claim 29, wherein a pattern defect is detected. 31. The determination condition of the specific area number is set as a matching mode with a good determination pattern, and when the detected image data is compared with the reference data in the specific area and determined as a defect, it is specified in advance. 30. The pattern inspection method according to claim 29, wherein a re-determination is performed by collating with a set good determination pattern, and when the pattern is determined to be good, the pattern inspection method is excluded from defects. 32. The determination condition of the specific area number is set as a position correction comparison determination mode, and when the detected image data is determined to be defective by comparing the detected image data with reference data in the specific area, 30. The pattern inspection method according to claim 29, wherein the comparison determination is performed again by shifting the positional relationship of the reference data by a predetermined amount, and when the determination is good, the pattern inspection method is excluded from the defects. 33. The determination condition of the specific area number is set as a position correction comparison determination mode, and when the detected image data is compared with the reference data in the specific area to determine a defect, the identification in the detected image is determined. The position shift of the target is detected, the reference data is shifted to a position correction area corresponding to the position shift and set, and the reference area of the contour of the specific area is set in a partial area excluding the position correction area from the specific area. The data is shifted and set, and the detected image data in the specific area and the position correction area are compared with the shifted and set reference data. The pattern inspection method according to claim 29, wherein: 34. A method for setting the specific area based on a uniform portion or a stepped portion of a predetermined pattern type, wherein a specific area number is set in a reference data generation condition parameter of the pattern type, When the reference data is generated, the specific area number is added to reference data of an area based on a uniform portion or a step portion of the pattern type or an area obtained by enlarging or reducing the area by a predetermined distance. The pattern inspection method according to any one of claims 29 to 33. 35. An image of the teaching pattern in the specific area, the position of the specific area is taught, and a predetermined value is set in reference data corresponding to the specific area, or a specific area number is set in the reference data. The pattern inspection method according to any one of claims 29 to 33, wherein 36. The reference image data is read, a partial pattern having a predetermined characteristic is searched for, a specific area is determined based on the detected position of the partial pattern, and a predetermined value is added to the reference data corresponding to the specific area. The pattern inspection method according to any one of claims 29 to 33, wherein the setting is made or a specific area number is added to the reference data. 37. The method according to claim 37, wherein the predetermined characteristic is a pattern shape, a brightness,
37. The pattern inspection method according to claim 36, wherein any one of chromaticity, distribution, and a combination thereof is used. 38. Image detecting means for imaging an inspection pattern or a reference pattern, illumination / imaging control means for controlling illumination and imaging, and detected image storage means for storing the captured detected image data or reference image data. A reference data automatic generation unit for generating reference data, a reference data storage unit for storing the generated reference data, and a determination unit for comparing the detected image data with the reference data to detect a pattern defect; Automatic data generation means classifies in advance into a plurality of pattern types according to the pixel value range of the reference image data obtained from the reference pattern, sets a reference data generation condition parameter for each pattern type, and then sets the reference pattern For each pixel of the reference image data obtained, the value of the pixel and a predetermined neighboring pixel are compared with the pixel value ranges of the plurality of pattern types. If it is within the pixel value range of a single pattern type, it is determined to be a uniform portion of the pattern type, and reference data is generated based on the reference data generation condition parameter of the pattern type. If it is not within the pixel value range, the pattern type inside and outside the step portion is determined as the step portion, and based on the reference data generation condition parameter of the pattern type inside and / or outside the step portion,
A pattern inspection apparatus, which is means for generating reference data. 39. A reference image data generating means for obtaining the reference image data by performing a calculation between the respective image data for each pixel or partial area by aligning the positions of the respective image data obtained by imaging and storing a plurality of reference patterns. The pattern inspection apparatus according to claim 38, wherein the pattern inspection is performed. 40. A specific area number is added to the reference data for a specific area, and when the detected image data is compared with the reference data to detect a pattern defect, the specific area number is added to the reference data. 40. The pattern inspection apparatus according to claim 38, further comprising a specific area determination unit configured to inspect a pattern according to the determination condition of the specific area number.